Hung Quoc To

---
name: author-commercial-teaching-pitch
description: |
  Author a Commercial Teaching pitch using the six-step choreography and polish it with SAFE-BOLD.

  Trigger this skill when you need to:
  - Write a sales pitch using commercial teaching methodology
  - Build a commercial teaching pitch from a validated insight
  - Structure a six-step pitch: warmer reframe rational drowning emotional impact new way solution
  - Author a sales narrative that leads to your solution rather than leading with it
  - Build a sales deck following challenger selling choreography
  - Write a challenger pitch or insight-led pitch for a B2B conversation
  - Draft a pitch script for a sales rep or sales enablement team
  - Differentiate a sales conversation by teaching customers something new about their business
  - Apply SAFE-BOLD to sharpen a teaching pitch before delivery

  NOT for: building the commercial insight itself (use build-commercial-insight first),
  tailoring the pitch to individual stakeholders (use tailor-pitch-by-stakeholder after),
  or diagnosing an existing pitch (use diagnose-pitch-for-commercial-teaching-fit).
version: 1.0.0
homepage: https://github.com/bookforge-ai/bookforge-skills/tree/main/books/the-challenger-sale/skills/author-commercial-teaching-pitch
metadata:
  openclaw:
    emoji: "🎤"
    homepage: "https://github.com/bookforge-ai/bookforge-skills"
status: draft
source-books:
  - id: the-challenger-sale
    title: "The Challenger Sale"
    authors:
      - Matthew Dixon
      - Brent Adamson
    chapters:
      - 5
tags:
  - sales
  - b2b-sales
  - challenger-sale
  - commercial-teaching
  - pitch-authoring
  - sales-messaging
depends-on:
  - build-commercial-insight
execution:
  tier: 1
  mode: hybrid
  inputs:
    - type: document
      description: "commercial-insight.md (output of build-commercial-insight) containing the validated Reframe and unique strength anchor, plus target customer segment"
  tools-required:
    - Read
    - Write
    - AskUserQuestion
  works-offline: true
discovery:
  goal: "Author a Commercial Teaching pitch using the six-step choreography (Warmer → Reframe → Rational Drowning → Emotional Impact → New Way → Your Solution) and polish it with the SAFE-BOLD framework. Outputs a pitch-script.md artifact ready for delivery."
  triggers:
    - "Write a sales pitch for my product"
    - "Build a commercial teaching pitch from my insight"
    - "I need a six-step challenger pitch"
    - "How do I structure a teaching conversation for my sales team?"
    - "Draft a pitch script for a B2B sales call"
    - "Apply SAFE-BOLD to my teaching pitch"
    - "I want an insight-led pitch, not a product pitch"
  audience:
    - Sales reps and account executives building teaching-led pitches
    - Sales enablement teams creating repeatable pitch assets
    - Product marketing managers turning insights into sales narratives
    - Sales managers coaching reps on commercial teaching delivery
  prerequisites:
    - "Run build-commercial-insight first to produce a validated commercial-insight.md"
  not-for:
    - Auditing or diagnosing an existing pitch — use diagnose-pitch-for-commercial-teaching-fit
    - Building the commercial insight itself — use build-commercial-insight
    - Tailoring the completed pitch to individual stakeholder roles — use tailor-pitch-by-stakeholder
---

# author-commercial-teaching-pitch

Author a Commercial Teaching pitch using the six-step choreography and polish it with SAFE-BOLD. Takes the validated `commercial-insight.md` from `build-commercial-insight` as its primary input and outputs a `pitch-script.md` ready for delivery or enablement packaging.

---

## When to Use

Use this skill after `build-commercial-insight` has produced a validated Reframe and unique strength anchor. This skill structures that insight into a complete, deliverable six-step conversation that engages both the customer's rational and emotional response — because logic alone is rarely enough to overcome the status quo.

The output is not a product presentation. It is a teaching narrative that arrives at your solution as the natural conclusion of a story about the customer's business.

---

## Context and Input Gathering

### Load the prerequisite artifact

If `commercial-insight.md` exists in the working directory, read it now. Extract:
- The validated Reframe statement (the headline insight)
- The unique strength anchor (what only your solution can deliver)
- The target customer segment
- The quantified cost-of-inaction evidence (for Step 3)
- The blocking worldview (what the customer currently believes)

If `commercial-insight.md` does not exist, use `AskUserQuestion` to collect:
1. The Reframe candidate: the unexpected perspective the customer has not considered
2. The unique strength anchor: what your solution delivers that competitors cannot
3. Target customer segment: role, industry, size, common behavior patterns
4. Quantified evidence: data or research showing the cost of the problem or size of the opportunity

Do not proceed without these four inputs. The pitch cannot be constructed from a generic or unvalidated insight.

---

## Process

### Step 1 — Draft the Warmer

Write an opening that names 2-3 challenges you are observing at companies similar to the customer's. Draw from the target segment profile in `commercial-insight.md`.

Structure the Warmer as:
- "We've worked with a number of companies in [segment], and we've consistently seen these challenges come up: [challenge 1], [challenge 2], [challenge 3]. Is that what you're seeing, or would you add something else to the list?"

**Emotional target:** Customer feels "they get us" — understood without being interrogated.

**Why:** This establishes credibility through demonstrated domain knowledge (Hypothesis-Based Selling), not through company credentials or product claims. Customers with "solutions fatigue" respond positively because they receive informed perspective rather than having to educate you.

**Self-check:** Does the Warmer lead with hypotheses, not open-ended questions? Does it avoid any mention of your solution or company? Does it invite the customer to confirm or extend the challenge list?

**Anti-pattern:** Asking "What's keeping you up at night?" (discovery framing). Also: transitioning directly from the Warmer to your solution — this is the next step a core-performing rep takes and the one move most likely to waste the goodwill just established.

---

### Step 2 — Draft the Reframe

Write the single headline insight that connects the challenges the customer just confirmed to a problem or opportunity they had not recognized. This is just the headline — not the full explanation.

The Reframe must:
- Introduce an unexpected angle that inverts or expands the customer's current assumption
- Be stated in 1-2 sentences
- Not mention your solution or company

**Emotional target:** "Huh, I never thought of it that way before."

**Why:** The Reframe is the pivot point of the entire conversation. Without a genuine perspective shift, everything that follows is either confirming existing beliefs (no urgency) or promoting your solution prematurely (ignored). The teaching moment lives specifically here — if the customer already believed this, no teaching has occurred.

**Self-check:** Does the customer's hypothetical first reaction suggest curiosity and surprise? Or would they say "I totally agree"? Enthusiastic agreement is a failure signal. Also check: is the Reframe just the headline at this stage, or have you overloaded it with explanation that belongs in steps 3 and 4?

**Anti-pattern:** Getting enthusiastic agreement — this means the insight confirms rather than challenges the customer's worldview, and they have likely already considered solutions (possibly competitor solutions). Timidity is also an anti-pattern: "If you're going to reframe, be sure you really reframe."

---

### Step 3 — Draft Rational Drowning

Write the quantified business case for why the Reframe matters. Use data, benchmarking figures, or research to show the hidden cost of the problem or the size of the missed opportunity.

If an ROI calculator is included, it must calculate the return on solving the problem — not the return on buying your solution.

Structure the Rational Drowning as:
- Here is what this costs most companies like yours (quantified)
- Here is why the cost is larger than it appears (hidden or indirect costs)
- Here is what inaction compounds to over time

**Emotional target:** "This is real, and bigger than I thought." The customer should feel they are "drowning" — not panicked, but acutely aware of a material problem they had underestimated.

**Why:** Rational evidence is necessary to justify action, but it must be framed against the problem the customer now believes they have (from Step 2), not against your solution. The ROI in Step 3 answers: "Is it worth fixing this at all?" Step 6 will answer: "Why buy from you?"

**Self-check:** Is the ROI explicitly about the challenge the Reframe revealed — not about your product? Is the data specific enough to feel credible? Would a skeptical CFO engage with this data?

**Anti-pattern:** Presenting ROI that requires customers to first believe in your product to understand the numbers. Also: staying in Step 3 when the customer is already convinced — move to Step 4 before the emotional urgency dissipates.

---

### Step 4 — Draft Emotional Impact

Write a narrative story that places companies like the customer's in the painful scenario your data just described. The story must feel immediately familiar — the customer should recognize their own organization in it.

Structure the Emotional Impact as:
- "I understand you're a little bit different, but let me show you how we've seen this play out at similar companies..."
- [Specific scenario that maps to the customer's role, industry, and common behavior pattern]
- The scenario ends at the moment of maximum pain — the unplanned cost, the scramble, the workaround — and pauses there

**Emotional target:** "This is MY problem, not just the company's." The customer stops seeing the data as applying to others and starts replaying the scenario in their own context.

**Why:** Customers often respond to strong rational arguments with "I see what you're saying, but we're different." More data never defeats this response — because it is not a logic problem, it is an emotional connection failure. The story creates the link between the abstract problem and the customer's lived experience.

**Self-check:** Can the customer immediately place themselves in the story? Does the scenario end at the pain moment — not at the solution? Does the narrative feel specific and credible rather than generic?

**Anti-pattern:** Responding to "we're different" with more charts. More charts intensify the same failed approach. Also: making the story too polished or corporate — the roughness of a real customer situation is what makes it land.

---

### Step 5 — Draft the New Way

Write a point-by-point description of the capabilities the customer would need to adopt in order to address the opportunity or solve the problem. Frame this as what the customer needs to do differently — not what your solution does.

Structure the New Way as:
- "Here is the type of organization you would need to become to capture this opportunity / eliminate this cost..."
- [Specific capability categories: what they would track, how they would operate, what relationships they would manage differently]

**Emotional target:** "We need to change how we operate." The customer agrees conceptually to the direction of change before being introduced to any specific vendor.

**Why:** Customers must buy the concept before they buy the product. Step 5 creates the buy-in to change that makes Step 6 feel like the natural answer rather than a sales pitch. If the customer agrees "that's what we need to do" in Step 5, Step 6 becomes a question of which supplier, not whether to act.

**Self-check:** Is Step 5 still about the customer's organizational capabilities — not about your product's features? Does it describe a world-class solution abstractly before naming your company? Does it feel like aspirational vision, not a product spec sheet?

**Anti-pattern:** Rushing to name your product in Step 5. This is the most common premature close in commercial teaching. The customer must first agree to the direction of change; then they are ready to hear who can deliver it.

---

### Step 6 — Draft Your Solution

Write the specific demonstration of how your solution is better positioned than any alternative to equip the customer to act on the New Way they agreed to in Step 5. This is the first time you name your company.

Structure Your Solution as:
- "Given everything we've talked about, here's how [your solution] delivers exactly what we described..."
- Map each capability in the New Way to a specific, unique feature or differentiator of your solution
- Close by pointing to the natural next step (diagnostic, pilot, or proposal)

**Emotional target:** "And you can deliver it." Customer sees the direct line from the New Way to your specific offer.

**Why:** The hard work of Steps 1-5 creates the context in which your capabilities feel like the natural answer rather than a sales pitch. If you have correctly identified unique capabilities that were not legible before the teaching, competitors cannot easily follow. The solution is legible only to a customer who has been taught to value it.

**Self-check:** Is every capability mentioned in Step 6 mapped directly to a point from the New Way (Step 5)? Is competition still potentially viable at this stage — if yes, the earlier steps may not have sufficiently differentiated the path. Does the close feel like a logical next step, not a closing technique?

**Anti-pattern:** Introducing new capabilities in Step 6 that were not set up in Step 5. Also: presenting your full product portfolio — Step 6 should only present capabilities that map to the specific New Way from this pitch.

---

### Step 7 — Per-Step Self-Check

Before the SAFE-BOLD pass, verify the sequence:

| Step | Emotional target met? | Success signal present? | Anti-patterns avoided? |
|------|----------------------|------------------------|----------------------|
| 1 Warmer | Customer would feel understood | Engagement + reaction invited | No discovery questions, no solution preview |
| 2 Reframe | Surprise + curiosity, not agreement | "I hadn't thought of it that way" | Not timid, not explained too early |
| 3 Rational Drowning | Problem feels large and material | Quantified, CFO-credible | ROI on problem, not on product |
| 4 Emotional Impact | Customer sees themselves in story | Rueful recognition, not generic | Story ends at pain, not solution |
| 5 New Way | Customer agrees to direction of change | "That's what we need to do" | No product name, no vendor mention |
| 6 Your Solution | Natural fit between need and offer | Clear next step proposed | Only capabilities set up in Step 5 |

If any row fails, return to that step and revise before continuing.

---

### Step 8 — SAFE-BOLD Polish Pass

Score the draft pitch against the four SAFE-BOLD dimensions (developed by Neil Rackham and KPMG). Each dimension is scored on a continuum: SAFE is weak, BOLD is strong.

| Dimension | SAFE (1) | BOLD (5) | Your score | Evidence from draft |
|-----------|----------|----------|------------|-------------------|
| **Big** | Small, narrow scope | Expansive, far-reaching | | |
| **Innovative** | Follower idea, common | Leading-edge, untested | | |
| **Risky** | Easily achievable | Asks significant change | | |
| **Difficult** | Easy to implement | Hard because of scale, uncertainty, or politics | | |

**Why Difficult matters:** If the problem is easy for the customer to solve without help, there is no reason to hire you. The difficulty of the problem is what creates the need for a capable partner.

For any dimension scoring below 3:
- **Big below threshold:** Broaden the scope of the Reframe or the Rational Drowning data — show a larger pattern, a longer time horizon, or a cross-functional impact
- **Innovative below threshold:** Sharpen the Reframe — if the customer could have thought of this themselves, it is not yet a teaching insight
- **Risky below threshold:** Ensure the New Way requires meaningful organizational change — if customers can adopt the direction casually, there is no urgency to act now
- **Difficult below threshold:** Strengthen the Rational Drowning and New Way — the problem should feel genuinely hard to solve at scale, which is exactly why your solution matters

Document SAFE-BOLD scores and any targeted rewrites in the `pitch-script.md` output.

---

### Step 9 — Write pitch-script.md

Create `pitch-script.md` with this structure:

```markdown
# [Pitch Title]

**Customer segment:** [Segment from commercial-insight.md]
**Unique strength anchor:** [From commercial-insight.md]
**SAFE-BOLD scores:** Big: [X/5] | Innovative: [X/5] | Risky: [X/5] | Difficult: [X/5]

---

## Step 1: Warmer
**Emotional target:** Customer feels understood
**Script:** [Draft language]
**Success signal:** [What the rep listens for]

## Step 2: Reframe
**Emotional target:** "Huh, I hadn't thought of it that way"
**Script:** [Draft language]
**Success signal:** [What the rep listens for]

## Step 3: Rational Drowning
**Emotional target:** "This is real and bigger than I thought"
**Script:** [Draft language with quantified evidence]
**Success signal:** [What the rep listens for]

## Step 4: Emotional Impact
**Emotional target:** "This is MY problem, not just the company's"
**Script:** [Narrative story]
**Success signal:** [What the rep listens for]

## Step 5: A New Way
**Emotional target:** "We need to change how we operate"
**Script:** [Draft language]
**Success signal:** [What the rep listens for]

## Step 6: Your Solution
**Emotional target:** "And you can deliver it"
**Script:** [Draft language]
**Success signal / Next step:** [What the rep proposes]
```

---

## Inputs

| Input | Required | Description |
|-------|----------|-------------|
| `commercial-insight.md` | Yes | Output of build-commercial-insight — validated Reframe, unique strength anchor, quantified cost evidence, target segment |
| Target customer segment | Yes | Role, industry, company size, common behavior patterns |
| Benchmarking or spend data | Optional | Quantified evidence for Rational Drowning — if not in commercial-insight.md, ask the user |

---

## Outputs

| Output | Path | Description |
|--------|------|-------------|
| `pitch-script.md` | `./pitch-script.md` | Six-step pitch with per-step language, emotional targets, success signals, and SAFE-BOLD scores |

---

## Key Principles

**The supplier appears last, not first.** Steps 1-5 contain no mention of your company or product. Your solution appears only in Step 6, as the natural answer to a problem the customer now believes is urgent. "Lead to, not with."

**Emotional and rational engagement are both required.** "No one ever sold anything off a spreadsheet alone." The six-step arc is designed to take the customer to a dark place (Steps 3-4) before showing them the way forward (Steps 5-6). If either the rational or the emotional layer is missing, the pitch is incomplete.

**The ROI belongs to the problem, not to the purchase.** In Step 3, if your ROI calculator explicitly references your product, you are answering the wrong question. The customer must first believe the problem is worth solving before they will listen to why your solution is the best way to solve it.

**Step 5 is the most commonly skipped.** Reps are trained to close at the moment of maximum customer agreement. After Step 4 emotional resonance, jumping to Step 6 feels natural but is wrong — it collapses the gap between the New Way and your specific product. Let the customer agree to the concept of change before introducing who should deliver it.

**SAFE-BOLD prevents organizational watering-down.** Teaching pitches that pass through multiple internal reviewers reliably lose their edge. SAFE-BOLD exists to measure and prevent this. A pitch that produces no discomfort internally has almost certainly lost the sharpness that makes it work with customers.

---

## Examples

See `references/worked-examples.md` for two detailed worked examples:
- W.W. Grainger "Power of Planning the Unplanned" — full six-step walkthrough with Pain Chain and Parts Orphanage emotional impact techniques
- ADP Dealer Services Profit Clinics — six-step walkthrough showing how commercial teaching delivered results during a 40% market contraction

Both examples trace the full sequence from Warmer through Solution, including how each step was operationalized and what customer reactions were targeted.

---

## References

- `references/worked-examples.md` — Full walkthrough of Grainger and ADP examples

**Source:** *The Challenger Sale* by Matthew Dixon and Brent Adamson. Chapter 5 (Teaching for Differentiation, Part 2). SAFE-BOLD Framework developed by Neil Rackham and KPMG.

---

## License

This skill is licensed under [CC-BY-SA 4.0](https://creativecommons.org/licenses/by-sa/4.0/). You are free to use, adapt, and redistribute with attribution.

Source book content is copyrighted by the authors. This skill contains no verbatim passages — all content is paraphrased, structured, and extended for agent execution.

---

## Related BookForge Skills

| Skill | Relationship |
|-------|-------------|
| `build-commercial-insight` | Run before this skill — validates the Reframe and unique strength anchor that anchors the pitch |
| `tailor-pitch-by-stakeholder` | Run after this skill — adapts the completed pitch to individual stakeholder roles and priorities |
| `diagnose-pitch-for-commercial-teaching-fit` | Alternative entry point — use to audit an existing pitch before rebuilding with this skill |
| `classify-rep-profile` | Use independently — confirms whether the rep is equipped to deliver a Challenger-style teaching pitch |

FILE:references/worked-examples.md
# Worked Examples: Commercial Teaching Pitch in Practice

Two real-world examples of companies executing the six-step Commercial Teaching choreography. Both examples are paraphrased from *The Challenger Sale* by Matthew Dixon and Brent Adamson, Chapter 5. They illustrate how each step functions in practice and what customer reactions to target.

---

## Example 1: W.W. Grainger — "The Power of Planning the Unplanned"

**Company:** W.W. Grainger, a $7 billion distributor of maintenance, repair, and operations (MRO) equipment serving approximately two million companies across North America.

**The business problem Grainger faced:** Customers had come to see Grainger as a transactional commodity supplier — a vendor of hammers, gloves, and light bulbs — rather than a strategic partner. When renewal conversations arrived, customers talked about price. Grainger wanted to be perceived as a partner who could improve the customer's bottom line, but customers saw no reason to engage at that level.

**The commercial insight:** Research into years of customer purchase data revealed that approximately 40% of a typical company's MRO spend is for unplanned, reactive purchases — items bought at the last minute in response to unexpected needs. Customers had never conceptualized their MRO spend this way. They thought of it as product categories (tools, safety, lighting). The "planned vs. unplanned" distinction was invisible to them — but the cost consequences of that invisibility were substantial.

**Unique strength anchor:** Only Grainger's combination of catalog breadth, nationwide branch network, and robust e-commerce platform could enable customers to shift unplanned purchases to planned ones at scale. Competitors might match one or two dimensions but not all three simultaneously.

---

### The Six-Step Execution

**Step 1 — Warmer**

The Grainger rep requests the meeting explicitly to share insights about how companies in the customer's industry can save money on MRO spend — not to present Grainger capabilities. The agenda lists the customer's challenges, not Grainger's offerings.

Opening language: *"We've worked with a number of companies in your space, and we've found that these three challenges come up again and again as most troubling: production line disruptions from unplanned equipment failures, workers' comp costs from safety incidents, and maintenance inefficiency. Is that what you're seeing too, or would you add something else?"*

Grainger found that this single framing shift — from "let me show you what we can do" to "let me share what we're seeing" — transformed sales conversations from defensive procurement exchanges into collegial discussions. The Warmer page alone reportedly generated robust conversations because it led with hypotheses rather than questions.

**Target reaction:** Customer engages, confirms challenges, adds context. Conversation feels like two colleagues rather than a sales pitch.

**Step 2 — Reframe**

The rep introduces a completely different way to think about MRO spend. Rather than organizing it by product category (tools, safety, lighting), the rep introduces the planned/unplanned distinction.

*"Here's something that surprises most companies when they first see it. If we look at your MRO spend not by what you buy but by how you buy — specifically, which purchases are planned in advance versus which are reactive and unplanned — something interesting emerges. Unplanned purchases are likely your second largest MRO category, bigger than any individual product category. And most companies have never tracked it that way."*

The key insight is structural, not factual — it reframes the organization of a concept the customer thought they already understood. They knew they bought hammers; they did not know they had a "reactive purchasing" category representing 40% of their spend.

**Target reaction:** "Huh, I'd never thought about it that way before." Curiosity and a slight lean forward, not enthusiastic agreement.

**Step 3 — Rational Drowning**

The rep quantifies the cost of the planned/unplanned distinction using Grainger's own analysis of customer purchase histories.

Key data points:
- 40% of typical MRO spend is for unplanned purchases
- Each unplanned purchase involves 5-10 people across the organization
- Process costs — invoice generation, supplier search, order placement, inventory management, paperwork and payment — often exceed the cost of the item itself
- A $17 hammer may actually cost $117 when process costs are included
- Most companies work with a small number of suppliers for planned purchases but hundreds of suppliers for unplanned purchases, eliminating all negotiating leverage and requiring full retail pricing

If Grainger has an existing relationship with the customer, reps pull actual purchase data to make the numbers specific to that company. *"Here's what we're actually seeing in your account..."*

**Target reaction:** *"Wow, I had no idea we were wasting that kind of money!"* Customer begins to feel uncomfortable about a problem they had not previously identified as a problem.

**Step 4 — Emotional Impact ("Pain Chain" and "Parts Orphanage")**

Rather than presenting more data, the rep tells a specific story. Grainger calls the visual they use the "Pain Chain."

The scenario (paraphrased): *"Let me show you how this typically plays out. Imagine the coil on the twenty-year-old air-conditioning unit in your CEO's office fails in the middle of summer. It's hot, it's the CEO, and it needs to be fixed immediately. What happens next?*

*You call your primary planned-purchase supplier. After twenty minutes on hold, you learn they sold out of that part and won't restock for two weeks. You call a second supplier you've used occasionally — they don't carry it. A third supplier tells you they should have two in stock but can't locate them in the warehouse. An hour and a half into this, you're getting desperate.*

*Finally, a fourth supplier across town has the part. You pull two people off the production line, put together the emergency paperwork, and send them across town in rush-hour traffic. An hour later they call you: 'Hey boss, they've got three of these. Should we grab an extra one just in case?'*

*And you say yes. You never want to do this again. So those two extra parts go on a shelf in the back of your warehouse — what we call the 'Parts Orphanage' — where they sit and gather dust until the entire system is obsolete and needs to be replaced anyway."*

Grainger reports that customers consistently respond to this story with recognition: *"Wow, it's like you work here. We do that all the time. It absolutely kills us."*

**Why this works:** The Pain Chain is not a generic industry story — it maps to real customer behavior observed across years of customer interviews. Its specificity is what creates the emotional connection. The customer does not merely understand the problem; they relive it.

**Target reaction:** Rueful head shake, wry smile, or a comment like "that's exactly what happened to us last month."

**Step 5 — A New Way**

The rep transitions from the emotional story to the organizational opportunity:

*"Now, that's the problem with just one unplanned purchase in one category. The challenge is, you do that again and again across every MRO category, every day. No company is currently structured to manage this spend proactively across all categories simultaneously. But imagine if you could. Unplanned purchases represent a huge amount of avoidable spend and unnecessary inventory cost. That's cash you could be using for things that actually matter to your business."*

Note: Grainger is still not mentioned by name. This step describes the solution category abstractly — a company that could proactively manage all unplanned MRO purchases — before introducing who can deliver it.

**Target reaction:** *"You're right — that makes total sense. We need to get our arms around this."*

**Step 6 — Your Solution (Grainger)**

Only now does the conversation turn to Grainger's specific capabilities and how they address the unplanned purchase challenge:

- Catalog breadth: Whatever part is needed, whatever the category, Grainger can supply it — eliminating the need to maintain relationships with hundreds of reactive suppliers
- Branch network + e-store: Wherever the customer is located, a Grainger branch or same-day online order eliminates the "drive across town in rush hour" scenario
- Purchase planning: Because Grainger supplies everything, it can help customers identify which historically unplanned purchases could be planned — and shift them into lower-cost bulk agreements

If it is an existing customer, Grainger uses actual purchase data to show exactly how much the customer has spent on unplanned purchases and what the projected savings would look like under a new agreement structure.

**Customer perception shift:** Grainger moves from "the place to buy $17 hammers" to "the partner that keeps us from buying $117 hammers." The result is not a pricing negotiation but a conversation about strategic partnership and long-term contract structure.

---

### SAFE-BOLD Assessment of the Grainger Pitch

| Dimension | Score | Evidence |
|-----------|-------|----------|
| Big | 4/5 | Reframes the entire category of MRO management, not just a product feature |
| Innovative | 5/5 | Planned/unplanned distinction had never been applied to MRO spend before |
| Risky | 3/5 | Requires restructuring procurement approach; moderate organizational change |
| Difficult | 4/5 | Managing unplanned spend across all MRO categories simultaneously is genuinely hard to do without a comprehensive partner |

---

## Example 2: ADP Dealer Services — Profit Clinics

**Company:** ADP Dealer Services, a division of Automatic Data Processing, providing enterprise software to automotive and vehicle dealerships.

**The business problem ADP faced:** The auto dealership market was contracting — approximately 15% of the total addressable market disappeared as dealerships closed between 2007 and 2010. Simultaneously, small competitors were winning deals with low-price, stripped-down, point solutions targeted at one part of the dealership (service center software, or sales office software alone). Dealers responded to ADP's comprehensive platform with: "That's great, but the other guy will do just the part I need for 30% less." ADP's reps were leading with their capabilities and losing on price.

**The commercial insight:** Dealers buying fragmented point solutions to save money were actually creating hidden redundancy — the average dealership worked with approximately 12 different software vendors, generating up to 40% redundant operational costs. Their software decisions to save money were costing them money.

**Unique strength anchor:** Only ADP's single-supplier, integrated platform could eliminate cross-system redundancy. Point-solution competitors, by definition, could not solve a problem caused by fragmentation — because they were the fragmentation.

---

### The Six-Step Execution

**Step 1 — Warmer**

ADP's Profit Clinic seminars open by naming the acute challenges facing automotive dealers in the current market: declining retail sales, margin compression, competitive pressure from online sales channels, and workforce efficiency concerns. Attendees confirm which pressures are most acute for them.

The meeting is framed as a free insight session about how to run a more profitable dealership — not as a product demonstration.

**Target reaction:** Dealers settle in, engage on business challenges, perceive the seminar as a "get" (valuable insight they are receiving) rather than a "give" (information they must provide to a sales process).

**Step 2 — Reframe**

*"Here is something that surprises most dealers when they first see the data: the software decisions you are making in order to save money are actually costing you money."*

This is a direct inversion of the dealer's current worldview. Dealers believed fragmenting their software purchases across multiple point-solution vendors was the cost-conscious choice. The Reframe challenges that belief at the structural level — not by arguing against any specific vendor but by questioning the logic of fragmentation itself.

**Target reaction:** *"Wait — how is that possible? I've been buying cheaper solutions specifically to cut costs."* Curiosity and mild disbelief — not agreement.

**Step 3 — Rational Drowning**

ADP presents its "Total Dealer Spend" analysis — a data-based examination of how IT fragmentation affects dealership profitability.

Key data points:
- Average dealership works with approximately 12 different software vendors
- Cross-system data entry, manual reconciliation, and workflow gaps create approximately 40% redundant operational costs
- These costs are invisible in individual vendor invoices but visible when total IT-related operational overhead is measured across the dealership

The framing: *"The line items on each vendor invoice look reasonable in isolation. The redundancy only becomes visible when you look at the full picture of what managing twelve systems actually costs your team."*

**Target reaction:** Dealers are "often deeply troubled to learn they are unnecessarily spending huge amounts of money at a time when they could least afford to do so."

**Step 4 — Emotional Impact**

ADP translates the aggregate data into a specific scenario: a day in the life of a service advisor at a multi-system dealership, navigating between disconnected platforms to complete a standard service workflow. Manual re-entry of customer data between systems. Lookups that require switching between vendor portals. A service estimate that requires information from three different software interfaces.

The scenario is drawn from real dealership observation and maps to behavior attendees recognize immediately. The emotional target: *"That is exactly what my team does. I didn't realize how much time we were losing."*

**Step 5 — A New Way**

*"Here is what a world-class dealership IT environment looks like: one integrated platform where customer data flows seamlessly from marketing to vehicle sales to service to parts. No re-entry. No reconciliation. The service advisor sees the customer's full history — what they bought, what service history they have, what current promotions apply — all in one view, without switching systems."*

Note: Still no mention of ADP. This step describes the architectural principle — integration over fragmentation — and invites dealers to agree that this is the direction they need to move.

**Target reaction:** *"That's exactly how our IT should work. That's the dealership I want to run."*

**Step 6 — Your Solution (ADP Dealer Services)**

*"That is exactly what ADP Dealer Services provides — and it is the reason we work with more of the top-performing dealerships in the industry than any other platform."*

ADP maps each element of the New Way to a specific capability of their integrated platform: unified customer database, seamless module integration, real-time reporting across all dealership functions. The key differentiator: because the integration is built into the platform rather than bolted on through third-party middleware, the redundancy costs quantified in Step 3 are structurally eliminated rather than just reduced.

The close is not a product pitch. It is an invitation to run a diagnostic: *"We can run a Total Dealer Spend analysis on your current vendor setup in about two weeks. That gives you a baseline of exactly what your IT fragmentation is costing you — and what the impact of moving to a unified platform would look like for your specific operation."*

---

### Outcome

In the year new car sales in the United States declined 40%, ADP Dealer Services revenue declined only 4%. The company not only preserved market share in a contracting market but established itself in the industry as the leading source of operational insight — earning "mind share" alongside market share.

ADP's head of sales operations noted that even after the automotive market stabilized and began recovering, the insight-led approach continued to resonate: *"Whether dealers need to survive or grow their business, they are still looking for interesting ways to better manage their operations — and that is exactly what the seminars provide."*

---

### SAFE-BOLD Assessment of the ADP Profit Clinic Pitch

| Dimension | Score | Evidence |
|-----------|-------|----------|
| Big | 5/5 | Reframes the entire IT procurement strategy of the dealership, not a single feature comparison |
| Innovative | 4/5 | "Software to save money is costing money" inverts the dealer's core assumption |
| Risky | 4/5 | Migrating from 12 vendors to one platform is a significant organizational commitment |
| Difficult | 5/5 | System consolidation at dealership scale involves vendor negotiations, data migration, staff retraining — genuinely complex |

---

## Cross-Example Patterns

Both examples share structural features worth noting when building a new pitch:

1. **The meeting is framed as insight delivery, not sales.** Neither Grainger nor ADP positions the conversation as a product demonstration. Both explicitly frame the meeting as sharing something valuable about the customer's business.

2. **The company name does not appear until late in the conversation.** In both cases, the first two-thirds of the conversation contains no company name-dropping, no logo pages, no capability lists.

3. **The Emotional Impact step uses a specific, recognizable scenario.** The Pain Chain and the service advisor scenario both draw from real customer behavior observed across many interactions. Generic scenarios do not produce the recognition response that makes this step work.

4. **The New Way is stated abstractly before the solution is named.** Both examples describe the ideal state (proactive MRO management; integrated dealership platform) before naming who delivers it. This creates buy-in to the concept before the vendor conversation begins.

5. **The close is a diagnostic, not a product demo.** Both companies offer a next step that continues to deliver insight (Grainger's purchase history analysis, ADP's Total Dealer Spend diagnostic) rather than a standard sales follow-up. The insight delivery continues past the pitch itself.

---

*Source: The Challenger Sale by Matthew Dixon and Brent Adamson, Chapter 5. All content paraphrased and structured for agent use. No verbatim passages.*

---
name: voting-system-strategist
description: "Analyze, design, or defend against voting system manipulation. Use this skill when a user needs to evaluate how a voting or election procedure will behave strategically — including which candidate or option will actually win under a given system, how an agenda-setter can engineer an outcome, whether a preference cycle makes the 'true will' of the group unknowable, how to choose the voting rule best suited to a group's goals, or when a voter should vote strategically rather than sincerely. Triggers include: user is designing a committee, board, or organizational voting process and wants to know which system is fairest or hardest to manipulate; user suspects the order of votes or the choice of voting method is being used against them; user needs to predict who wins under plurality, runoff, Condorcet pairwise, Borda count, or approval voting; user wants to know whether their group's preferences form a cycle that makes any outcome unstable; user is a voter or participant wondering whether to vote sincerely or strategically; user is analyzing a legislative, judicial, or board vote where agenda control may be shaping the outcome; user needs to apply the median voter theorem to predict where competing positions will converge; user wants to evaluate the pivotal-voter principle to understand when a single vote actually changes outcomes. This skill covers social choice theory, Arrow's impossibility theorem, the Condorcet paradox, agenda control via sequential voting, strategic (insincere) voting, comparison of voting rules, the median voter theorem, approval voting, and pivotal voter analysis. It does NOT cover simultaneous-move strategic games (use nash-equilibrium-analyzer), sequential multi-player negotiation (use other negotiation skills), or auction design."
version: 1.0.0
homepage: https://github.com/bookforge-ai/bookforge-skills/tree/main/books/the-art-of-strategy/skills/voting-system-strategist
metadata: {"openclaw":{"emoji":"📚","homepage":"https://github.com/bookforge-ai/bookforge-skills"}}
status: draft
source-books:
  - id: the-art-of-strategy
    title: "The Art of Strategy"
    authors: ["Avinash K. Dixit", "Barry J. Nalebuff"]
    chapters: [12]
tags: [game-theory, voting, social-choice, decision-making, group-decisions]
depends-on: []
execution:
  tier: 1
  mode: plan-only
  inputs:
    - type: document
      description: "Description of the voting situation: participants, options or candidates, known preference orderings or rankings, voting procedure in use or under consideration, and any agenda-setting information"
  tools-required: [Read, Write]
  tools-optional: []
  mcps-required: []
  environment: "Any agent environment; user describes the voting situation in text or structured form"
discovery:
  goal: "Identify who wins under the current or proposed voting system, flag strategic manipulation risks, detect preference cycles, recommend a voting rule aligned with the group's fairness goals, and — for voters — determine whether sincere or strategic voting is optimal"
  tasks:
    - "Classify the voting situation: election, committee vote, sequential judicial/board decision, resource allocation, or candidate positioning problem"
    - "Elicit the complete preference profile: how many voters, how many options, and each group's preference ranking from best to worst"
    - "Identify the voting rule in use: plurality, runoff, Condorcet pairwise, Borda count, approval, or quota-based"
    - "Apply the Condorcet cycle check: conduct all pairwise comparisons to detect whether majority preferences are intransitive"
    - "Analyze agenda control: if votes are sequential, apply backward induction to reveal which outcome the vote sequence produces and whether it can be engineered"
    - "Assess strategic voting incentives: identify who is a pivotal voter and whether any voter gains by misrepresenting preferences"
    - "Apply the median voter theorem where applicable: find the median preference position and assess whether it is a stable equilibrium"
    - "Recommend or evaluate the voting rule: compare plurality, runoff, Condorcet, Borda, and approval voting on fairness, manipulability, and practicality for the situation"
    - "Deliver: predicted winner under each relevant rule, identified manipulation risks, Condorcet cycle diagnosis, and a recommended design or strategy"
  audience: "Organizational leaders, committee chairs, policy designers, negotiators, board members, political analysts, voters, and anyone designing or participating in group decision-making"
  when_to_use:
    - "User is choosing or evaluating a voting system for a committee, board, election, or organization"
    - "User suspects the agenda or vote order is being manipulated against their interests"
    - "User wants to know who would actually win under different voting rules given known preferences"
    - "User is a voter wondering whether to vote sincerely or strategically when a preferred candidate is unlikely to win"
    - "User is designing rules for a process (judicial, legislative, board) and needs to understand how procedure affects outcome"
    - "User wants to predict where competing candidates or proposals will converge when positioning is strategic"
  quality:
    correctness: null
    depth: null
    actionability: null
    specificity: null
---

# Voting System Strategist

## When to Use

Use this skill whenever a group must aggregate individual preferences into a collective decision and the choice of procedure — or individual voter behavior — is itself strategic.

The foundational insight: **the outcome of a vote is determined by the voting system just as much as by the voters' preferences.** The same set of preferences can produce different winners depending on whether you use plurality voting, runoff, Condorcet pairwise comparison, Borda count, or approval voting. Anyone who controls the procedure controls significant power over the result.

**This skill applies when:**

- Three or more candidates, options, or proposals are under consideration (two-option majority vote is strategically trivial)
- You need to predict who wins, or design a system that produces fair or manipulation-resistant outcomes
- You suspect a cycle in group preferences (majority prefers A over B, B over C, but also C over A)
- Vote order, agenda setting, or sequential decisions may be shaping outcomes
- A voter is considering whether to misrepresent preferences to get a better outcome

**This skill does NOT apply to:**

- Two-option majority votes (vote sincerely; no manipulation is possible)
- Simultaneous-move strategic games among competitors (use the Nash equilibrium skill)
- Price-setting, bidding, or auction design (use auction strategy skills)
- Infinite-horizon political games where reputation and coalition dynamics dominate

---

## Context and Input Gathering

### Required (ask if missing)

- **Options and participants:** How many candidates or proposals are there? How many voters or decision-makers?
  -> Ask: "List every option being considered and every voter or voter-group with a significant block of votes."

- **Preference profile:** What is each voter or voter-group's ranking of the options from most to least preferred?
  -> Ask: "For each distinct group of voters, rank the options from best to worst. Approximate sizes are sufficient."

- **Voting rule in use:** What procedure is being used — plurality (most first-place votes wins), runoff (top two advance), sequential pairwise votes, Borda count (points by rank), approval (vote for as many as you like), or something else?
  -> Ask: "How is the winner actually determined?"

- **Agenda information (if sequential):** If votes are taken in sequence, what is the order?
  -> Ask: "What gets voted on first, and what does the loser of that vote face next?"

### Useful (gather if present)

- Whether voters have accurate information about each other's preferences (affects whether strategic voting is feasible)
- Whether any voter or group controls agenda-setting (order of votes, which options are paired)
- Whether the group is positioned on a single ideological dimension or multiple dimensions (affects median voter applicability)
- Whether the goal is designing a new system or diagnosing an existing one

---

## Execution

### Step 1 — Classify the Voting Situation

**Why:** Different voting situations call for different analytical tools. Misclassifying a sequential agenda-control problem as a simple election leads to wrong predictions. Spending one minute on classification prevents wasted analysis.

**Classification questions:**

**1a. How many options?**
- Two options: simple majority applies; no manipulation possible. This skill is not needed.
- Three or more: proceed — all the interesting strategic dynamics arise here.

**1b. Are votes simultaneous or sequential?**
- Simultaneous: all voters express preferences at once (standard elections). Analyze rule first, then strategic voting.
- Sequential: issues voted in a specified order (legislative votes, judicial panels, board approval sequences). Apply backward induction to identify how agenda order determines the outcome (Step 4).

**1c. What is the goal?**
- Predict who wins under the current system
- Diagnose a manipulation or cycle problem
- Design or recommend a better voting system
- Advise a specific voter on sincere vs. strategic voting

---

### Step 2 — Build the Preference Profile

**Why:** All voting analysis depends on knowing how voters rank options. A preference profile that seems obvious often conceals cycles or surprises when laid out systematically. Constructing it explicitly prevents errors.

**Format:** Build a table with voter groups as columns and options ranked top-to-bottom in each column. Record group sizes (as vote shares or raw counts).

**Example (Condorcet's original three-group setup):**

```
             Group L (40%)   Group M (25%)   Group R (35%)
1st choice       A               B               C
2nd choice       B               C               A
3rd choice       C               A               B
```

**Key rule:** Even if you only care about who wins under one system, complete the full ranking. Incomplete rankings hide cycles and make agenda-control analysis impossible.

---

### Step 3 — Diagnose the Preference Cycle (Condorcet Paradox Check)

**Why:** Majority preferences can be intransitive even when every individual's preferences are perfectly rational and transitive. This is the Condorcet paradox: Group prefers A over B (majority), B over C (majority), and C over A (majority). No option beats all others in pairwise comparison. When a cycle exists, there is no stable "will of the people" — any outcome can be justified, and the voting system or agenda-setter chooses who wins.

**Procedure:**

1. For every pair of options (A vs. B, A vs. C, B vs. C, etc.), count which option is preferred by a majority.
2. Construct a directed graph: draw an arrow from the winner to the loser in each pairwise matchup.
3. Check for cycles: if you can trace a loop (A beats B, B beats C, C beats A), a Condorcet cycle exists.

**Interpreting the result:**

- **No cycle, one option beats all others:** That option is the Condorcet winner. It is the most defensible choice — it would win a head-to-head election against any alternative.
- **Cycle exists:** No option is unambiguously "best." The winner will be determined by the voting system and agenda, not by voter preferences alone. Flag this prominently.

**Worked example (Condorcet's Revolutionary France case):**
- R vs. D: Left (40) + Right (35) = 75 prefer R; Middle (25) prefer D. R wins 75–25.
- R vs. L: Left (40) prefer R; Middle (25) + Right (35) = 60 prefer L. L wins 60–40.
- L vs. D: Left (40) + Middle (25) = 65 prefer D; Right (35) prefers L. D wins 65–35.
- Cycle: R beats D, D beats L, L beats R. No stable winner.

---

### Step 4 — Analyze Agenda Control and Sequential Votes

**Why:** When preferences cycle, the agenda-setter — the person who decides which option gets voted on first and who the loser faces next — effectively chooses the winner. This is not a minor procedural detail; it is the most powerful unrecognized form of strategic influence in committees, legislatures, and judicial panels.

**Backward induction procedure for sequential votes:**

Apply the same backward induction logic used for sequential games (see backward-reasoning-game-solver if needed):

1. Identify the last vote in the sequence. Determine which option wins that final matchup based on majority preferences.
2. Replace the final vote with its outcome. Now the penultimate vote is between the option that lost the previous step and whatever just became the "resolved" outcome.
3. Repeat — move one step back at a time — until you reach the first vote.

**Case study: Three judicial procedures, same preferences**

Three judges with this preference profile (a cycle):

```
               Judge A         Judge B         Judge C
1st choice   Death penalty    Life in prison   Acquittal
2nd choice   Life in prison   Acquittal        Death penalty
3rd choice   Acquittal        Death penalty    Life in prison
```

Three agenda sequences produce three different outcomes:

| Procedure | Vote 1 | If loses → Vote 2 | Winner |
|-----------|--------|-------------------|--------|
| Status quo (innocence/guilt first) | Guilt vs. Innocence → Guilty wins (A+B). Then: Death vs. Life → Life wins (B+C votes for life/acquittal). | Acquittal tied → | **Acquittal** (B tips) |
| Roman tradition (most serious first) | Death vs. not-Death → Death wins (A+C). Then Life vs. Acquittal → Life wins (A+B). | | **Death penalty** |
| Mandatory sentencing (sentence first) | Life vs. Death as required sentence → Life wins (A+B prefer conviction). Then: Convict vs. Acquit → Convict (A+B). | | **Life in prison** |

The same three judges with identical fixed preferences produce death, acquittal, or life in prison depending solely on vote order. **Whoever sets the agenda chooses the outcome.**

**Practical implication:** When you cannot change the preferences, change the agenda. When someone else controls the agenda, predict their preferred outcome and work backward to see what vote order would produce it.

---

### Step 5 — Compare Voting Systems

**Why:** Arrow's impossibility theorem proves no voting system can satisfy all fairness criteria simultaneously. Every system is flawed — but they are not equally flawed. The choice of system is a real design decision with real consequences.

**The five main systems:**

**Plurality (first-past-the-post)**
- Rule: The option with the most first-place votes wins.
- Problem: Spoiler effect. A third candidate drawing votes from a similar alternative can hand victory to the most opposed option. The 2000 US election: Nader's 97,488 Florida votes (majority would have chosen Gore) handed the state to Bush by 537 votes.
- Strategic incentive: Voters abandon sincere first choices for viable alternatives ("wasting a vote").
- Use when: Fast, low-stakes decisions with two dominant options and no third-candidate risk.

**Runoff (two-round)**
- Rule: If no candidate wins an absolute majority in round one, top two advance to a runoff.
- Problem: Eliminates the Condorcet winner if they come third in round one (Danton example: comes third with 25% but would beat both in pairwise). Creates strategic first-round voting by supporters of front-runners who fear the runoff.
- Example: 2002 French election — left-wing voters "naively" voted for fringe candidates in round one, eliminating Jospin (their real preference) and forcing a choice between Chirac and Le Pen in round two.
- Use when: Legitimacy requires a majority winner; strategic coordination risk is low.

**Condorcet method (pairwise)**
- Rule: Each option faces every other in a head-to-head majority vote. The option that beats all others (the Condorcet winner) is elected. Voters submit a full ranking; a computer calculates all matchups.
- Advantage: Selects the option most voters would choose in any direct comparison. Resistant to spoiler effects.
- Problem: When a cycle exists, no Condorcet winner exists — the method fails to produce a result.
- Practical implementation: Voters rank candidates once; software derives all pairwise votes from the ranking. Used successfully at Yale School of Management for teaching prizes.
- Use when: Fairness across all pairwise comparisons is the priority; group is willing to provide full rankings.

**Borda count (point-scoring)**
- Rule: Each voter ranks all options. An option gets n-1 points for each first-place vote, n-2 for second-place, etc. Highest total points wins.
- Advantage: Incorporates intensity of preference across all positions.
- Problem: Highly manipulable — strategic voters can bury a strong opponent by ranking them last, regardless of their true view.
- Use when: Selecting among many options where overall support matters and strategic manipulation risk is low (e.g., sports awards, internal committees with aligned interests).

**Approval voting**
- Rule: Each voter may vote for as many candidates as they approve of. No exclusion — voting for one person does not cost votes on others. The option with the most approval votes wins (or all above a threshold are selected).
- Advantage: Eliminates the spoiler effect entirely. Voters can express true preferences without strategic calculation in threshold-based systems. Proposed by Steven Brams and Peter Fishburn; used by many professional societies.
- Residual problem: When candidates compete for a fixed number of slots, strategic misrepresentation reappears — a voter may withhold approval from a strong candidate to help a weaker favorite.
- Use when: Multiple winners are selected, or a minimum-threshold rule applies. Especially valuable for elections with many candidates and concern about vote-splitting.

**Comparison summary:**

| System | Spoiler risk | Cycle risk | Manipulation ease | Complexity |
|--------|-------------|------------|-------------------|------------|
| Plurality | High | N/A | Low-medium | Minimal |
| Runoff | Medium | N/A | Medium | Low |
| Condorcet | None | Fails if cycle | Low | Medium |
| Borda | Low | N/A | High | Medium |
| Approval | None (threshold) | N/A | Low (threshold) | Low |

---

### Step 6 — Assess Strategic Voting Incentives

**Why:** Sincere voting — voting for your true first preference — is not always rational. When your vote may be pivotal (deciding the outcome), the strategic question is: which vote produces the best outcome for you, given what others will do? Arrow's impossibility theorem guarantees that in any voting system, some voter in some situation will have an incentive to misrepresent their preferences.

**The pivotal voter principle:**

A vote matters only when it breaks or creates a tie. If the outcome is already decided (by a large margin), your vote is a "voice" — it affects margins but not outcomes. Think about your vote as if you are the pivotal voter:

- If you are not pivotal (outcome is decided either way): vote sincerely. Express your true preference.
- If you are potentially pivotal: compare what happens if you vote sincerely vs. strategically, and choose the action that produces your preferred outcome.

**Strategic voting procedure:**

1. Identify your true preference ranking (best to worst).
2. Identify the realistic candidates (options with a realistic chance of being pivotal).
3. Determine which realistic candidate you most prefer.
4. If your most-preferred realistic candidate differs from your sincere first choice, strategic voting means supporting the realistic candidate.

**Case study: The 2000 Nader dilemma**
- Nader supporters' true ranking: Nader > Gore > Bush
- Realistic options: Gore or Bush (Nader cannot win)
- Strategic analysis: If you are in Florida (a swing state), your vote is potentially pivotal in the Gore/Bush decision. Voting for Nader means voting for the option that cannot be pivotal — throwing away influence over the Gore/Bush outcome.
- Strategic vote: Gore

**The paradox:** You can afford to vote sincerely only when your vote does not matter. When your vote does matter, you must vote strategically. Truthful expression is ethical only when it is costless.

**First-mover preference distortion:**

When voting is sequential and your stated preference influences others (foundation funding example, pre-1974 Congressional budgets), there is an incentive to overstate or distort preferences strategically:
- Acting early and committing your resources to one option forces others to cover the alternatives.
- Small actors can exercise disproportionate influence by committing first to secondary priorities, knowing large actors will cover primary needs.

---

### Step 7 — Apply the Median Voter Theorem (Where Applicable)

**Why:** When voters' preferences can be ordered along a single dimension (e.g., liberal to conservative, low-tax to high-tax, lenient to strict policy), the median voter theorem predicts where competing candidates or proposals will converge. The median position is a Nash equilibrium — any candidate who deviates loses votes.

**Conditions for applicability:**

- Preferences must be one-dimensional (can be placed on a single spectrum)
- Each voter has a single "ideal point" on that spectrum and prefers options closer to it
- At least two competing candidates or proposals are adjusting their positions to win

**The median voter theorem:**

The platform that beats all others in majority voting is the one at the median voter's ideal point — the position where exactly half the voters prefer a shift in each direction.

**Why the median is stable:**

- A voter to the left of the median: exaggerating leftward does not move the median position; only rightward pressure affects the median, which works against this voter.
- A voter at the median: their position is adopted; no incentive to distort.
- A voter to the right of the median: symmetric to left-voter case.

**Critical advantage:** The median rule is the unique voting rule (for one-dimensional preferences) where no voter has an incentive to misrepresent their position. Truthful voting is a dominant strategy for everyone.

**Limits:**
- Fails for multi-dimensional preference spaces (taxes AND social issues simultaneously). In two dimensions, no stable equilibrium may exist — preferences cycle across the plane.
- Fails when voters can strategically claim extreme positions to shift the computed average (mean-based rules invite this; median-based rules do not).

**The constitutional stability insight:**

In multi-dimensional preference spaces, simple majority rule (50% threshold) creates cyclical instability — any outcome can be overturned by a new majority coalition. The U.S. Constitution's two-thirds amendment requirement reflects a key finding: a supermajority threshold of approximately 64% or higher creates a stable position at the average of all voter preferences that cannot be beaten. This explains why constitutions require supermajorities to amend: not rigidity, but calibrated stability.

---

### Step 8 — Deliver the Analysis

Structure your output as:

**Voting system diagnosis:** [Which rule is in use; what it produces given the stated preferences]

**Condorcet check result:** [Does a Condorcet winner exist? If so, who? If a cycle exists, what does that mean for stability?]

**Predicted winner(s) by rule:** [For each relevant voting system, who wins and why]

**Agenda control risk:** [If sequential, what does backward induction reveal about the agenda's effect?]

**Strategic voting assessment:** [Who has incentive to vote strategically, and what should they do?]

**Recommendation:** [If designing a system: which voting rule best fits this group's goals and manipulation-resistance needs. If advising a voter: sincere or strategic, and how.]

**Limits:** [Where the analysis depends on preference estimates, single-dimensionality assumptions, or information about others' votes that may not hold]

---

## Key Principles

**Arrow's impossibility theorem: no perfect system exists.** No voting rule simultaneously satisfies all fairness criteria (unanimity, non-dictatorship, independence of irrelevant alternatives, transitivity of group preferences). Every system will be gamed in some scenario. The design question is which failures are least harmful for your context.

**The Condorcet paradox: group irrationality from individual rationality.** Even when every voter is perfectly rational and has transitive preferences, the group's majority preferences can cycle. A cycle means there is no objectively correct winner — only the system or agenda determines the outcome.

**Agenda control is the hidden lever.** Whoever controls the order of sequential votes controls the outcome. This is not corruption — it is mathematics. Backward induction reveals exactly how the agenda maps to the result.

**Strategic voting is rational, not cynical.** When your vote is pivotal, voting sincerely for an unwinnable option is equivalent to not voting at all on the choice that matters. The ethical case for sincere voting holds only when your vote is not pivotal.

**The median voter theorem works for one dimension, fails for two.** On a single axis (liberal/conservative), candidates converge to the median. Add a second axis and convergence disappears — no stable center exists, and any position can be attacked from a direction that assembles a winning coalition.

**Supermajority rules create stability, not just barriers.** A 64%+ threshold can prevent preference cycles in multi-dimensional spaces by making some positions unbeatable. Constitutional supermajority requirements are not mere tradition — they are an engineered solution to the instability of simple majority rule.

**A vote matters only when it creates or breaks a tie.** The vice president's tie-breaking vote is equal in power to any senator's vote, because it decides exactly the same set of outcomes — those decided by a 50–50 split. A vote's impact is not its frequency but the leverage it applies when decisive.

---

## Examples

### Example 1: Diagnosing a Condorcet Cycle (Committee Vote)

**Setup:** A product committee of 12 must choose between Feature A (performance), Feature B (usability), Feature C (integration). Preference profile:

```
          Engineering (5)   Design (4)   Business (3)
1st           A                B             C
2nd           B                C             A
3rd           C                A             B
```

**Pairwise check:**
- A vs. B: Engineering (5) prefer A; Design + Business (7) prefer B. B wins 7–5.
- B vs. C: Engineering + Design (9) prefer B; Business (3) prefer C. B wins 9–3.
- A vs. C: Engineering + Business (8) prefer A; Design (4) prefer C. A wins 8–4.

**Result:** B beats A and C. No cycle. B is the Condorcet winner — the option that would win any head-to-head vote. Recommend Feature B regardless of voting system.

---

### Example 2: Agenda Control in a Board Vote

**Setup:** A board of 7 must choose among Status Quo (S), Proposal A, and Proposal B. The board chair can set the vote order. Preference cycle: A beats S, S beats B, B beats A.

**Backward induction by agenda:**

- Agenda 1: A vs. B first → A wins. Then A vs. S → S wins. **Winner: S**
- Agenda 2: A vs. S first → A wins. Then A vs. B → B wins. **Winner: B**
- Agenda 3: S vs. B first → S wins. Then S vs. A → A wins. **Winner: A**

**Diagnosis:** The chair can produce any of the three outcomes by choosing the agenda. If the chair prefers A, use Agenda 3. If the chair prefers B, use Agenda 2.

**Countermeasure:** If you are not the agenda-setter and you see a cycle, identify which option the agenda is designed to produce by tracing backward induction on the proposed vote order. If it does not serve you, propose a different vote order — or propose a voting system change (Condorcet pairwise) that eliminates agenda sensitivity.

---

### Example 3: Strategic Voting Decision

**Setup:** A primary election with three candidates: your preferred candidate X (progressive, 15% polling), candidate Y (moderate, 45%), and candidate Z (conservative, 40%). You prefer X > Y > Z.

**Analysis:**
- Is X viable? No — 15% cannot win.
- Are you potentially pivotal in Y vs. Z? Yes — this is close.
- What is your pivotal choice? Y vs. Z, and you strongly prefer Y.
- Strategic vote: Y.

**Why sincere voting hurts you:** Voting for X means your vote does not participate in the Y vs. Z decision. If Y loses to Z by a small margin, your sincere vote contributed to your worst outcome.

**The paradox stated plainly:** It is only okay to vote sincerely (for X) when the election is not close — when your vote will not matter anyway. The moment your vote matters, strategic voting for your best viable option is the rational choice.

---

### Example 4: Choosing a Voting System for an Organization

**Setup:** A professional society with 200 members needs to elect 3 people to an advisory board from 12 nominees. Currently uses plurality with each voter casting 3 votes.

**Problem diagnosis:** With 12 candidates and 3 votes each, coordinated blocs can dominate — a minority with focused votes can sweep all three seats. The Joe DiMaggio effect: the obvious strongest candidate gets abandoned by strategic voters who "know they're safe" and redirect votes to favorites who need help. Result: strongest candidate sometimes fails to be elected.

**System options:**

- Plurality (current): High manipulation risk from vote concentration. Weakest option.
- Approval voting: Each voter approves as many of the 12 as desired; top 3 by approval votes win. Eliminates strategic vote-rationing — no cost to approving a strong candidate. Best fit for threshold-based selection with many candidates.
- Condorcet pairwise: Elects the option that beats all others head-to-head. Not well-defined for selecting 3 of 12 simultaneously; complex to implement.
- Borda count: Full ranking of all 12; points assigned by rank. More expressive but highly manipulable — voters can bury strong opponents.

**Recommendation:** Approval voting. Approving a deserving candidate never hurts them. Strategic misrepresentation requires complex reasoning about competitors' mutual approval rates — unlikely to be widespread. Implement with a threshold (e.g., elected if approved by >50% of voters) or seat-limit (top 3 by approval count).

---

## References

- `references/voting-systems-comparison.md` — Detailed rules, examples, fairness criteria, and failure modes for plurality, runoff, Condorcet, Borda, and approval voting
- `references/condorcet-cycle-detection.md` — Step-by-step pairwise comparison procedure, cycle diagnosis, worked examples with three and four candidates
- `references/agenda-control-backward-induction.md` — Applying backward induction to sequential vote sequences; all-outcomes-by-agenda table construction; legislative and judicial examples
- `references/arrows-impossibility-theorem.md` — The five fairness criteria, formal statement of impossibility, and practical implications for voting system design
- `references/median-voter-theorem.md` — One-dimensional convergence proof, strategic preference exaggeration under mean-based rules, multi-dimensional failure, constitutional supermajority insight

## License

This skill is licensed under [CC-BY-SA-4.0](https://creativecommons.org/licenses/by-sa/4.0/).
Source: [BookForge](https://github.com/bookforge-ai/bookforge-skills) — The Art of Strategy by Avinash K. Dixit, Barry J. Nalebuff.

## Related BookForge Skills

This skill is standalone. Browse more BookForge skills: [bookforge-skills](https://github.com/bookforge-ai/bookforge-skills)

FILE:references/agenda-control-backward-induction.md
# Agenda Control and Backward Induction in Sequential Votes

How to apply backward induction to sequential voting sequences, construct all-outcomes-by-agenda tables, and identify or counter agenda manipulation.

---

## Why Agenda Order Determines Outcomes

When preferences cycle (a Condorcet cycle exists), there is no option that beats all others. This means:

- For any option X, there exists some option Y that beats X in a pairwise majority vote
- For any option Y, there exists some option Z that beats Y
- And so on, cycling back to X

The agenda-setter exploits this by choosing which pair votes first. The loser of each vote is eliminated; the winner faces the next option. By controlling the sequence, the agenda-setter can engineer any option to be the final winner.

**This is not corruption — it is mathematics.** The agenda-setter is applying backward induction. Any participant who knows preference profiles can do the same analysis to predict or counter the manipulation.

---

## Backward Induction Procedure for Sequential Votes

This is the same backward induction used in sequential game analysis (see backward-reasoning-game-solver), applied to vote sequences.

### Step 1: Map the agenda sequence

Identify the vote order: which two options are voted on first, what the loser is eliminated from, and who the winner faces next.

Example with three options (A, B, C) and agenda "A vs. B first, winner faces C":

```
Round 1: A vs. B → winner
Round 2: [Round 1 winner] vs. C → final winner
```

### Step 2: Resolve the last vote first

Using the preference profile, determine who wins the final matchup (Round 2 in the example). This is known because all pairwise outcomes are determined by majority preference.

### Step 3: Fold backward to the first vote

Now that you know what winning Round 1 leads to (either the Round 1 winner or C in the final), replace Round 2 with its result. Round 1 becomes a choice between: "what happens if A wins Round 1" vs. "what happens if B wins Round 1."

If voters look ahead (as sophisticated decision-makers do), they vote in Round 1 based on which final outcome they prefer — not necessarily based on their honest A-vs-B preference.

### Step 4: Identify the actual winner

The option that emerges from this backward induction analysis is the predicted winner under sophisticated voting. Under naive voting (voters ignore future rounds), use the direct pairwise results.

---

## All-Outcomes-by-Agenda Table

For three options and a cycle, construct this table to show what each agenda produces.

**Three options with cycle:** A beats B, B beats C, C beats A.

| Agenda sequence | Round 1 | Round 2 | Naive winner | Sophisticated winner |
|----------------|---------|---------|-------------|---------------------|
| A vs. B → winner vs. C | A beats B | [A] vs. C → C wins | C | C |
| A vs. C → winner vs. B | C beats A | [C] vs. B → B wins | B | B |
| B vs. C → winner vs. A | B beats C | [B] vs. A → A wins | A | A |

**Reading the table:** The option placed last (facing the round-2 winner) always wins in a complete cycle under naive voting. The "protected" option — the one that doesn't have to win in round 1 — is the ultimate victor.

**For agenda-setters:** If you want option X to win, construct the agenda so X faces the final opponent in the last round, having been protected through the early votes.

**For agenda-challengers:** If you suspect manipulation, identify which option is being "protected" by the proposed agenda. Then propose an alternative agenda that protects your preferred option instead.

---

## Case Study: Three Judicial Procedures

This is the Pliny the Younger problem from Chapter 12. Three judges, three possible outcomes (Death penalty, Life in prison, Acquittal), with a preference cycle.

**Preference profile:**

```
               Judge A          Judge B         Judge C
1st choice   Death penalty   Life in prison    Acquittal
2nd choice   Life in prison  Acquittal         Death penalty
3rd choice   Acquittal       Death penalty     Life in prison
```

**Pairwise results:**
- Death vs. Life: A prefers Death (1), B prefers Life (1), C prefers Death (1). Death wins 2–1.
- Death vs. Acquittal: A prefers Death (1), B prefers Acquittal (1), C prefers Acquittal (1). Acquittal wins 2–1.
- Life vs. Acquittal: A prefers Life (1), B prefers Life (1), C prefers Acquittal (1). Life wins 2–1.

**Cycle:** Death beats Life, Life beats Acquittal, Acquittal beats Death.

**Agenda analysis (sophisticated voters who look forward and reason backward):**

Procedure 1 — Status Quo (guilt/innocence first, then sentencing):
- Stage 2 if guilty: Death vs. Life → Death wins (A+C prefer Death over Life). No wait — check: A prefers Death, C prefers Death over Life? C's ranking: Acquittal > Death > Life. So C prefers Death over Life. Death wins 2–1.
- Stage 1: Guilt (leading to Death) vs. Innocence (Acquittal). A prefers Death > Acquittal → votes Guilty. B prefers Acquittal > Death → votes Innocent. C prefers Acquittal > Death → votes Innocent. Acquittal wins 2–1.
- **Sophisticated winner: Acquittal.**

Procedure 2 — Roman Tradition (most serious first):
- Stage 2 if Death rejected: Life vs. Acquittal → Life wins (A+B prefer Life over Acquittal). A: Life > Acquittal. B: Life > Acquittal. C: Acquittal > Life. Life wins 2–1.
- Stage 1: Death vs. [if rejected → Life]. A prefers Death over Life → votes Death. B prefers Life over Death → votes against Death. C prefers Death over Life → votes Death. Death wins 2–1.
- **Sophisticated winner: Death penalty.**

Procedure 3 — Mandatory Sentencing (sentence first, then guilt):
- Stage 2: Conviction vs. Acquittal, given that conviction means Life (if that's the set sentence). A+B prefer conviction (Life) over acquittal. C prefers acquittal. Conviction wins 2–1.
- Stage 1: Which sentence? Death vs. Life. If Death is the sentence, C votes against conviction in Stage 2, so acquittal wins. If Life is the sentence, conviction wins. Judges look ahead: A prefers Death > Life but knows death sentence leads to acquittal. So A prefers Life sentence (leads to conviction and life) over Death sentence (leads to acquittal). B prefers Life > Acquittal, and Life sentence leads to conviction+life — B votes for Life. C prefers acquittal over both — C votes for Death sentence (knowing it leads to acquittal). Life wins 2–1.
- **Sophisticated winner: Life in prison.**

**Summary:** Same three judges, identical fixed preferences, three different outcomes from three procedures. The choice of judicial system determines the verdict.

---

## The "Love a Loathed Enemy" Pattern

A more subtle form of agenda control: committing resources early to force others to bear costs.

**Mechanism:**
1. Multiple parties share responsibility for a common priority (the "top priority" that all must fund together).
2. One party acts first, committing all their resources to their secondary priority.
3. The remaining parties are forced to bear the full cost of the common priority, leaving nothing for their own secondary priorities.

**Result:** The first mover achieves their secondary priority at the expense of the others' secondary priorities — using the shared top priority as a lever.

**Congressional budget analog (pre-1974 Budget Act):** Congress voted on individual expenditure items first. Unimportant items were approved early; by the time critical items came to a vote, the budget was nearly exhausted, and cutting them was politically impossible. The 1974 Budget Act reformed this by requiring votes on budget totals first.

**Countermeasure:** Require commitment to shared priorities before secondary priorities are addressed. Vote on budget totals before line items. Establish that shared responsibilities must be satisfied proportionally before any party can divert resources to secondary goals.

FILE:references/condorcet-cycle-detection.md
# Condorcet Cycle Detection

Step-by-step procedure for checking whether a group's majority preferences are transitive or form a cycle.

---

## What a Condorcet Cycle Is

A Condorcet cycle (also called a preference cycle or voting paradox) occurs when majority preferences are intransitive:

- Majority prefers A over B
- Majority prefers B over C
- Majority prefers C over A

This is possible even when every individual voter has perfectly transitive preferences. The paradox arises at the group level through aggregation. It was first formally described by the Marquis de Condorcet in the 18th century.

**Why it matters:** When a cycle exists, there is no option that represents "the will of the people." Any option can be defeated by another in a pairwise vote. The winner is determined entirely by the voting procedure and agenda, not by voter preferences.

---

## Detection Procedure

### Step 1: Construct the preference profile table

List all voter groups as columns. List all options as rows within each column, ranked from most to least preferred.

```
              Group 1 (size)   Group 2 (size)   Group 3 (size)
1st choice       [option]         [option]         [option]
2nd choice       [option]         [option]         [option]
3rd choice       [option]         [option]         [option]
```

### Step 2: Enumerate all pairwise matchups

For n options, there are n(n-1)/2 pairwise matchups.
- 3 options: 3 matchups (A-B, A-C, B-C)
- 4 options: 6 matchups
- 5 options: 10 matchups

### Step 3: For each matchup, count majority preference

For each pair (X vs. Y), add up the voters whose ranking places X above Y. If more than half, X beats Y in the pairwise comparison. Record the winner with an arrow: X → Y means X beats Y.

### Step 4: Build the dominance graph

Draw each option as a node. For each pairwise result, draw an arrow from the winner to the loser.

### Step 5: Check for cycles

Trace paths through the graph. If any path forms a loop (A → B → C → A), a cycle exists.

- No cycle + one node with arrows to all others: Condorcet winner exists. This is the option to elect under the Condorcet method.
- Cycle involving all options: No Condorcet winner. The outcome is entirely procedure-determined.
- Partial cycle (some options in cycle, others not): The options outside the cycle may be stable; options within the cycle have no stable majority ranking among themselves.

---

## Worked Example: The Revolutionary France Case

**Preference profile:**

```
             Left (40)   Middle (25)   Right (35)
1st choice      R            D             L
2nd choice      D            L             R
3rd choice      L            R             D
```

**Pairwise matchups:**

R vs. D:
- Left prefers R (40), Right prefers R (35): total 75 prefer R
- Middle prefers D (25): total 25 prefer D
- R beats D, 75–25

R vs. L:
- Left prefers R (40): 40 prefer R
- Middle prefers L (25), Right prefers L (35): total 60 prefer L
- L beats R, 60–40

D vs. L:
- Left prefers D (40), Middle prefers D (25): total 65 prefer D
- Right prefers L (35): 35 prefer L
- D beats L, 65–35

**Dominance graph:**

```
R → D
D → L
L → R
```

**Result:** Complete cycle. R beats D, D beats L, L beats R. No Condorcet winner exists.

**Implication:** Under any voting procedure, someone will be able to argue their preferred option should win by choosing the right comparison. The agenda-setter decides the winner.

---

## Worked Example: No Cycle (Product Committee)

**Preference profile:**

```
          Engineering (5)   Design (4)   Business (3)
1st           A                B             C
2nd           B                C             A
3rd           C                A             B
```

**Pairwise matchups:**

A vs. B: Engineering (5) prefer A; Design + Business (7) prefer B. B wins 7–5.
B vs. C: Engineering + Design (9) prefer B; Business (3) prefer C. B wins 9–3.
A vs. C: Engineering + Business (8) prefer A; Design (4) prefer C. A wins 8–4.

**Dominance graph:**

```
B → A
B → C
A → C
```

**Result:** No cycle. B beats A and B beats C — B is the Condorcet winner. B would win any head-to-head vote. Recommend B regardless of which voting system is used.

---

## Four-Candidate Cycle Check

With four options (A, B, C, D), run six pairwise matchups: A-B, A-C, A-D, B-C, B-D, C-D.

Possible outcomes:
- One option beats all three others: Condorcet winner, no cycle.
- One option loses to all three others (Condorcet loser): Can still have a Condorcet winner among the remaining three.
- Partial cycle among three: Two options may rank clearly; the third three form a cycle.
- Complete four-way cycle: Very unusual but possible.

For complex profiles, a systematic approach is to build the full dominance matrix: a grid where entry (X, Y) = 1 if X beats Y in pairwise, 0 otherwise. The Condorcet winner is the row with all 1s (except the diagonal).

---

## What To Do When a Cycle Exists

1. **Switch to a cycle-resistant rule:** Condorcet method with a tiebreaker (e.g., elect the option with the smallest maximum pairwise defeat).

2. **Use approval voting:** Approval voting does not depend on transitivity. Each voter simply approves options above their personal quality threshold.

3. **Restrict the agenda:** Identify which option the current vote sequence is engineered to produce. If it does not serve you, propose a different vote order or a system change.

4. **Accept the instability:** Acknowledge that the group's preferences are genuinely cyclical — any outcome requires a procedure-based choice, not a preference-based one. Make the choice transparent rather than pretending the system is neutral.

FILE:references/voting-systems-comparison.md
# Voting Systems Comparison

Detailed rules, fairness criteria, failure modes, and worked examples for the five main voting systems covered in Chapter 12.

---

## Fairness Criteria (Arrow's Framework)

Any voting system can be evaluated against these criteria. Arrow proved no system satisfies all of them simultaneously.

1. **Unanimity (Pareto efficiency):** If every voter prefers A over B, the group result must prefer A over B.
2. **Non-dictatorship:** No single voter's preferences always determine the group outcome regardless of others.
3. **Independence of irrelevant alternatives (IIA):** The group ranking of A vs. B should not change if a third option C is introduced or removed.
4. **Transitivity:** If the group prefers A over B and B over C, it must prefer A over C.
5. **Unrestricted domain:** The system must handle any possible combination of individual preference orderings.

Arrow's impossibility theorem: No voting system with 3+ options can satisfy all five criteria simultaneously. Every system violates at least one.

---

## Plurality (First-Past-the-Post)

**Rule:** Each voter casts one vote for their top choice. The option with the most votes wins.

**Fairness criteria violated:** Independence of irrelevant alternatives (IIA). Adding or removing a third candidate changes the outcome between the two leaders.

**Failure mode — spoiler effect:** A third candidate who cannot win draws votes from a similar candidate, handing victory to the opposing side. The 2000 US election is the textbook case: Nader's 97,488 votes in Florida (most would have gone to Gore) decided a race Bush won by 537.

**Strategic incentive:** Voters abandon sincere first preferences for "viable" candidates. Polls become self-fulfilling: candidates announced as long-shots lose support independent of merit.

**When to use:** Two dominant options with negligible third-candidate risk. Fast decisions where simplicity matters. Internal straw polls.

**When not to use:** Elections with three or more competitive candidates. Any context where the spoiler effect would produce an outcome most voters strongly oppose.

---

## Runoff (Two-Round)

**Rule:** All candidates compete in round one. If no candidate wins an absolute majority (>50%), the top two advance to a runoff election. The runoff winner wins the election.

**Fairness criteria violated:** Independence of irrelevant alternatives. The Condorcet winner can be eliminated in round one if they come third in first-round plurality, even though they would beat both finalists head-to-head.

**Failure mode — first-round naive voting:** Voters who "vote with their hearts" for fringe candidates in round one may accidentally eliminate their second-choice viable candidate. The 2002 French election: left-wing voters split among fringe parties in round one, eliminating Jospin (socialist, second choice for most leftists). The runoff was between Chirac and Le Pen — an outcome no left-winger wanted.

**Strategic incentive:** In round one, supporters of weak candidates should consider whether their sincere vote risks eliminating the viable candidate they prefer. Robespierre supporters in the worked example could vote strategically for Danton in round one to ensure their preferred outcome.

**When to use:** When a majority winner is required for legitimacy. When voter coordination in round one is reliable enough to avoid the Jospin failure mode.

**When not to use:** When the Condorcet winner is likely to poll third in first-round plurality (weak plurality support but strong pairwise performance). Environments where first-round strategic coordination is difficult.

---

## Condorcet Method (Pairwise Majority)

**Rule:** Each option competes against every other in a head-to-head majority vote. The option that beats all others — the Condorcet winner — is elected. In practice, voters submit a complete ranking once; a computer calculates all pairwise matchups from the single ranking.

**Fairness criteria violated:** None of Arrow's five criteria in isolation — but the method fails entirely (produces no result) when a Condorcet cycle exists.

**Advantage:** Selects the option a majority would choose in any direct comparison. No spoiler effect — adding or removing a candidate who is not the Condorcet winner does not change the result. Voters can express true preferences without strategic calculation (in the absence of cycles).

**Failure mode — Condorcet cycle:** When majority preferences are intransitive (A beats B, B beats C, C beats A), no Condorcet winner exists. The method cannot produce a result without supplementary rules (e.g., elect the option with the smallest maximum defeat).

**Practical implementation:** Rank your candidates. The computer derives all pairwise votes from your ranking. In a six-candidate election, this replaces 15 separate binary votes with a single ranked ballot. Used at Yale School of Management for annual teaching awards.

**When to use:** When fairness across all pairwise comparisons is the primary criterion. Committees or organizations willing to submit full rankings. Situations where a spoiler effect would be catastrophic.

**When not to use:** When preference cycles are likely (diverse preferences, many options). When voters are unwilling or unable to rank all options completely.

---

## Borda Count (Point-Scoring)

**Rule:** Each voter ranks all options. Option ranked 1st gets n-1 points, 2nd gets n-2 points, ..., last gets 0 points (where n = number of options). The option with the highest total points wins.

**Fairness criteria violated:** Independence of irrelevant alternatives. Adding a new candidate changes the point allocation for existing candidates.

**Advantage:** Captures intensity of preference across all positions. A candidate widely regarded as second-best by everyone may beat a candidate who is first choice for some and last choice for others — this may better reflect overall support.

**Failure mode — strategic burial:** Voters can improve their preferred candidate's relative standing by ranking a strong opponent last, regardless of their true view. This is both common and effective, making Borda count among the most manipulable systems.

**Strategic incentive:** High. Any voter who knows others' preferences has an incentive to rank their preferred candidate first and their main competitor last.

**When to use:** Sports awards, internal committee rankings where strategic manipulation is unlikely (aligned interests, small groups, social norms against gaming). Academic awards with professional norms.

**When not to use:** Political elections or competitive organizational contexts. Any environment where participants have strong incentives to misrepresent preferences.

---

## Approval Voting

**Rule:** Each voter may vote for (approve of) as many candidates as they wish. Voting for one candidate does not cost votes on any other. The candidate with the most approval votes wins (threshold variant: all candidates above a set percentage are elected).

**Fairness criteria violated:** Independence of irrelevant alternatives in quota/competitive contexts (see failure mode below).

**Core advantage:** Eliminates the spoiler effect in threshold-based elections. Approving a deserving candidate never hurts them — there is no cost to honest expression of support for multiple candidates. Voters need not consider electability when casting approval votes.

**Failure mode — competitive quota context:** When exactly n seats are filled by top-n vote-getters, candidates compete with each other indirectly in voters' minds. A voter who approves both a strong candidate (who will get in anyway) and a borderline candidate inadvertently helps the borderline candidate as much as the strong one. If the voter cares about the composition of the group (e.g., "not two sluggers in the same year"), they may withhold approval from a deserving candidate to affect the slate composition.

**The DiMaggio problem:** In a quota election with Joe DiMaggio (obvious winner), Marv Throneberry, and Bob Uecker, a voter who knows DiMaggio is safe may give both votes to a weaker candidate they prefer — concentrating votes on borderline candidates at the expense of the strongest. If all voters reason this way, DiMaggio is shut out.

**When to use:** Elections where all candidates meeting a quality threshold should be selected (Baseball Hall of Fame model, professional society boards). Elections with many candidates and concern about vote-splitting. Any context where the spoiler effect is the main problem and quota competition is low.

**When not to use:** Competitive fixed-slot elections where voters have strong preferences about the composition of the winning group, not just the quality of individual winners.

---

## System Selection Guide

| Goal | Recommended system |
|------|-------------------|
| Eliminate spoiler effect | Condorcet or Approval |
| Require majority winner | Runoff |
| Capture overall support across ranks | Borda (low-manipulation environment only) |
| Select multiple winners from large field | Approval (threshold rule) |
| Resistance to preference distortion | Condorcet (no cycle) or Approval (threshold) |
| Simplicity | Plurality |
| Strategic robustness for political elections | Condorcet |

---
name: strategic-situation-analyzer
description: "Classify any strategic situation and route to the right game-theory skill. Use this skill whenever a user describes any situation involving multiple decision-makers whose outcomes depend on each other's choices. Triggers include: user says 'I'm not sure how to think about this strategically'; user faces a competitive or cooperative decision and doesn't know where to start; user asks which game theory concept applies to their situation; user describes a negotiation, competition, auction, vote, or incentive design problem and wants to know the right framework; user asks 'is this a prisoners' dilemma?'; user wants to understand whether their situation calls for cooperation or competition; user has a business, political, or personal strategic dilemma and needs a diagnostic before diving into analysis; user says 'what kind of game am I playing?'; user describes any interaction where their best action depends on what others will do; user is unsure whether to look for dominant strategies, equilibria, or use backward reasoning; user needs to decide whether to move first or second; user wonders whether they should cooperate, compete, randomize, commit, signal, or negotiate. This is the ENTRY POINT skill for the entire Art of Strategy skill set. It diagnoses the game type and routes to the specialized skill best suited to the situation. It does NOT replace the specialized skills — it prepares the user to use them effectively."
version: 1.0.0
homepage: https://github.com/bookforge-ai/bookforge-skills/tree/main/books/the-art-of-strategy/skills/strategic-situation-analyzer
metadata: {"openclaw":{"emoji":"📚","homepage":"https://github.com/bookforge-ai/bookforge-skills"}}
status: draft
source-books:
  - id: the-art-of-strategy
    title: "The Art of Strategy"
    authors: ["Avinash K. Dixit", "Barry J. Nalebuff"]
    chapters: [1, 4]
tags: [game-theory, strategy, decision-making, strategic-analysis, situation-classifier]
depends-on: [backward-reasoning-game-solver, nash-equilibrium-analyzer]
execution:
  tier: 1
  mode: full
  inputs:
    - type: document
      description: "Description of the strategic situation: who is involved, what they can do, what they want, and how their choices interact"
  tools-required: [Read, Write]
  tools-optional: []
  mcps-required: []
  environment: "Any agent environment; user describes their strategic situation in text"
discovery:
  goal: "Identify the type of strategic game the user is in, classify it along the key structural dimensions (sequential vs. simultaneous, zero-sum vs. non-zero-sum, one-shot vs. repeated), name the game type if it matches a known pattern, and route to the correct specialized skill with targeted guidance"
  tasks:
    - "Elicit the game structure: players, moves, payoffs, information, and timing"
    - "Classify as sequential, simultaneous, or mixed"
    - "Classify as zero-sum or non-zero-sum"
    - "Identify one-shot vs. repeated interaction"
    - "Match to a named game type: prisoners' dilemma, chicken, coordination, battle of sexes, assurance/stag hunt, or other"
    - "Route to the appropriate specialized skill with a specific briefing on what to bring to it"
    - "Flag multiple applicable skills when the situation has more than one relevant dimension"
  audience: "Anyone facing a strategic situation and unsure which analytical tool to apply — managers, negotiators, policy designers, analysts, students, or individuals in personal strategic decisions"
  when_to_use:
    - "User is at the start of a strategic analysis and does not yet know which game-theory framework applies"
    - "User describes a situation and asks 'what kind of game is this?'"
    - "User wants to know whether to cooperate, compete, commit, signal, or negotiate"
    - "User needs a diagnostic before investing time in deeper game-theoretic analysis"
    - "User is unsure whether their situation is sequential, simultaneous, or mixed"
  quality:
    correctness: null
    depth: null
    actionability: null
    specificity: null
---

# Strategic Situation Analyzer

## When to Use

Use this skill as the **starting point** for any strategic analysis. A game is a situation of strategic interdependence: the outcome of your choices depends on the choices of others acting purposefully. Before applying any specialized technique, you must know which kind of game you are in.

The five rules that govern all strategic games (from Dixit and Nalebuff's Part I Epilogue):

- **Rule 1:** Sequential games — look forward and reason backward (backward induction)
- **Rule 2:** Simultaneous games — check for dominant strategies first
- **Rule 3:** No dominant strategy — eliminate dominated strategies successively
- **Rule 4:** No dominant or dominated strategies — find Nash equilibrium (mutual best responses)
- **Rule 5:** Zero-sum game with no pure equilibrium — mix strategies randomly

This skill diagnoses which rules apply to your situation, names the game type, and routes you to the specialized skill that implements the right rule. It does NOT perform the deep analysis itself — the specialized skills do that. Think of this as triage: fast, accurate classification that saves the user from applying the wrong framework.

---

## Context and Input Gathering

### Required (ask if missing)

- **Players:** Who are the decision-makers? (Include silent players — a passive institution, a market, or a future version of yourself all count as players)
  -> Ask: "Who are the active decision-makers whose choices affect your outcome?"

- **Moves and timing:** Does each player choose before or after seeing the other's choice?
  -> Ask: "When you make your decision, do you already know what the other party has done — or are you choosing at the same time?"

- **Payoffs and interests:** Does one player's gain require another's loss, or can both benefit?
  -> Ask: "If you get what you want, does that necessarily hurt the other side, or could both of you end up better off?"

- **Time horizon:** Is this a one-time interaction or an ongoing relationship?
  -> Ask: "Will you interact with this person or organization again? How often?"

### Useful (gather if present)

- The specific domain (business negotiation, competitive bidding, collective decision, contract design, information asymmetry)
- Whether the user wants to change the rules of the game rather than play within them
- Whether trust, reputation, or credibility is at stake
- Whether there is a hidden-information component (one party knows something the other doesn't)

---

## Execution

### Step 1 — Identify the Players, Moves, and Payoffs

**Why:** Game theory requires knowing exactly who is playing, what they can do, and what they want. Without these three elements, any analysis is guesswork. The key insight from the Introduction: "The key lesson of game theory is to put yourself in the other player's shoes." You cannot do this without first knowing who the other players are and what motivates them.

**Three questions to answer:**

1. **Players:** List every decision-maker whose choices affect the outcome — including those who may seem passive. A regulator, a future customer, or your own future self (as in commitment problems) may be a player.

2. **Moves:** What actions are available to each player? Are these actions chosen once (one-shot) or repeatedly? Are there actions that change the rules of the game itself (commitments, threats, promises)?

3. **Payoffs:** What does each player ultimately care about? Be precise: a competitor may care about relative standing, not absolute profit. A negotiating counterpart may value fairness or face-saving alongside money.

**Output of this step:** A compact statement of the form: "There are [N] players: [names/roles]. [Player A] can [actions]. [Player B] can [actions]. [Player A] prefers [outcomes ranked]. [Player B] prefers [outcomes ranked]."

---

### Step 2 — Classify: Sequential vs. Simultaneous vs. Mixed

**Why:** This is the single most important structural distinction in game theory. Sequential and simultaneous games require completely different analytical tools. Applying backward induction to a simultaneous game or Nash equilibrium to a purely sequential game yields wrong answers.

**Decision rule:**

| Timing Structure | Definition | Analytical Tool |
|---|---|---|
| **Sequential** | Players move in turns; each player observes previous moves before choosing | Backward induction (Rule 1) |
| **Simultaneous** | Players choose at the same time, or without observing the other's current move | Dominant strategies → dominated strategy elimination → Nash equilibrium (Rules 2-4) |
| **Mixed** | Some stages are sequential, others simultaneous | Combine both tools: solve simultaneous sub-games with Nash equilibrium, substitute results into the sequential tree, then apply backward induction |

**Diagnostic questions:**
- "By the time you make your decision, will you know what the other party has already decided?" → If yes: sequential. If no: simultaneous.
- "Is there a first-mover advantage here?" → Likely sequential.
- "Would you want to go first or second?" → If the answer is obvious and asymmetric, this is likely sequential.
- "Does your best choice depend on what you *think* the other party will do simultaneously?" → Likely simultaneous.

**Football example (mixed game):** The offense and defense choose run/pass simultaneously (neither knows the other's call), but the play call game sits inside a larger sequential game of drive management and clock strategy. Both tools are required.

---

### Step 3 — Classify: Zero-Sum vs. Non-Zero-Sum

**Why:** This classification determines whether cooperation is even theoretically possible. In a zero-sum game, every gain for one player is an exact loss for another — there is no pie to expand, only to divide. In a non-zero-sum game, there are zones of mutual benefit or mutual harm that strategic choices can navigate. Treating a non-zero-sum game as zero-sum leaves value on the table. Treating a zero-sum game as non-zero-sum invites exploitation.

**Decision rule:**

| Payoff Structure | Definition | Implication |
|---|---|---|
| **Zero-sum** | One player's gain = another's loss. Total payoff is constant regardless of outcome. | Pure competition. No cooperative outcome exists. Mix strategies if no pure equilibrium. |
| **Non-zero-sum** | Players' interests partly conflict, partly align. Outcomes exist that are better (or worse) for all. | Cooperation may be possible. Look for jointly beneficial moves, and be alert to coordination failures and collective action problems. |

**Diagnostic questions:**
- "If the other party wins completely, do you lose exactly what they gain?" → Zero-sum.
- "Is there any outcome where both of you are better off than if you simply competed?" → Non-zero-sum.
- "Could you both lose simultaneously?" → Non-zero-sum (mutual loss is only possible when interests are not strictly opposed).

**Common error:** Many real-world situations that feel like competitions are actually non-zero-sum. Arms races, price wars, and litigation are non-zero-sum games where both parties can lose — which means cooperation-based resolutions may be available.

---

### Step 4 — Match to a Named Game Type

**Why:** Named game types are cognitive anchors. Recognizing that you are in a prisoners' dilemma immediately tells you the structure of the problem and the range of solutions. The ten tales in Chapter 1 illustrate that the same structural patterns appear across radically different domains — campaign finance, criminal interrogations, collective action problems, and corporate negotiations are all structurally identical.

**Game type reference:**

| Game Type | Structure | Signature | Example |
|---|---|---|---|
| **Prisoners' Dilemma** | Simultaneous, non-zero-sum. Each player has a dominant strategy that leads to mutual harm. Cooperation is individually irrational but collectively better. | "Both of us defect even though we'd both prefer to cooperate" | Price wars, arms races, campaign spending, confession under interrogation |
| **Chicken** | Simultaneous, non-zero-sum. Two players race toward a collision; the one who swerves loses face but avoids disaster. Both swerving is best collectively; neither swerving is catastrophic. | "Someone has to back down — but who?" | Nuclear brinkmanship, labor negotiations at impasse, road rage |
| **Coordination Game** | Simultaneous, non-zero-sum. Multiple equilibria exist; players just need to coordinate on the same one. Pure alignment of interest — both prefer the same equilibrium once selected. | "We both want to do the same thing, but which thing?" | Driving on the left vs. right, technology standards, meeting-point problems |
| **Battle of Sexes** | Simultaneous, non-zero-sum. Multiple equilibria; players prefer to coordinate but each prefers a different equilibrium. Conflict over which coordination point to reach. | "We both want to meet, but we each prefer our own venue" | Joint ventures with competing HQ preferences, co-authorship credit disputes |
| **Assurance / Stag Hunt** | Simultaneous, non-zero-sum. Cooperation pays off big if both cooperate, but defection is safe if uncertain about the other. Trust is the barrier. | "I'll cooperate if I can trust you will too" | Supply chain partnerships, open-source contributions, international environmental agreements |
| **Pure zero-sum / matching pennies** | Simultaneous, zero-sum. No pure Nash equilibrium; any predictable strategy is exploitable. Must randomize. | "Whatever I do, you have a counter" | Rock-paper-scissors, penalty kicks, IRS audit targeting |
| **Sequential commitment game** | Sequential, non-zero-sum. First mover can lock in advantage by credibly committing to a strategy before the other moves. | "If I can commit first, I win" | Market entry deterrence, collective bargaining postures, international negotiation |

**Matching procedure:**
1. Is it zero-sum? → If yes, skip to mixed-strategy logic or zero-sum sequential analysis.
2. Is it simultaneous? → Check payoff structure against the prisoners' dilemma, chicken, coordination, battle of sexes, and assurance patterns.
3. Is it sequential? → Determine if commitment or credibility is the key issue.
4. Does one party have private information the other lacks? → Flag information asymmetry as a separate dimension.

---

### Step 5 — Route to the Specialized Skill

**Why:** Each specialized skill in this set is optimized for one game type or strategic problem. Sending the user to the wrong skill wastes time and produces misdirected analysis. The routing table below maps the game classification to the correct skill, with specific guidance on what to bring to the skill.

**Routing table:**

| Situation | Primary Skill | What to bring to it |
|---|---|---|
| Sequential game — need the optimal opening move and full strategy | `backward-reasoning-game-solver` | Players, move order, actions at each node, terminal payoffs or preference rankings |
| Simultaneous game — need to find equilibrium or dominant strategies | `nash-equilibrium-analyzer` | Players, strategies available to each, payoff matrix or ranking of outcomes |
| Prisoners' dilemma — cooperation problem, repeated interaction | `prisoners-dilemma-resolver` | Whether the game is one-shot or repeated, who the players are, what defection and cooperation look like in this context |
| Information asymmetry — one party knows something the other doesn't | `information-asymmetry-strategist` | Who has private information, what actions they could take to signal or hide it, what the uninformed party could do to screen |
| Auction or competitive bidding | `auction-bidding-strategist` | Auction format (open/sealed), number of bidders, whether values are private or shared, your value estimate |
| Voting, collective decision, or agenda control | `voting-system-strategist` | Number of voters, voting rule, preference rankings of key voters, who controls the agenda |
| Need to change the game — commitment, threats, promises | `strategic-commitment-designer` | The current game structure, what commitment or threat is being considered, whether it is credible given the mover's incentives |
| Negotiation, bargaining, or deal-making | `negotiation-strategist` | Both parties' alternatives to agreement (BATNAs), the zone of possible agreement, what is being divided and what can be traded |
| Incentive design — principals, agents, moral hazard | `incentive-scheme-designer` | Who the principal and agent are, what actions the agent can take that the principal cannot observe, what the principal wants the agent to do |

**Multiple-skill situations:** Many real situations involve more than one dimension. Common combinations:
- Negotiation with information asymmetry → use `negotiation-strategist` + `information-asymmetry-strategist`
- Sequential game where the first move is a credible commitment → use `strategic-commitment-designer` first, then `backward-reasoning-game-solver`
- Prisoners' dilemma in a repeated relationship → use `prisoners-dilemma-resolver` (it handles both one-shot and repeated cases)
- Auction with information asymmetry (winner's curse) → use `auction-bidding-strategist` (it covers this directly)

---

### Step 6 — Deliver the Diagnosis and Routing

Structure your output as:

**Game classification:**
- Timing: [Sequential / Simultaneous / Mixed]
- Payoff structure: [Zero-sum / Non-zero-sum]
- Time horizon: [One-shot / Repeated]
- Named game type: [Prisoners' dilemma / Chicken / Coordination / Battle of sexes / Assurance / Zero-sum mixing / Sequential commitment / Other]

**Key structural insight:** [The one observation about this game's structure that most shapes the strategic options — e.g., "Both players have dominant strategies that lead to mutual harm, so this is a prisoners' dilemma; the solution must change the payoff structure or introduce repetition"]

**Primary skill to use:** [Skill name] — [One sentence on why this skill is the right one]

**What to bring to it:** [Specific inputs the next skill will need, drawn from what the user has already described]

**Secondary skills (if applicable):** [Any additional skills for dimensions of the situation not covered by the primary skill]

**One thing to watch for:** [The most common error or trap in this type of game, stated concisely]

---

## Key Principles

**Diagnosis before prescription.** Applying a technique before classifying the game type is a common and costly error. The prisoners' dilemma and a coordination game are both simultaneous non-zero-sum games, but their solutions are almost opposite. Classification is not overhead — it is the analysis.

**Put yourself in the other player's shoes.** The core discipline of game theory is modeling the other player accurately: what they know, what they want, and how they will reason. George Bernard Shaw's warning applies: do not treat others as you would want to be treated — their preferences may differ from yours.

**You may be playing a larger game.** The taxi story (Tale #10) illustrates that the immediate transaction may be embedded in a reputation game, a social game, or a future-interaction game. Always ask: what larger game does this interaction sit inside?

**The game type determines the solution, not vice versa.** Do not start with a preferred solution and fit the game type to it. Start with the game type (classification determines what solutions are even theoretically available) and then find the best solution within that type.

**Non-zero-sum games have traps that zero-sum thinking misses.** The prisoners' dilemma is dangerous precisely because it feels like pure competition while actually containing a cooperate-cooperate outcome that both players prefer. Treating it as zero-sum guarantees the bad equilibrium.

**You may not be playing the game you think you are.** Buffett's campaign finance dilemma (Tale #7) looks like a cooperation problem but is actually a clever prisoners' dilemma construction where supporting the bill is a dominant strategy for both parties. Re-examine the payoff structure before committing to any classification.

---

## The Ten Tales: Game Type Reference

Chapter 1 illustrates the breadth of strategic situations. Each tale maps to a game type:

| Tale | Domain | Game Type |
|---|---|---|
| #1: Pick a Number | Adversarial search | Sequential game of expectations; backward induction on predicted opponent behavior |
| #2: Winning by Losing | Survivor | Sequential game; backward reasoning from finale structure |
| #3: The Hot Hand | Sports/defense | Simultaneous mixed-strategy game; zero-sum between offense and defense |
| #4: To Lead or Not | Sailboat racing | Sequential commitment; copying strategy vs. innovation |
| #5: Here I Stand | Negotiation | Sequential commitment; credibility and intransigence as power |
| #6: Thinning Strategically | Self-commitment | Commitment game against your future self; removing options to gain credibility |
| #7: Buffett's Dilemma | Campaign finance | Prisoners' dilemma; dominant strategies leading to mutual harm |
| #8: Mix Your Plays | Rock Paper Scissors | Zero-sum simultaneous; no pure equilibrium, must randomize |
| #9: Never Give a Sucker an Even Bet | Auctions/betting | Information asymmetry; winner's curse |
| #10: Game Theory Can Be Dangerous | Negotiation/taxi | Bargaining; importance of modeling the other player's perspective and payoffs |

---

## References

- `references/game-type-field-guide.md` — Detailed criteria and worked examples for each named game type; disambiguation rules for borderline cases
- `references/five-rules-quick-reference.md` — One-page summary of the five rules from the Part I Epilogue with decision flow and tool mapping
- `references/situation-to-skill-routing-guide.md` — Extended routing logic including multi-skill combinations and common situational patterns

## License

This skill is licensed under [CC-BY-SA-4.0](https://creativecommons.org/licenses/by-sa/4.0/).
Source: [BookForge](https://github.com/bookforge-ai/bookforge-skills) — The Art of Strategy by Avinash K. Dixit, Barry J. Nalebuff.

## Related BookForge Skills

Install related skills from ClawhHub:
- `clawhub install bookforge-backward-reasoning-game-solver`
- `clawhub install bookforge-nash-equilibrium-analyzer`

Or install the full book set from GitHub: [bookforge-skills](https://github.com/bookforge-ai/bookforge-skills)

FILE:references/five-rules-quick-reference.md
# Five Rules Quick Reference

The five rules from the Part I Epilogue of *The Art of Strategy* (Dixit and Nalebuff). These rules are the complete decision procedure for any strategic game — sequential or simultaneous, zero-sum or non-zero-sum. Use this reference as a one-page lookup.

---

## The Five Rules

### Rule 1: Sequential games — Look forward and reason backward

**When it applies:** Players move in turns; each player observes previous moves before choosing; game ends in a finite number of moves.

**Mechanism:** Construct a game tree. Starting at the terminal nodes (where outcomes are known), fold backward by replacing each node with the optimal choice of the player who moves there. Continue until you reach the first move.

**Key insight:** Future actions by rational players are predictable, not uncertain. "Look forward" means identify what the game ultimately leads to. "Reason backward" means calculate what choice at the first move leads to that desired outcome.

**Tool:** `backward-reasoning-game-solver`

**Failure mode:** Forward reasoning — choosing the action that looks best immediately without tracing the full chain of consequences.

---

### Rule 2: Simultaneous games — Check for dominant strategies

**When it applies:** Players choose without observing the other's current move. Applied first in any simultaneous game analysis.

**Mechanism:** For each player, check whether any one strategy outperforms all other strategies regardless of what the opponent does. If a dominant strategy exists, use it. If your opponent has a dominant strategy, predict they will use it and respond accordingly.

**Key insight:** A dominant strategy eliminates the "what will they do?" problem. You don't need to predict the opponent's action — the strategy is best no matter what they do.

**Tool:** `nash-equilibrium-analyzer` (handles dominant strategy detection)

**Failure mode:** Searching for a "clever" response to anticipated opponent moves when a dominant strategy already makes that search unnecessary.

---

### Rule 3: No dominant strategy — Eliminate dominated strategies successively

**When it applies:** No player has a dominant strategy (or after dominant strategies have been identified and played, the residual game still has multiple strategies per player).

**Mechanism:** A dominated strategy is one that is uniformly worse than some other strategy, regardless of opponent choices. Remove it. After removal, re-examine the smaller game — new dominant or dominated strategies may now appear. Repeat until the game resolves or cannot be further reduced.

**Key insight:** Dominated strategies will never be played by rational players. Knowing this shrinks the game and may reveal solutions that were hidden in the full game.

**Tool:** `nash-equilibrium-analyzer` (covers iterated elimination)

**Failure mode:** Analyzing all combinations in a large game matrix without first eliminating dominated strategies, wasting effort on branches that rational players would never reach.

---

### Rule 4: No dominant or dominated strategies — Find Nash equilibrium

**When it applies:** After Rules 2 and 3 have been applied (or found not to apply), the game still has no clear solution.

**Mechanism:** A Nash equilibrium is a combination of strategies such that each player's strategy is their best response to the other players' strategies. Neither player can benefit by unilaterally deviating. Find the equilibrium by checking all remaining strategy combinations for mutual best-response.

**Key insight:** Nash equilibrium is the logical resting point of strategic reasoning — the only combination of strategies where all players are simultaneously satisfied with their choices given what the others are doing. Multiple equilibria may exist; in that case, a focal point or coordination mechanism is needed.

**Tool:** `nash-equilibrium-analyzer`

**Failure mode:** Stopping at a strategy combination without verifying it is a mutual best response — finding "a good strategy" rather than "an equilibrium strategy."

---

### Rule 5: Zero-sum game with no pure equilibrium — Mix strategies

**When it applies:** The game is zero-sum (one player's gain is exactly the other's loss) and there is no pure-strategy Nash equilibrium (which is always the case in zero-sum games like Rock-Paper-Scissors and penalty kicks where any predictable choice can be exploited).

**Mechanism:** Calculate mixed-strategy probabilities such that the opponent is indifferent between their available strategies. At a mixed-strategy equilibrium, randomizing in the right proportions makes you unpredictable and exploitable-proof.

**Key insight:** In zero-sum games with no pure equilibrium, predictability is exploitability. The goal of mixing is not to confuse — it is to make the opponent indifferent, which is the only state where neither player can improve by deviating from the mix.

**Tool:** `nash-equilibrium-analyzer` (covers mixed strategies)

**Failure mode:** Rotating strategies in a predictable pattern (thinking that "changing it up" is the same as mixing). Any detectable pattern — even a rotation — can be exploited. True mixing requires randomization.

---

## Decision Flow

```
Start: Describe the game (players, moves, payoffs)
         |
         v
Is the game SEQUENTIAL?
(players observe moves before choosing)
  |
  Yes --> Apply Rule 1: Look forward, reason backward
          --> backward-reasoning-game-solver
  |
  No  --> Game is SIMULTANEOUS
          |
          v
       Does any player have a DOMINANT STRATEGY?
         |
         Yes --> Apply Rule 2: Use it (and predict opponent uses theirs)
                 --> nash-equilibrium-analyzer
         |
         No  --> Can any strategy be ELIMINATED as dominated?
                 |
                 Yes --> Apply Rule 3: Eliminate successively
                         --> nash-equilibrium-analyzer
                 |
                 No  --> Look for NASH EQUILIBRIUM (mutual best responses)
                         |
                         Found: unique --> Apply Rule 4: Use it
                         Found: multiple --> Need focal point/coordination
                         Not found (zero-sum game) --> Apply Rule 5: Mix
                         --> nash-equilibrium-analyzer
```

---

## Combined Game Note

In practice, many games have both sequential and simultaneous stages (football, auctions with rounds, multi-stage negotiations). The approach:

1. Identify all simultaneous sub-games and solve them using Rules 2-5
2. Substitute the equilibrium payoffs from each simultaneous sub-game back into the larger sequential game tree
3. Apply Rule 1 to the full sequential structure with these substituted payoffs

This combination is why `strategic-situation-analyzer` is the entry point: it identifies the structure so the right tools are applied in the right order.

---

## What the Rules Do NOT Cover

These five rules address how to find optimal strategies within a given game. They do not address:

- **Changing the game itself** — modifying payoffs, timing, or information structure before the game is played → `strategic-commitment-designer`
- **Repeated game dynamics** — how cooperation can be sustained over time through reciprocity and reputation → `prisoners-dilemma-resolver`
- **Hidden information** — when one player knows something the other doesn't → `information-asymmetry-strategist`
- **Institutional design** — designing rules, incentives, or mechanisms for others to play within → `incentive-scheme-designer`, `auction-bidding-strategist`, `voting-system-strategist`
- **Bargaining over the surplus** — when the game produces joint value and the question is how to divide it → `negotiation-strategist`

FILE:references/game-type-field-guide.md
# Game Type Field Guide

Detailed criteria, payoff structures, worked examples, and disambiguation rules for each named game type. Use this reference when the main skill's routing table leaves ambiguity.

---

## Prisoners' Dilemma

### Structure
- Simultaneous moves
- Non-zero-sum
- Each player has a **dominant strategy** (defect/compete) that makes them better off regardless of what the other does
- The dominant-strategy equilibrium is worse for both players than if both had chosen the cooperative option

### Payoff signature (ordinal ranking per player)
Defect while other cooperates > Both cooperate > Both defect > Cooperate while other defects

The key: "Both defect" is the equilibrium even though "both cooperate" is better for both. Individual rationality produces collective irrationality.

### Diagnostic test
Ask: "If you knew for certain the other party would cooperate, would you still be tempted to defect? And if you knew they would defect, would you be forced to defect too?" If yes to both: prisoners' dilemma.

### Examples from the Ten Tales
- Buffett's Dilemma (#7): Campaign finance. Each party's dominant strategy is to support the reform bill, regardless of what the other party does. Both support it → the billionaire's mechanism works for free.
- In Cold Blood (classic illustration): Both suspects defect (confess) even though mutual silence would serve both better.
- Collective action / belling the cat: Each individual free-rides even though collective action would benefit everyone.

### Common real-world instances
- Price wars: each firm cuts prices to gain share; both end up worse
- Arms races: each side builds weapons; both end up less secure at higher cost
- Advertising spending battles: mutually escalating spend with no change in market share
- Environmental non-compliance: each firm pollutes to save costs; industry faces regulation

### Solution routes
- **One-shot:** No escape within the game. Must redesign the game (change payoffs, add enforcement) → `strategic-commitment-designer`
- **Repeated, indefinite horizon:** Tit-for-tat or trigger strategies can sustain cooperation → `prisoners-dilemma-resolver`
- **Many players (collective action):** Coordination mechanisms, selective incentives, threshold dynamics → `prisoners-dilemma-resolver`

---

## Chicken (Hawk-Dove)

### Structure
- Simultaneous moves
- Non-zero-sum
- Two Nash equilibria in pure strategies: (You swerve, They don't) and (You don't, They swerve)
- A third equilibrium in mixed strategies exists
- The worst outcome for both is mutual non-swerving (collision/catastrophe)

### Payoff signature (ordinal ranking per player)
You don't swerve, they do > Both swerve > You swerve, they don't > Neither swerves (catastrophe)

### Diagnostic test
Ask: "Is there an outcome that would be catastrophic for both of us if neither backs down?" If yes: likely chicken. The key feature is mutual catastrophe at the extreme.

### Examples
- Nuclear brinkmanship: mutual destruction is worse than either side backing down
- Labor strike negotiations: protracted dispute damages both sides; one must concede
- Corporate acquisition battles: bidding wars can destroy value for the "winning" acquirer

### Disambiguation from prisoners' dilemma
In prisoners' dilemma, both players have the same dominant strategy (defect). In chicken, there is no dominant strategy — each player's best response depends on what the other does. If you expect the other to swerve, you should hold firm. If you expect the other to hold firm, you should swerve.

### Solution routes
- Creating credible commitment to not swerve forces the other side to swerve → `strategic-commitment-designer`
- Negotiating a coordinated solution before either commits → `negotiation-strategist`

---

## Coordination Game

### Structure
- Simultaneous moves
- Non-zero-sum
- Multiple Nash equilibria; players prefer to coordinate on the same one
- Interests are perfectly aligned once a coordination point is selected

### Payoff signature
Both choose A > Both choose B > Miscoordinate (any combination)
(or: Both choose A = Both choose B > Miscoordinate — pure coordination with indifferent equilibria)

### Diagnostic test
Ask: "Would both of you be happy with any outcome where you both make the same choice, and unhappy whenever you choose differently?" If yes: coordination game.

### Examples
- Driving on the left vs. right side of the road: either convention works; miscoordination is catastrophic
- Technology standards (USB-C, Blu-ray vs. HD DVD): everyone benefits from the same standard
- Meeting a stranger in a city with no pre-arranged location: need a focal point (Schelling point)

### Disambiguation from battle of sexes
In a pure coordination game, both players are indifferent between equilibria, or equally prefer the same one. In battle of sexes, both want to coordinate but each prefers a different coordination point.

### Solution routes
- Focal points (Schelling points): shared cultural or contextual anchors that players naturally coordinate on → `nash-equilibrium-analyzer` (covers focal point selection)
- Communication and pre-play agreements
- Standards-setting institutions

---

## Battle of Sexes

### Structure
- Simultaneous moves
- Non-zero-sum
- Two Nash equilibria in pure strategies; players prefer to coordinate but each prefers a different equilibrium
- Conflict over which coordination point to reach

### Payoff signature (per player)
Your preferred outcome > Other's preferred outcome > Miscoordinate

Both players prefer either coordination point over miscoordination, but they rank the two coordination points differently.

### Diagnostic test
Ask: "Would both of you prefer to be doing the same thing, but you each prefer a different 'same thing'?" If yes: battle of sexes.

### Examples
- Opera vs. football: partners want to be together; each prefers their own venue
- Joint venture HQ location: both firms prefer to co-locate, but each prefers its own city
- API standards negotiations: both parties want compatibility, but each prefers their own protocol as the standard

### Solution routes
- Alternating coordination (trade off between preferred outcomes over time)
- Pre-commitment to your preferred equilibrium to force the other to follow → `strategic-commitment-designer`
- Negotiation with side payments → `negotiation-strategist`

---

## Assurance Game (Stag Hunt)

### Structure
- Simultaneous moves
- Non-zero-sum
- Two Nash equilibria: mutual cooperation (payoff-dominant) and mutual defection (risk-dominant)
- Unlike prisoners' dilemma, cooperation is individually rational IF the other party cooperates
- The barrier is trust and mutual assurance, not individual incentive to defect

### Payoff signature
Both cooperate > You defect, they cooperate > Both defect > You cooperate, they defect

The key distinction from prisoners' dilemma: cooperating while the other defects is the *worst* outcome for you (not just bad), which means you should only cooperate if you are confident the other will too.

### Diagnostic test
Ask: "If you were confident the other party would cooperate, would you definitely cooperate? And if you were uncertain, would you fall back to the safe, individual option?" If yes to both: assurance game.

### Examples
- International environmental agreements: cooperating on emissions cuts is good if all do it; unilateral cuts while others don't is the worst outcome
- Open-source contributions: valuable if many contribute; contributing alone while others free-ride is wasteful
- Supply chain partnerships: sharing proprietary information benefits both if both share; exposing information while the other doesn't is catastrophic

### Solution routes
- Building mutual trust through repeated interaction and observable signals
- Transparency and verification mechanisms that make cooperation observable
- Upfront commitments and credible signals of cooperative intent → `strategic-commitment-designer`
- Sequencing to reveal cooperative intent early → `prisoners-dilemma-resolver`

---

## Zero-Sum Simultaneous Game (Matching Pennies / Rock-Paper-Scissors)

### Structure
- Simultaneous moves
- Zero-sum: what one player gains, the other loses exactly
- No pure Nash equilibrium: any predictable strategy is exploitable
- Mixed-strategy Nash equilibrium exists: randomize to become unpredictable

### Diagnostic test
Ask: "Does my opponent benefit from knowing in advance what I will do?" If yes, and if one player's gain is exactly the other's loss: this is a zero-sum simultaneous game requiring mixing.

### Examples
- Rock Paper Scissors (Christie's vs. Sotheby's)
- Penalty kicks in soccer: kicker and goalkeeper choose simultaneously
- IRS audit targeting: if the formula is known, only compliant taxpayers get audited
- Bluffing in poker: if you only raise with strong hands, opponents fold every time

### Solution routes
- Calculate the mixed-strategy equilibrium probabilities from the payoff matrix → `nash-equilibrium-analyzer`
- Use randomization devices to achieve unpredictability
- Consider whether the game can be restructured to avoid the zero-sum dynamic

---

## Borderline Cases and Disambiguation

### "Is this a prisoners' dilemma or chicken?"
The test: what happens if both players choose the "hard" option (both defect vs. neither swerves)?
- Prisoners' dilemma: both defect is a **Nash equilibrium** (bad but stable)
- Chicken: both don't swerve is a **catastrophe** (not a Nash equilibrium — both would want to deviate)

### "Is this a coordination game or assurance game?"
The test: what happens if you cooperate and the other doesn't?
- Coordination game: you are unhappy (miscoordination), but not catastrophically so
- Assurance game: cooperating while the other defects is the **worst possible outcome** for you (zero payoff or negative — you shared information / resources and got nothing back)

### "Is this sequential or just asymmetric simultaneous?"
A game is sequential if Player B *observes* Player A's choice before making their own — not just if Player A happens to go first in calendar time. If B must commit without observing A's actual choice, it is simultaneous even if the decisions happen at different calendar times.

### "Is this zero-sum or just competitive?"
A competitive game is not necessarily zero-sum. A market where firms compete for share is competitive but non-zero-sum — both can grow or both can shrink together. Zero-sum means the total payoff is fixed regardless of outcome; every unit transferred is a gain for one and an equal loss for the other.

FILE:references/situation-to-skill-routing-guide.md
# Situation-to-Skill Routing Guide

Extended routing logic for the `strategic-situation-analyzer` hub skill. Covers multi-skill combinations, common situational patterns, and disambiguation between skills with overlapping applicability.

---

## Single-Skill Routes

### Route A: Pure Sequential Game
**Trigger:** Moves alternate; each player observes the previous move; game ends in finite steps.
**Skill:** `backward-reasoning-game-solver`
**Inputs to prepare:**
- The complete sequence of who moves when
- Available actions at each decision point
- Terminal payoffs or preference rankings (ordinal is sufficient)
- Whether any stages involve simultaneous choices embedded in the sequence

**Watch for:** Natural uncertainty (chance nodes like weather, market outcomes) — treat these as probability-weighted branches, not opponent moves.

---

### Route B: Simultaneous Game — Finding Equilibrium
**Trigger:** Players choose without observing each other's current action; need to find the best strategy.
**Skill:** `nash-equilibrium-analyzer`
**Inputs to prepare:**
- List of strategies available to each player
- Payoff matrix: what each player gets for each combination of choices
- Whether the game is zero-sum or non-zero-sum
- Whether the game is one-shot or repeated (repeated changes the equilibrium significantly)

**Watch for:** Multiple equilibria — bring context on any focal points or conventions that might select between them.

---

### Route C: Cooperation Under Recurring Temptation
**Trigger:** Players face repeated prisoners' dilemma-type structure; each round they are tempted to defect but mutual cooperation would serve both better long-term.
**Skill:** `prisoners-dilemma-resolver`
**Inputs to prepare:**
- The payoff structure of the one-shot game (confirm it has the prisoners' dilemma signature)
- Whether the game is repeated and, if so, for how many rounds (known end vs. indefinite)
- The players' discount rates or patience (how much they value future payoffs vs. current ones)
- Whether communication is possible before each round

---

### Route D: Hidden Information Advantage/Disadvantage
**Trigger:** One party has private information the other lacks; the uninformed party must infer or screen; the informed party may want to signal or conceal.
**Skill:** `information-asymmetry-strategist`
**Inputs to prepare:**
- Who has private information and what kind (quality, intentions, costs)
- What actions the informed party can take that might reveal or conceal their type
- What actions the uninformed party could take to distinguish between types (screening mechanisms)
- Whether signaling is costly (costly signals are more credible)

---

### Route E: Competitive Bidding or Auction
**Trigger:** Multiple parties bid for a prize; the highest bid wins; bidders have private value estimates.
**Skill:** `auction-bidding-strategist`
**Inputs to prepare:**
- Auction format (English/ascending, Dutch/descending, first-price sealed bid, second-price/Vickrey)
- Number of competing bidders (approximately)
- Whether values are private (you know only your own value) or common (the prize has a single true value; everyone has imperfect signals about it)
- Your value estimate for the prize

**Watch for:** Winner's curse in common-value auctions — if winning means all others bid less than you, it likely means you overestimated value.

---

### Route F: Collective Decision or Voting
**Trigger:** A group must reach a collective decision through voting; agenda control or voting rule design is relevant.
**Skill:** `voting-system-strategist`
**Inputs to prepare:**
- The voting rule in use (majority, supermajority, sequential elimination, approval voting)
- The number of voters and key preference rankings
- Whether any player controls the agenda (what gets voted on and in what order)
- Whether the vote is one-shot or part of a multi-round process

---

### Route G: Changing the Structure of the Game
**Trigger:** The user does not want to play the current game optimally — they want to change it. Or they need to make a threat or promise credible before the game begins.
**Skill:** `strategic-commitment-designer`
**Inputs to prepare:**
- The current game structure and its equilibrium outcome
- What commitment, threat, or promise is being considered
- Whether the commitment is reversible (limits credibility) or irreversible (credible but costly)
- What the opponent's anticipated response to the commitment would be

---

### Route H: Negotiation or Bargaining
**Trigger:** Two or more parties are trying to reach a deal that creates joint value; the question is how to get to agreement and how the surplus is divided.
**Skill:** `negotiation-strategist`
**Inputs to prepare:**
- Each party's best alternative to a negotiated agreement (BATNA)
- The range of possible deals (zone of possible agreement)
- What is being divided and what can be traded across issues
- Time pressure: whether delay costs one party more than the other

---

### Route I: Incentive Design or Contract Structure
**Trigger:** A principal wants to induce an agent to take certain actions that the principal cannot directly observe or verify.
**Skill:** `incentive-scheme-designer`
**Inputs to prepare:**
- Who the principal and agent are; what the principal wants the agent to do
- What actions the agent can take that the principal cannot observe (the moral hazard dimension)
- What outcome metrics are observable and contractible
- Whether there is an adverse selection problem (the agent has private information before contracting) as well as a moral hazard problem

---

## Multi-Skill Combinations

### Combination 1: Sequential Commitment + Backward Reasoning
**When:** The user wants to make a first-mover commitment and then predict how the game unfolds from there.
**Order:** Use `strategic-commitment-designer` first to design the credible commitment. Then use `backward-reasoning-game-solver` with the modified game tree (where the commitment changes the first-mover's available actions or the opponent's anticipation).

---

### Combination 2: Negotiation + Information Asymmetry
**When:** A negotiation where one party has private information that affects the deal terms — typical in M&A, hiring, or supply contracts.
**Order:** Use `information-asymmetry-strategist` first to understand the signaling/screening dynamics (should you reveal your information? how do you interpret their signals?). Then use `negotiation-strategist` with the information structure clarified.

---

### Combination 3: Repeated Prisoners' Dilemma + Commitment
**When:** The players are in an ongoing prisoners' dilemma relationship and the user wants to change the equilibrium — either by changing payoffs (commitment) or by establishing a reciprocity norm (repeated game).
**Which to use first:** Depends on the user's goal.
- If the goal is to change the game's structure → `strategic-commitment-designer` (changes payoffs so cooperation becomes dominant)
- If the goal is to sustain cooperation within the existing structure → `prisoners-dilemma-resolver` (tit-for-tat, trigger strategies)

---

### Combination 4: Auction + Information Asymmetry
**When:** The user is bidding in an auction where other bidders may have better information about the true value of the prize (common-value auction with information asymmetry).
**Order:** Use `auction-bidding-strategist` directly — it covers the winner's curse and bid shading for common-value auctions as part of its core content.

---

### Combination 5: Voting + Strategic Commitment
**When:** The user controls the agenda and wants to sequence votes to achieve a preferred outcome.
**Order:** Use `voting-system-strategist` first (it covers agenda manipulation and strategic voting directly). Use `strategic-commitment-designer` if the key issue is pre-committing to a position before the vote to influence other voters' choices.

---

## Disambiguation: When Two Skills Seem Applicable

### Nash Equilibrium Analyzer vs. Backward Reasoning Game Solver
**Distinguishing question:** "By the time you make your decision, will you know what the other party has already chosen?"
- Yes → `backward-reasoning-game-solver` (sequential)
- No → `nash-equilibrium-analyzer` (simultaneous)

**Common confusion:** "I move first, so isn't this sequential?" — Not necessarily. If the second party chooses *without observing your move* (as in sealed-bid auctions), it is simultaneous even though you "moved first" in calendar time.

---

### Prisoners' Dilemma Resolver vs. Nash Equilibrium Analyzer
**Distinguishing question:** "Is the problem fundamentally about mutual defection in a repeated or structured relationship, or about finding the right strategy in a one-shot game?"
- Repeated relationship with cooperation breakdown → `prisoners-dilemma-resolver`
- One-shot equilibrium analysis → `nash-equilibrium-analyzer`

**Overlap:** The one-shot prisoners' dilemma can be solved by `nash-equilibrium-analyzer` (dominant strategy = defect; Nash equilibrium = both defect). Use `prisoners-dilemma-resolver` when the goal is to actually achieve cooperation despite the dominant-strategy trap.

---

### Strategic Commitment Designer vs. Negotiation Strategist
**Distinguishing question:** "Are you trying to change the rules before the game starts, or are you trying to navigate a bargaining process that is already underway?"
- Changing the game → `strategic-commitment-designer`
- Navigating bargaining → `negotiation-strategist`

**Overlap:** Commitment tactics (making threats credible, establishing BATNAs) are part of both skills. If the commitment is purely internal to the negotiation (not a structural change to the game), use `negotiation-strategist`.

---

## Situations That Don't Fit Cleanly

### "I'm not sure if my situation is a game at all"
A game requires **strategic interdependence**: your outcome depends on others' choices, and they know your outcome depends on their choices, and you know that they know, etc. If the other party is passive (a natural environment, a machine with fixed rules, a bureaucracy with no decision-maker), this is a decision problem, not a game. Standard optimization applies.

**Test:** "Does the other party adjust their behavior in response to what I do?" If no, this is not a game theory problem.

---

### "There are more than two players"
Game theory applies to n-player games, but the analysis is more complex. For this skill set:
- Voting and collective decisions → `voting-system-strategist` (designed for multi-player settings)
- Auctions with multiple bidders → `auction-bidding-strategist`
- Multi-player prisoners' dilemma / collective action → `prisoners-dilemma-resolver`
- Multi-player negotiations → `negotiation-strategist` (covers coalition dynamics)
- Multi-player simultaneous games → `nash-equilibrium-analyzer` (handles n players)

---

### "The other party doesn't behave rationally"
Game theory assumes rational players who maximize their payoffs. When players deviate from rationality (due to emotion, pride, status concerns, or cognitive bias), pure game-theoretic analysis may be wrong. Adjustments:
- Model the non-rational payoff directly: if the other party cares about face-saving, include that in their payoff function
- Use behavioral benchmarks: the ultimatum game shows real players reject unfair offers even at personal cost
- Build in slack: if you cannot predict whether the other party will behave rationally, choose strategies that perform reasonably well under both rational and non-rational opponent behavior

**Important note from Tale #10 (the taxi):** Pride, spite, and perceived dishonor can dominate monetary calculations. Always ask: "Is this person optimizing money, or optimizing something else?"

---
name: strategic-commitment-designer
description: "Design credible strategic moves — commitments, threats, and promises — to change the game in your favor before play begins. Use this skill when a user needs to lock in a position and prevent backtracking; deter an adversary from an unwanted action; compel a counterpart to take a desired action; make a negotiation stance, policy, or business pledge actually believable; or structure incentive mechanisms that hold even when renegotiation is tempting. Triggers include: user wants to commit to a course of action in a way that others will believe; user is setting a credible deterrent threat (e.g., retaliation policy, penalty clause, price floor); user must compel action by a deadline and needs the right move type and deadline design; user suspects their threat or promise will be dismissed as a bluff; user needs to choose between issuing a threat vs. a promise for deterrence or compellence; user wants to practice brinkmanship and needs to calibrate the risk level; user is designing a contract or commitment mechanism and needs to close renegotiation loopholes; user is countering an opponent's commitment or threat. This skill covers the full taxonomy of strategic moves (commitment / threat / promise, deterrence / compellence, warnings / assurances) and all eight credibility mechanisms. It does NOT perform the underlying game tree analysis — use backward-reasoning-game-solver for that before applying this skill."
version: 1.0.0
homepage: https://github.com/bookforge-ai/bookforge-skills/tree/main/books/the-art-of-strategy/skills/strategic-commitment-designer
metadata: {"openclaw":{"emoji":"📚","homepage":"https://github.com/bookforge-ai/bookforge-skills"}}
status: draft
source-books:
  - id: the-art-of-strategy
    title: "The Art of Strategy"
    authors: ["Avinash K. Dixit", "Barry J. Nalebuff"]
    chapters: [6, 7]
tags: [game-theory, strategy, negotiation, commitment, credibility]
depends-on: [backward-reasoning-game-solver]
execution:
  tier: 1
  mode: plan-only
  inputs:
    - type: document
      description: "Description of the strategic situation: the players, the action you want to influence (prevent or compel), your current and desired positions, any existing threats or promises in play, and context about your ability to follow through"
  tools-required: [Read, Write]
  tools-optional: []
  mcps-required: []
  environment: "Any agent environment; user describes the situation in text"
discovery:
  goal: "Classify the strategic move needed, determine its purpose (deterrence or compellence), select and design one or more credibility mechanisms, and produce a concrete action plan the user can execute to change the game before play begins"
  tasks:
    - "Classify the move type: unconditional commitment, deterrent threat, compellent threat, deterrent promise, or compellent promise"
    - "Determine whether the goal is deterrence (no deadline needed) or compellence (deadline required)"
    - "Distinguish genuine strategic moves from warnings and assurances to avoid wasted effort on informational-only signals"
    - "Run the renegotiation failure test: check whether both parties would benefit from renegotiating — if so, the proposed commitment is hollow"
    - "Calibrate brinkmanship risk: only invoke if a hard commitment is not credible; size the risk to the minimum needed to compel/deter"
    - "Select and design credibility mechanisms from the Eightfold Path: contracts, reputation, cut communication, burn bridges, leave to chance, small steps, teamwork, mandated agents"
    - "Check for anti-patterns: threat too large, renegotiable commitment, commitment that forecloses a winning position, brinkmanship when you would blink first"
    - "Produce a structured action plan: move type, purpose, credibility mechanism(s), concrete implementation steps, and monitoring criteria"
  audience: "Negotiators, executives, policymakers, managers, lawyers, and anyone designing incentive structures or making strategic pledges"
  when_to_use:
    - "User needs to lock in a position to gain first-mover advantage"
    - "User wants to deter an adversary from taking an unwanted action"
    - "User must compel a counterpart to act by a specific deadline"
    - "User suspects their threat or promise will be treated as a bluff"
    - "User is designing a contract, policy, or incentive mechanism and needs it to be renegotiation-proof"
    - "User is in a brinkmanship situation and needs to calibrate the risk level"
    - "User wants to counter or defuse another player's strategic move"
  quality:
    correctness: null
    depth: null
    actionability: null
    specificity: null
---

# Strategic Commitment Designer

## When to Use

Use this skill when you want to change the game — not just play it well — by taking a prior action that alters what others expect you to do or what you are capable of doing.

**Prerequisite:** Before designing a commitment, threat, or promise, you need to know the underlying game structure: who moves when, what the payoffs are, and what the current equilibrium looks like without any strategic move. If you have not already done this, run `backward-reasoning-game-solver` first. This skill takes up where backward induction leaves off: once you know what the game predicts, you use strategic moves to change the prediction.

The core principle: **a strategic move must change the game, not just describe your intentions.** Mere words are not strategic moves. The move works only if it alters the other party's rational expectations about your future behavior — and that requires changing either your payoffs or your available actions so that follow-through becomes your dominant choice, not just your stated preference.

---

## Step 1: Classify the Move Type

**WHY:** The taxonomy determines both what needs to be communicated and what credibility problem must be solved. Misclassifying the move type leads to designing the wrong mechanism.

Use the 3x2 taxonomy:

| | Deterrence (prevent action) | Compellence (induce action) |
|---|---|---|
| **Commitment** (unconditional first move) | Occupy a position before others can; create a fait accompli | Announce an irreversible action the other must respond to |
| **Threat** (conditional: punish non-compliance) | "If you do X, I will hurt you" — tripwire, no deadline needed | "Do Y by deadline T, or I will hurt you" — deadline is required |
| **Promise** (conditional: reward compliance) | "If you refrain from X, I will reward you" | "If you do Y by deadline T, I will reward you" |

**Warnings and assurances are not strategic moves.** A warning merely informs the other party what you would naturally do in your own interest anyway — it does not change your response rule. An assurance tells them what cooperative behavior would naturally elicit. Neither changes the game. If your "threat" is already in your interest to carry out, it is a warning. If your "promise" is already in your interest to keep, it is an assurance. Treat these as informational only and skip to credibility design only for the genuine strategic moves.

**Selecting the right move type:**
- If you want to stop something that has not yet started, and timing is open-ended → deterrent threat
- If you want to stop something already happening, or need action by a specific date → compellent threat (must attach a deadline or the opponent can use salami tactics to procrastinate)
- If punishment is too costly or reputation-destroying to be credible → consider a compellent promise instead (reward compliance rather than punish defiance)
- If you can act first without the other player being able to undo it → unconditional commitment; this is the cleanest move because it eliminates the response-rule credibility problem entirely

---

## Step 2: Run the Renegotiation Failure Test

**WHY:** A commitment that both parties would prefer to renegotiate is not a commitment at all — it collapses the moment temptation appears. This is the most common reason well-designed commitments fail in practice.

Apply the test before designing any credibility mechanism:

1. Identify the moment in the future when the commitment would be costly to honor (the "moment of temptation").
2. At that moment, ask: would both parties be better off agreeing to set the commitment aside?
3. If yes → the commitment is vulnerable to renegotiation. You cannot simply add a clause saying "no renegotiation"; you must change who holds the enforcement power so that at least one party at the table has an independent interest in enforcing the original agreement.

**The Nick Russo failure case:** A dieter offers $25,000 to anyone who catches him eating fattening food. At the moment of temptation, the dieter can offer the restaurant patron a free drink to look the other way. The patron prefers the drink to the slim chance of claiming the $25,000. Both parties benefit from renegotiation. The contract is hollow. **Fix:** enforcement must be held by a party who is not present at the moment of temptation and who has independent reasons to enforce (e.g., StickK.com's third-party charity pledge, or a supplier-producer penalty clause where the producer genuinely needs delivery, not just the fine).

---

## Step 3: Determine Purpose and Deadline Requirements

**WHY:** Deterrence and compellence have different structural requirements. Using deterrence logic for a compellence goal is a common error that allows the opponent to stall indefinitely.

**Deterrence:**
- Goal: prevent the other party from taking an action they otherwise would
- Timing: no deadline required — the threat is standing ("ever do X and I will respond")
- Better achieved with threats than promises (a threat is costless if it succeeds; a promise must be paid if it succeeds)
- Example: Cold War nuclear deterrence, beat-the-competition pricing clauses

**Compellence:**
- Goal: get the other party to take a positive action they would not otherwise take
- Timing: a specific deadline is required — without it, salami tactics defeat the move (the opponent defers action one small step at a time, and each step is too small to trigger the costly response)
- Better achieved with promises than threats for inducing genuine action (promises provide incentive not to procrastinate; threats require the threatener to invoke costs when partial progress has been made)
- Example: "Clean your room before 5 p.m. or lose dessert" vs. "Clean your room" with no deadline

**Threat vs. promise cost tradeoff:**
- A successful threat costs nothing to carry out (the action is never triggered)
- A successful promise must be paid (the reward must be delivered when the other party complies)
- A failed threat requires carrying out a costly action or accepting reputational damage from backing down
- Choose threat when deterrence is the goal and you can size the threat so the opponent complies; choose promise when compellence is the goal and you need to create a positive gradient toward compliance

---

## Step 4: Select Credibility Mechanisms (The Eightfold Path)

**WHY:** Words alone cannot make a strategic move credible. The other party knows that once they have moved, you have the incentive to renege on your threat or promise. You must change the game so that follow-through is in your interest at the moment of execution.

The eight mechanisms fall under three principles. Apply the principle that fits your situation first, then choose the specific mechanism.

### Principle 1 — Change Your Payoffs (Make Follow-Through Profitable)

**Mechanism 1: Write a contract.** Create a legal agreement with penalty clauses that make non-compliance more costly than compliance. Works best in commercial settings where the enforcer (court, arbitration panel) has independent reasons to enforce. Vulnerability: renegotiation by mutual consent. Must ensure enforcer has interests independent of both parties. Practical application: milestone-based payment contracts, performance bonds, penalty clauses for delay.

**Mechanism 2: Establish and use reputation.** Publicly commit your reputation to a position so that backing down destroys future credibility in other games. Works across repeated interactions and multiple audiences. The key is that the public declaration must be specific and visible enough that deviation is clearly observed. Vulnerability: reputation effects require future interactions — in a one-shot game, reputation has no value. Practical application: public speeches committing to a policy, publishing pricing policies in catalogs, establishing industry standing. Caution: a reputation built by public declaration can be demolished by a single visible breach (e.g., "Read my lips: no new taxes").

### Principle 2 — Limit Your Ability to Back Out (Make Retreat Impossible or Costly)

**Mechanism 3: Cut off communication.** Make yourself unavailable for renegotiation after the commitment is announced. If you cannot be reached, you cannot be persuaded to back down. Effective for creating irreversibility. Practical application: wills and trusts (the testator is unavailable after death), publishing a policy before entering negotiations, sending a certified letter. Vulnerability: you also cannot receive confirmation of compliance; you must designate a proxy to monitor.

**Mechanism 4: Burn bridges behind you.** Physically eliminate the retreat option so that advance is your only rational choice. Remove the alternative to following through. Practical application: Cortés burning his ships on arrival in Mexico (soldiers had no choice but to fight); Polaroid concentrating all resources in instant photography (committed to aggressive defense against entrants); announcing a product launch before the product is ready; signing a lease before leaving a job. The goal is to create a situation where your only rational action is the committed action.

**Mechanism 5: Leave the outcome beyond your control, or even to chance.** Delegate the trigger to an automatic mechanism or a probabilistic process that you genuinely cannot stop. Because you no longer control the response, the opponent cannot negotiate with you to avoid it. This is the mechanism underlying brinkmanship. Practical application: automatic penalty clauses triggered by observable metrics (credit ratings, delivery dates), doomsday devices, graduated penalty schedules. The power comes from genuine loss of control — simulated loss of control is transparent and ineffective.

### Principle 3 — Use Others to Help You Maintain the Commitment

**Mechanism 6: Move in small steps.** Break a large commitment into a sequence of small ones, each credible on its own. This solves the credibility problem when the full commitment is too large to be believable, and it reduces the damage if either party defects. Works against salami tactics when you are the compeller (each small step delivered honestly builds credibility for the next). Warning: the step-by-step structure creates an end-game problem — rational players anticipate defection on the last round and unravel the whole sequence. Fix: ensure there is no clearly defined last step (leave continuation open-ended). Practical application: milestone-based contracts, escalating relationship investments, progressive payment schedules.

**Mechanism 7: Develop credibility through teamwork.** Engage a group whose collective payoffs change when any individual defects. Social enforcement mechanisms — honor codes, peer accountability groups, organizational culture — make individual defection costly through shame, ostracism, or retaliation by the group. Practical application: Alcoholics Anonymous (breaking a commitment destroys standing in the group), honor codes at universities (failure to report cheating is itself a violation), labor unions (the leader is accountable to a constituency that will remove him if he backs down). The team structure changes the individual's payoffs without requiring an external enforcer.

**Mechanism 8: Employ mandated negotiating agents.** Delegate authority to an agent whose own incentives prevent flexibility. The agent genuinely cannot concede because doing so would cost them their position, reputation, or legal standing. Two forms: (a) human agents with restrictive mandates (union leaders whose members must ratify any contract; lawyers without settlement authority; real estate agents with published listing prices); (b) mechanical agents (vending machines, automated pricing systems, fixed-rule bureaucracies). The agent's constraint is credible because it is externally visible and costly to override. Vulnerability: the opponent can attempt to go directly to the principal — counter by making the principal unavailable or by ensuring the agent's constraint is itself renegotiation-proof.

---

## Step 5: Calibrate Brinkmanship (If Applicable)

**WHY:** Brinkmanship is a last resort when a hard commitment is impossible and the threat too large to be credible as written. It deliberately creates shared risk to compel compliance. Used incorrectly, it destroys both parties.

Brinkmanship is the **controlled loss of control**: the threatener controls the size of the risk but not the outcome. The risk is genuinely real — this is not a bluff. If the opponent calls the bluff, there is a real chance the catastrophic outcome occurs.

**Apply brinkmanship only when:**
- A hard, credible threat is not available (the action is too costly to be credible as a certain response)
- You need a graduated response mechanism (you do not know the minimum threat size that will work)
- You assess that your opponent's tolerance for risk is lower than yours (if you would blink first, do not start)

**Calibration protocol:**
1. Start with the smallest risk increment that has a reasonable chance of inducing compliance
2. Escalate gradually, observing the opponent's reactions
3. Maintain genuine loss of control — automated or observable mechanisms are more credible than personal discretion
4. Have a de-escalation path ready (face-saving compromise for both parties)
5. Know your own tolerance threshold before starting — if the risk will exceed your own threshold before exceeding the opponent's, do not begin

**Cuban Missile Crisis benchmark:** Kennedy estimated 1/3 to 1/2 odds of nuclear war. This was not a bluff — the risk was real. The strategy worked because Khrushchev's risk tolerance was lower. Before using brinkmanship, honestly estimate both tolerance levels.

---

## Step 6: Check for Anti-Patterns

**WHY:** Strategic moves that are poorly designed can backfire, destroying credibility, locking you into losing positions, or creating outcomes worse than no move at all.

**Anti-pattern 1 — Threat too large.** A threat exceeding what is necessary and proportionate generates terror and is socially unacceptable, making it incredible even if technically feasible. Size threats at the minimum level needed to achieve deterrence or compellence. The escalation cost if the threat fails is too high to ignore.

**Anti-pattern 2 — Renegotiable commitment.** Any commitment both parties would prefer to renegotiate will be renegotiated. Do not rely on legal language alone; ensure enforcement is held by a party with independent interests. Check: if the commitment were tested, would both you and your counterpart benefit from quietly setting it aside?

**Anti-pattern 3 — Commitment that forecloses your winning position.** Before committing, run backward induction on the new game the commitment creates. Some commitments eliminate your own optimal future moves. De Lesseps committed to a sea-level canal at Panama without checking the engineering; the commitment destroyed his ability to adapt when the geology made sea-level construction fatal. First check: does this commitment leave me with a rational path to my goal?

**Anti-pattern 4 — Commitment without a credibility mechanism.** An unconditional announcement is not a commitment; it is just a statement. If you announce a commitment without changing payoffs, eliminating retreat options, or engaging others to enforce it, the announcement will be treated as cheap talk.

**Anti-pattern 5 — Starting brinkmanship when you would blink first.** If your honest assessment is that the shared risk will exceed your own tolerance before the opponent's, brinkmanship is irrational. Your opponent will read your tolerance correctly and call your escalation.

---

## Step 7: Compose the Action Plan

**WHY:** Strategic moves must be implemented as specific prior actions, not future intentions. The plan must be executable before the game begins.

Produce a structured plan covering:

1. **Move type and purpose:** [commitment / deterrent threat / compellent threat / deterrent promise / compellent promise] for [deterrence / compellence]
2. **Deadline (if compellent):** Specific, observable deadline with graduated consequences for partial progress
3. **Credibility mechanism(s):** Which of the eight mechanisms applies, and the concrete prior action that implements it
4. **Renegotiation firewall:** Who holds enforcement authority, and why they have independent interests in enforcing
5. **Opponent's counter-options:** How the opponent might try to undermine your credibility (p. 212-215), and pre-emptive countermeasures
6. **Monitoring criteria:** Observable indicators that the move is working or has been defied
7. **Contingency:** What you will do if the move is tested (having thought this through in advance is itself a credibility signal)

---

## Discovery

This skill surfaces when the user is trying to change what others will do by taking prior action — not just responding optimally to the current game. Key phrases: "make my threat credible," "how do I commit to this," "they won't believe me," "I need them to take action by X date," "how do I deter," "designing an incentive structure," "contract that can't be renegotiated," "burned my bridges," "how do I back this up."

For detailed reference material, see:
- `references/move-taxonomy-reference.md` — full 3x2 taxonomy with examples
- `references/eightfold-path-reference.md` — detailed mechanism descriptions and case studies
- `references/renegotiation-failure-test.md` — test protocol with worked examples
- `references/brinkmanship-calibration.md` — risk calibration guide and anti-patterns

## License

This skill is licensed under [CC-BY-SA-4.0](https://creativecommons.org/licenses/by-sa/4.0/).
Source: [BookForge](https://github.com/bookforge-ai/bookforge-skills) — The Art of Strategy by Avinash K. Dixit, Barry J. Nalebuff.

## Related BookForge Skills

Install related skills from ClawhHub:
- `clawhub install bookforge-backward-reasoning-game-solver`

Or install the full book set from GitHub: [bookforge-skills](https://github.com/bookforge-ai/bookforge-skills)

FILE:references/brinkmanship-calibration.md
# Brinkmanship Calibration Guide

Source: The Art of Strategy, Ch. 6–7 (Dixit & Nalebuff)

## What Brinkmanship Is (and Is Not)

**Brinkmanship is not:** Threatening to push the adversary off the brink (cold-bloodedly executing a catastrophic response). That threat is not credible — the adversary would take you down with them, and they know it.

**Brinkmanship is:** The deliberate creation of a shared risk — a risk that neither party fully controls. The threatener sets the size of the risk but cannot guarantee the outcome. The goal is to make the risk sufficiently intolerable to the opponent that they eliminate it by complying.

**Schelling's definition:** Brinkmanship is a "controlled loss of control." The shooter controls how many bullets are in the chamber (the size of the risk) but not whether the firing chamber contains the bullet (the outcome).

**Why it works:** Cold-blooded execution of a catastrophic punishment is often not credible because it would hurt both sides. But creating a risk of that punishment — a genuine, uncontrolled probability — can be credible because the escalation has already been set in motion. The adversary cannot negotiate the risk away because the risk has partially moved beyond anyone's control.

## The Brinkmanship Sequence

**Phase 1 — Establish the threshold.**
Define the action you want to prevent or compel. Establish the link between that action and the risk of a mutual catastrophe. The link must be genuine (not simulated) or the adversary will recognize the bluff.

**Phase 2 — Start at the minimum risk level.**
Begin with the smallest risk increment that has a reasonable chance of inducing compliance. Do not start large — you do not know in advance the minimum risk that will work, and starting large wastes your escalation space and raises the danger unnecessarily.

**Phase 3 — Escalate gradually.**
Increase the probability of the bad outcome step by step, observing the opponent's response. Each escalation signals that you are willing to go further and that the risk is real.

**Phase 4 — Maintain genuine loss of control.**
The escalation must involve genuinely giving up control over the outcome, not just threatening to. Mechanisms that create real risk (naval blockades where individual officers make decisions in the field; economic sanctions that trigger automatic financial market responses; military postures where miscommunication is possible) are more credible than centrally controlled, perfectly reversible moves.

**Phase 5 — Have a de-escalation path ready.**
Brinkmanship requires a face-saving exit for both parties. If the opponent backs down, they need a way to do so without total humiliation — otherwise they may prefer the catastrophe to the capitulation. The Cuban Missile Crisis resolution involved a face-saving element: the eventual withdrawal of U.S. missiles from Turkey, allowing Khrushchev to claim a quid pro quo.

## The Cuban Missile Crisis — Calibration Reference

**Context (1962):** The Soviet Union installed nuclear missiles in Cuba, 90 miles from the American mainland. Kennedy announced a naval quarantine on October 22.

**The risk level:** Kennedy himself estimated the probability of nuclear war at "between one out of three and even" — between 33% and 50%.

**Why this was brinkmanship, not a hard commitment:**
- Kennedy could not credibly promise to launch nuclear war if a single Soviet ship crossed the quarantine line (the cost to the U.S. was too catastrophic)
- Instead, the quarantine created a risk of escalation: any confrontation could have spiraled through "fog of war" dynamics into nuclear exchange
- Kennedy did not fully control the naval operations — he attempted to move the blockade from 800 to 500 miles but evidence suggests it was never moved; the first ship boarded was a Lebanese freighter under Soviet charter

**The outcome:** Khrushchev "looked over the nuclear brink, did not like what he saw, and pulled back." He accepted a face-saving compromise: Soviet missiles withdrawn from Cuba in exchange for a U.S. pledge not to invade Cuba and (eventually) U.S. missile withdrawal from Turkey.

**The calibration lesson:** Kennedy's risk tolerance was high enough to accept a 1/3 to 1/2 probability of nuclear war. Khrushchev's was lower. The brinkmanship worked because Kennedy correctly assessed that the opponent would blink first. If Khrushchev had had a higher tolerance for nuclear risk, the survivors would have condemned Kennedy rather than praised him.

## Pre-Conditions Checklist

Before invoking brinkmanship, verify all of the following:

- [ ] A hard commitment is not available (the action is too costly to be credible as a certain response)
- [ ] You can create a genuine (not simulated) escalating risk
- [ ] Your honest estimate of your risk tolerance is higher than your honest estimate of the opponent's risk tolerance
- [ ] A face-saving de-escalation path exists for the opponent
- [ ] The potential catastrophe, even if it occurs, leaves you in a survivable position (do not use brinkmanship when the catastrophic outcome is existential for you)
- [ ] You can escalate gradually — the threat is not binary or indivisible

**If any box is unchecked, do not use brinkmanship.**

## Calibrating Risk Size

**Too small:** The opponent ignores the risk; no behavior change occurs. The result is continued confrontation at no cost to the opponent.

**Too large:** You reach your own tolerance threshold before the opponent reaches theirs; you must back down or face the catastrophe yourself. This destroys your credibility and may trigger the catastrophe.

**Just right:** The risk level causes the opponent to perceive that the expected cost of continued defiance exceeds the cost of compliance, while the risk remains within your own tolerance level.

**Practical guidance on sizing:**
- Start at 10-20% of the eventual risk you are willing to accept
- Escalate in steps of comparable size
- Watch for signals of the opponent's tolerance: delays in response, partial concessions, back-channel communications
- Do not escalate faster than the opponent can process and respond — allow time for the signal to register

## Tiananmen Square Counter-Case

The 1989 Tiananmen Square confrontation illustrates brinkmanship failure.

**The setup:** Chinese students occupied Tiananmen Square; hard-line government leaders wanted to clear it. Both sides were in a game of brinkmanship — neither backing down, each hoping the other would concede first.

**The failure:** When neither side blinked, the catastrophic outcome (military crackdown) occurred. Both sides had reached their tolerance thresholds simultaneously, and the government's hard-liners chose the catastrophic response.

**The lesson for calibration:** Brinkmanship fails when both parties have similar risk tolerance and neither blinks before the catastrophic outcome occurs. When you cannot assess which party has the higher risk tolerance, the brinkmanship strategy is gambling, not strategy. In the aftermath, other communist governments that learned from Tiananmen (East Germany, Czechoslovakia) chose different strategies — conceding to democracy protests rather than escalating to mutual catastrophe.

## The "Blink First" Self-Assessment

Before starting brinkmanship, conduct this honest self-assessment:

1. What is the catastrophic outcome you are both risking? Describe it concretely.
2. At what probability of that outcome occurring would you back down? (Your tolerance threshold)
3. At what probability do you estimate your opponent would back down? (Their tolerance threshold)
4. How quickly can you escalate to the opponent's threshold without exceeding your own?
5. If your estimate in Step 2 is lower than your estimate in Step 3, do not proceed.

This assessment requires genuine honesty — the temptation is to overestimate your own tolerance and underestimate the opponent's. Kennedy and his advisors conducted exactly this assessment before the Cuban quarantine decision.

## Brinkmanship vs. Salami Tactics (Defender's Perspective)

When an opponent is using brinkmanship against you:

**Salami tactics** — defying the opponent's wishes in increments so small that each step is below the threshold that would trigger the costly response. Each slice is too thin to eat on its own; the whole salami is consumed before the opponent notices.

This is the child's strategy: told not to go in the water, they sit on the bank and submerge their feet; then wade to the knee; then argue that since they go back and forth it all averages out. Eventually they are swimming out of sight.

**Counter to salami tactics:** Move to a commitment or automatic mechanism that triggers on accumulated behavior, not on any single step. Graduated penalty schedules (a few grade points per minute of overtime in an exam) are immune to salami tactics because each small step is independently penalized. All-or-nothing thresholds (the exam is rejected if it is even one minute late) are vulnerable because no single small step seems worth the catastrophic response.

FILE:references/eightfold-path-reference.md
# The Eightfold Path to Credibility

Source: The Art of Strategy, Ch. 7 (Dixit & Nalebuff)

## Overview

Commitments, threats, and promises are not credible by default. After the other party has moved, you have the incentive to renege — either not to carry out a costly threat or not to pay a costly reward. The eight mechanisms below are organized under three principles that transform your announced move into a binding one.

The principles:
1. **Change the payoffs** — make follow-through profitable by raising the cost of reneging
2. **Limit your ability to back out** — remove the retreat option physically or informationally
3. **Use others** — engage parties or mechanisms whose interests make them enforce your commitment independently

---

## Principle 1: Change the Payoffs

### Mechanism 1 — Write a Contract

**How it works:** Agree to pay a penalty if you fail to follow through, or accept a reward-and-penalty structure that makes compliance your dominant choice. The contract shifts your payoffs so that breaking the commitment is more costly than keeping it.

**Critical requirement — the renegotiation-proof condition:** The party that enforces the penalty must have independent interests in enforcement. If both parties to the commitment prefer to renegotiate, they will — regardless of what the contract says. A third-party enforcer (court, arbitration panel, charity, trusted institution) with independent incentives is essential.

**When it works best:** Commercial dealings where the enforcer (typically a court) receives no benefit from renegotiation, and where the enforcing party would rather receive the contracted performance than the penalty sum.

**When it fails:** Personal behavioral commitments where enforcers are present at the moment of temptation and can be bought off (the Nick Russo $25,000 diet contract failure — the restaurant patron prefers a free drink to waiting to claim the bounty).

**Variants:**
- Milestone-based payment (contractor gets paid for progress, incentivizing delivery)
- Self-incriminating letters held by a rehabilitation center (Denver cocaine addiction treatment model)
- StickK.com / Commitment Store (third party holds the stake, gains reputation from enforcement, has no financial motive to renegotiate)
- ABC Primetime bikini photo contract (effective when the program producer's reputation was staked on enforcement)

**Contractual escalation — building toward irrevocability:** Some contracts are designed so that the person holding the contract would lose their job if they allowed renegotiation, thereby creating a chain of enforcement that extends well beyond the original commitment.

### Mechanism 2 — Establish and Use Reputation

**How it works:** Publicly commit your reputation to a position so that backing down causes visible, costly damage to your standing in future games. Because you play multiple games with the same or overlapping audiences, your credibility in future interactions depends on your follow-through in current ones.

**What makes reputation work:**
- The commitment is specific, visible, and observable
- Deviation is clearly detectable
- You interact with the same or overlapping opponents in the future
- The future value of reputation exceeds the short-term gain from reneging

**Classic example:** Kennedy's inaugural address and Berlin crisis speeches created a public reputation for nuclear deterrence. The cost of backing down in Berlin or Cuba exceeded the cost of confrontation, because it would have destroyed U.S. credibility with all NATO allies, all Soviet proxies, and all future adversaries simultaneously.

**The public declaration device:** Making a commitment in public puts your reputation on the line in a visible way. The Mafia example: the only thing that makes a hit man trustworthy is a track record of committing acts of violence. Reputation is built by actually following through — it cannot be faked through signals that are cheap to produce.

**Vulnerability:** A single visible breach can destroy a reputation that took years to build. George H.W. Bush's "Read my lips: no new taxes" pledge — the breach of that public commitment was cited as a major factor in his 1992 defeat. Public declarations should be made only when you are highly confident you can maintain them, or when the benefits of the reputation outweigh the risks of breach.

---

## Principle 2: Limit Your Ability to Back Out

### Mechanism 3 — Cut Off Communication

**How it works:** Make yourself unavailable for renegotiation after announcing the commitment. If the opponent cannot reach you, they cannot negotiate the commitment away.

**Core effect:** Creates irreversibility without requiring physical action — the mere act of becoming unavailable removes the renegotiation option.

**Examples:**
- Wills and testaments: once the testator has died, the commitment is irreversible (it took an act of British Parliament to override Cecil Rhodes's will on female Rhodes Scholars)
- Mailing a certified letter: signing for it proves the recipient has read it — creating a binding informational fact
- General Ripper in Dr. Strangelove: sealed the base, cut communications, destroyed radios, and killed himself to prevent any recall of the attack

**Limitations:**
- You cannot monitor compliance if you are incommunicado — you must designate an agent with authority to observe and act
- Ripper's attempt failed because the communication cut was imperfect (a radio was found playing music; a payphone was discovered; Ripper's obsessive doodling revealed the recall code to Officer Mandrake)
- Theory vs. practice: even the most thorough communication cut leaves "unknown unknowns" that can unravel the commitment

### Mechanism 4 — Burn Bridges Behind You

**How it works:** Physically eliminate the alternative to following through. When there is literally no option to retreat, advance is rational regardless of its cost.

**The logic:** Burning bridges does not make you stronger — it makes you less free. But in a strategic interaction, the freedom you are eliminating is the freedom to be the "chicken." By removing the retreat option, you transfer the entire burden of avoiding disaster to the opponent, who must back down or face mutual catastrophe.

**Historical examples:**
- Hernán Cortés, Mexico 1519: burned or disabled all but one of his ships. His soldiers, vastly outnumbered, had no choice but to fight to victory or die. "Had [Cortés] failed, it might well seem an act of madness.... Yet it was the fruit of deliberate calculation. There was no alternative in his mind but to succeed or perish."
- William the Conqueror, England 1066: burned his own ships after landing, committing his forces to fight rather than flee
- Russian Captain Ramius in The Hunt for Red October: mailed a letter to Admiral Padorin announcing his intention to defect before departing. The Soviets would now try to sink the submarine. There was no turning back.
- Polaroid Corporation: refused to diversify out of instant photography for decades, committing credibly to aggressive defense of that market. When Kodak entered in 1976, Polaroid sued and won a $909.4 million judgment.

**Inverse: building bridges as commitment:** The East German government dismantling parts of the Berlin Wall in December 1989 committed itself to reform credibly — the opening made reversal visible and costly. The promise of reform was credible precisely because the government had now made exodus possible, meaning failure to reform would produce mass emigration.

### Mechanism 5 — Leave the Outcome Beyond Your Control, or Even to Chance

**How it works:** Delegate the trigger to an automatic mechanism or probabilistic process that you genuinely cannot override. Because follow-through no longer depends on your willingness to act, the opponent cannot negotiate with you to prevent it.

**The power of genuine loss of control:** When the response is automated and irreversible, the question "will you really do it?" becomes unanswerable — the mechanism will execute regardless of anyone's subsequent preferences. This is the source of the Doomsday Machine's deterrent power in Dr. Strangelove: "It is not anything a sane man would do. The Doomsday Machine is designed to trigger itself automatically.... It is designed to explode if any attempt is ever made to untrigger it."

**Two forms:**
1. **Complete automatic response:** The response is certain if the trigger condition is met (Doomsday Machine; automatic penalty clauses; bond covenants triggered by rating downgrades). Maximally credible; also maximally dangerous if accidentally triggered.
2. **Probabilistic response (brinkmanship):** The response is not certain but has a real probability. The threatener controls the size of the probability, not the outcome. This allows graduated escalation and avoids the all-or-nothing problem of large indivisible threats. See the brinkmanship calibration reference for full treatment.

**Application examples:**
- Soviet "automatic rocket response" in Berlin: Khrushchev threatened that Soviet rockets would fly automatically in the event of armed conflict — removing his own discretion to de-escalate
- Automated repricing systems: prices change without human intervention when competitor prices are observed, making the threat to match prices mechanically credible
- Covenant triggers in bond agreements: specific financial metrics (debt ratios, credit ratings) automatically trigger acceleration or restructuring, removing management's discretion

---

## Principle 3: Use Others

### Mechanism 6 — Move in Small Steps

**How it works:** Break a large commitment into a sequence of smaller ones, each credible on its own. The small-step structure solves two problems simultaneously: the individual step is too small to be worth defecting on, and each defection terminates the entire profitable relationship.

**When it applies:** When the full commitment would require trusting the other party with more than either party can credibly commit to at once. The step-by-step structure caps the maximum loss from defection to one step's worth of value.

**The cocaine transaction example:** Engaging in 1,000 sequential $1,000 transactions rather than one $1 million transaction. Defecting for $1,000 is not worth ending a relationship worth $999,000 in future transactions. Defecting for $1 million is worth ending the relationship entirely.

**The end-game problem:** Rational parties using backward induction anticipate defection on the last round, defect on the penultimate round to preempt it, and the whole cooperation unravels. **Fix:** ensure there is no clearly defined last step. As long as continued interaction remains possible, defection is not worth the cost. When a store announces a "going out of business sale with massive price reductions," be especially skeptical about quality — the end-game has arrived and defection is now rational.

**Connection to brinkmanship:** Moving in small steps also reduces the scale of each commitment, making each step credible in the same way brinkmanship escalates risk in small increments. The difference: small steps build cooperative credibility upward; brinkmanship escalates threatening credibility upward.

### Mechanism 7 — Develop Credibility Through Teamwork

**How it works:** Create or join a group structure where individual defection triggers collective enforcement. The group changes your payoffs by adding social costs (shame, ostracism, loss of group membership) to any commitment breach.

**Why teams work:** Individuals may be too weak — or too easily tempted — to maintain commitments on their own. The team establishes a social institution where pride and self-respect are lost when commitments are broken. Sometimes the team goes further, applying direct coercion.

**Examples:**
- Alcoholics Anonymous: the AA approach creates a social institution where breaking sobriety is witnessed by the group, destroying standing and self-respect. The group changes the payoffs of relapse.
- Diet and exercise groups: public accountability creates shame costs for non-compliance
- Roman army decimation protocol: soldiers who saw a comrade falling behind in an attack were ordered to kill the deserter immediately — and failing to do so was also a capital offense. The credible mutual enforcement made advance individually rational even for soldiers who would rather retreat.
- University honor codes: failure to report observed cheating is itself a violation, making silence as costly as cheating. This creates a self-enforcing network of accountability.
- Dean Karlan's weight contract with his friend (Yale economist): each party collected the other's penalty when the other failed. Karlan actually collected $15,000 from his friend. Knowing the friend would collect if Karlan failed was the commitment's credibility — mutual enforcement, not external enforcement.

### Mechanism 8 — Employ Mandated Negotiating Agents

**How it works:** Delegate negotiation authority to an agent whose own incentives genuinely prevent them from conceding beyond your stated position. The agent's constraints are externally visible and costly for the agent to override — making your negotiating position credible without requiring your personal willpower.

**Why it works:** The agent is accountable to a constituency (union members, shareholders, law, institutional rules) that will penalize the agent for compromising. The opponent cannot negotiate a better deal by applying social pressure to you personally, because you are not the decision-maker.

**Two types of mandated agents:**
1. **Human agents with restrictive mandates:** Union leaders who must have contracts ratified by members (they cannot accept less without losing their position); attorneys without settlement authority; sports agents (useful when you have personal bonds with the other party that would lead you to concede too much); real estate agents with published listing prices
2. **Mechanical agents (machines and rules):** Vending machines do not negotiate; store clerks following fixed policies can truthfully say the decision is "above their grade." The mechanical nature of the agent creates absolute non-negotiability.

**When mandated agents are especially valuable:** When you share social bonds with the other party that make you reluctant to hold firm. An impersonal agent avoids the trap of conceding too much for the sake of the relationship. Professional athletes use agents partly for this reason; authors use agents in dealings with editors and publishers.

**Vulnerability:** The opponent can try to bypass the agent and go directly to the principal. Counter: make the principal unavailable (cut communication), or ensure the agent's constraint is itself renegotiation-proof (the mandate is publicly visible and the agent's career depends on not breaching it).

**Important qualification:** If the agent voluntarily chose an inflexible position (vs. being externally mandated), observers may treat it differently. A labor leader who voluntarily committed his prestige to a position invites a different interpretation than one externally constrained by a ratification requirement.

FILE:references/move-taxonomy-reference.md
# Move Taxonomy Reference

Source: The Art of Strategy, Ch. 6 (Dixit & Nalebuff)

## The Three-by-Two Taxonomy

Strategic moves are classified on two axes: the **structure** of the move (unconditional vs. conditional) and the **purpose** of the move (deterrence vs. compellence).

### Axis 1: Structure

**Commitment — Unconditional First Move**
You act before the other player moves. There is no "if" condition. The other player is now the follower and must respond to a fait accompli. Because you have no further moves in the game at the time the other player responds, the response-rule credibility problem does not arise. The commitment must, however, be irreversible — if the other player can simply wait you out or undo the commitment, the move has no force.

Example: The night self sets the alarm clock on the wardrobe across the room. By morning, the morning self must get out of bed to silence it. The night self has no further moves; the irreversibility is structural.

**Threat — Conditional Move that Punishes**
You announce in advance: "If you do X (which I do not want), I will respond with action Y (which hurts you, and costs me something too)." The response rule must be in place before the other player acts. You must also have the second move — some way of observing the other player's action and responding after it.

A threat requires:
1. A response rule that is genuinely not in your natural interest to follow through (otherwise it is a warning)
2. Some mechanism to make that response credible

**Promise — Conditional Move that Rewards**
You announce in advance: "If you do X (which I want), I will respond with action Y (which rewards you, and costs me something too)." A promise also requires being in place before the other player acts, and also has the incentive problem: once the other player has complied, you no longer need to pay the reward.

A promise requires:
1. A response rule that rewards the other player in a way that is not in your natural interest to deliver once compliance is secured
2. Some mechanism to make that reward delivery credible

### Axis 2: Purpose

**Deterrence — Preventing an Action**
Goal: stop the other player from doing something they would naturally do.
- No deadline required (the deterrent is standing)
- Better achieved with a threat than a promise (a deterrent threat is free if it succeeds)
- Creates a "tripwire" — the other player decides whether to trigger it

Deterrence formula: "If you ever do X, I will respond with Y (costly to you, costly to me)."
U.S. Cold War nuclear deterrence: "If the Soviet Union attacks any NATO country, we will respond with nuclear weapons."

**Compellence — Inducing an Action**
Goal: get the other player to do something they would not naturally do.
- Deadline is required; without it, salami tactics defeat the move
- Better achieved with a promise than a threat (promise provides an incentive gradient; threat requires costly follow-through against partial compliance)
- The opponent must know exactly what action is required and by when

Compellence formula: "Do X by time T, or I will respond with Y."
Parent example: "Clean your room before 5 p.m. or no dessert tonight."

## The Full Table

| | **Deterrence** | **Compellence** |
|---|---|---|
| **Threat** | "Don't do what I don't want you to do... then I will respond with an action that hurts you (and also hurts me)" | "Do what I want you to do... or I will respond with an action that hurts you (and also hurts me)" |
| **Promise** | "Don't do what I don't want you to do... then I will respond with an action that rewards you (at some cost to me)" | "Do what I want you to do... then I will respond with an action that rewards you (at some cost to me)" |

## Warnings and Assurances (Not Strategic Moves)

A **warning** is a statement that your "threat" is actually something you would carry out in your own interest regardless. It is informational, not strategic — it cannot change the game because it does not change your response rule.

Example: If B.B. Lean would naturally respond to Rainbow's End's price cut by cutting its own price by 40 cents for each dollar of Rainbow's cut (because this maximizes B.B. Lean's profit given market conditions), then announcing this is a warning — it merely informs Rainbow's End of B.B. Lean's natural behavior.

An **assurance** is a statement that your "promise" is actually in your interest to fulfill anyway. Again, informational only.

When to use warnings/assurances: They serve a communication function — ensuring the other party is not misinformed about your intentions. But they do not require credibility mechanisms and cannot change equilibrium behavior.

## Threat vs. Promise: Choosing Between Them

**Cost structure:**
- A successful threat costs nothing to carry out (deterrence works; the trigger is never pulled)
- A failed threat requires carrying out the costly action or accepting the reputational cost of backing down
- A successful promise must always be paid
- A failed promise (promise made but not fulfilled) destroys reputation for future promises

**When to prefer a threat:**
- Deterrence is the goal (no action needed from the other party, just restraint)
- The threatened action is credible without a mechanism (it lies in your natural interest, making it a warning — but if it needs slight reinforcement, a threat is cheaper than a promise)
- You can size the threat just large enough to deter (keeping follow-through costs manageable if the threat fails)

**When to prefer a promise:**
- Compellence is the goal
- The other party's compliance genuinely benefits you and the benefit covers the reward cost
- Punishment would be disproportionate or legally/socially unacceptable

**The Stalin error:** Using threats (sticks) instead of promises (carrots) fails when the monitoring system for compliance is arbitrary and corrupt. If the other party cannot predict whether punishment will be triggered, both effort and shirking receive the same expected punishment — removing all incentive to comply.

## Other Players' Strategic Moves

You can also benefit from or be harmed by other players' strategic moves:

**Allowing the other party to make an unconditional commitment first** can be beneficial when the game has a second-mover advantage. Sun Tzu's advice to leave an enemy an escape route aims to prevent the enemy from making a credible commitment to fight to the death.

**Never allow threats directed at you** — they can only restrict your options. But allowing promises directed at you can benefit both parties (mutual promises in a prisoner's dilemma can resolve the coordination problem).

**Undermining opponent credibility:** In zero-sum or conflict situations, undermine the opponent's commitment by: exposing renegotiation opportunities (contracts), destroying reputation (keep agreements private), cutting their communication channels, leaving escape routes (burning bridges counter), using salami tactics against their compellent threats (moving in steps defuses them), or going directly to the principal when they use a mandated agent.

FILE:references/renegotiation-failure-test.md
# Renegotiation Failure Test

Source: The Art of Strategy, Ch. 7 (Dixit & Nalebuff)

## The Core Problem

Any commitment that both parties would prefer to renegotiate will be renegotiated.

No matter how well-designed a contract or commitment appears, if both the committer and the enforcer would benefit from quietly setting it aside at the moment of temptation, they will. The third-party enforcement apparatus (courts, contracts, reputation mechanisms) only comes into play if at least one party brings a dispute. If both parties prefer to avoid the dispute, the commitment evaporates.

## The Test Protocol

Apply this test to any proposed commitment before investing in credibility mechanisms.

**Step 1 — Identify the moment of temptation.**
When specifically will the commitment become costly to honor? Describe the concrete scenario: what temptation appears, when it appears, and who is present.

**Step 2 — Map all parties present at that moment.**
Who is at the table when the commitment is tested? The committer, the enforcer, any third parties?

**Step 3 — Check mutual benefit from renegotiation.**
For each party present: would they be better off agreeing to release the commitment in exchange for some immediate side deal?
- If yes for all parties → renegotiation is likely; commitment will fail unless redesigned
- If no for at least one party → that party has an independent interest in enforcing; proceed to assess their power to enforce

**Step 4 — Assess enforcer independence.**
If one party would not benefit from renegotiation, do they have the power to prevent it? (Legal standing, physical control of the assets, reputational stakes that require enforcement.)

**Step 5 — Redesign if needed.**
If all present parties benefit from renegotiation, restructure the commitment so that enforcement authority rests with a party who:
(a) is not present at the moment of temptation, or
(b) has independent interests that are not served by renegotiation

## Case Analysis: Nick Russo's $25,000 Diet Contract

**The setup:** Nick Russo, frustrated with weight loss programs, offered $25,000 (to charity) to anyone who spotted him eating fattening food in a restaurant. He posted "wanted" pictures of himself in local restaurants.

**The failure:**
- Moment of temptation: Russo sees a dessert he wants
- Parties present: Russo and the restaurant patron
- Does Russo benefit from renegotiation? Yes — he gets the dessert
- Does the patron benefit from renegotiation? Yes — Russo offers them a free round of drinks, which is worth more to the patron than the slim chance they would actually receive the $25,000 bounty (they would need to catch Russo, claim the bounty, and receive payment — all uncertain)
- Both parties prefer renegotiation → the contract is hollow

**The fix:** Replace the patron-as-enforcer with an enforcer who has independent interests:
- Russo's family (who also want him thin and would not accept a round of drinks)
- A third-party institution (StickK.com, a charity) whose reputation depends on strict enforcement
- A formal court contract with a plaintiff who would rather receive the promised performance than the penalty (as in supplier-producer delivery contracts)

## Case Analysis: Supplier-Producer Penalty Clause

**The setup:** A producer demands a penalty from a supplier who fails to deliver on time. The contract specifies a large penalty for delay.

**Does this pass the renegotiation test?**
- Moment of temptation: the supplier is late and tries to renegotiate the penalty
- The supplier argues: "The penalty is so large it will always be honored, so you will never actually collect it — just release me from it."
- Does the producer benefit from renegotiation? **No.** The producer wants the delivery, not the penalty. Renegotiating the penalty does not get the producer its supply. The producer would rather enforce the contract to get delivery.
- Result: the contract is renegotiation-resistant because the enforcer's interest is in the *action promised*, not in the penalty.

## Case Analysis: ABC Primetime Bikini Photo Contract (First Round)

**The setup:** Overweight participants agreed to be photographed in bikinis. Anyone who failed to lose 15 pounds in two months would have the photos displayed on national television. One participant failed narrowly (lost 13 pounds, dropped two dress sizes).

**Round 1 outcome:** ABC forgave the participant. The producers exercised discretion and chose not to air the photos.

**Why this failed the renegotiation test:**
- At the moment of decision, the participant threatened (implicitly) a lawsuit if the photos aired
- ABC's interest was not in punishment per se but in program credibility
- With the lawsuit threat, both parties had reasons to renegotiate → ABC backed down

**The consequence:** Future participants could not regard the contract as credible. ABC had to redesign the mechanism entirely for round 2 (using the baseball team's jumbotron instead of national television — different enforcer, different stakes).

## Case Analysis: Dean Karlan's Mutual Commitment Contract

**The setup:** Dean Karlan (Yale economist) and a friend each agreed that if either went above 175 pounds, the overweight one would owe the other $1,000 per pound.

**Why this passes the renegotiation test:**
- Moment of temptation: Karlan's friend reached 190 pounds
- Karlan could have offered to renegotiate (let the friend off the hook)
- But Karlan's incentive was to actually collect — doing so demonstrated that the threat was real, which protected Karlan's own future compliance
- Karlan collected $15,000 from his friend, even though he personally did not enjoy doing so
- Both parties knew the other would collect → neither could renegotiate without destroying the bilateral enforcement mechanism

**The key:** Mutual stakes, bilateral enforcement. Neither party can renegotiate without destroying the mechanism that protects them from their own future temptation.

## Diagnostic Questions

Use these questions to quickly test any commitment:

1. At the moment when follow-through becomes costly, who is in the room?
2. Could either party make the other better off by quietly renegotiating?
3. Does the enforcer have an independent interest in enforcement (beyond the penalty itself)?
4. Can the enforcer be persuaded or compensated to look the other way?
5. If the commitment is tested today, who would bring the dispute to the enforcement mechanism — and why?

If you cannot answer question 5 with a clear, motivated party, redesign the enforcement structure before proceeding.

---
name: prisoners-dilemma-resolver
description: |
  Diagnose whether a multi-player conflict is a prisoners' dilemma and design a cooperation mechanism to resolve it. Use when parties are locked in a mutually destructive pattern even though all would benefit from cooperation — price wars, overfishing, arms races, advertising spirals, commons depletion, collective action failures. Distinguishes prisoners' dilemmas (dominant strategy to defect) from coordination problems (no incentive to deviate once aligned) and tailors the remedy accordingly. Produces a structured cooperation design plan: diagnosis, payoff assessment, discount-rate threshold calculation, mechanism selection from a resolution menu (self-enforcement through repeated play, tit-for-two-tats, mutual promises with escrow, linkage, reputation systems, third-party enforcement, Ostrom commons governance), and implementation checklist. Use when someone says 'everyone would be better off if we all cooperated but no one does', 'we keep undercutting each other even though it hurts everyone', 'how do we stop a race to the bottom', 'we need a collective agreement that actually holds', 'our cartel keeps collapsing', 'how do I stop a defection spiral', 'we need to solve a commons problem', or 'is this a coordination problem or a cooperation problem'.
version: 1.0.0
homepage: https://github.com/bookforge-ai/bookforge-skills/tree/main/books/the-art-of-strategy/skills/prisoners-dilemma-resolver
metadata: {"openclaw":{"emoji":"📚","homepage":"https://github.com/bookforge-ai/bookforge-skills"}}
status: draft
source-books:
  - id: the-art-of-strategy
    title: "The Art of Strategy: A Game Theorist's Guide to Success in Business and Life"
    authors: ["Avinash K. Dixit", "Barry J. Nalebuff"]
    chapters: [3, 9]
tags: [game-theory, cooperation, negotiation, collective-action, prisoners-dilemma]
depends-on: []
execution:
  tier: 1
  mode: plan-only
  inputs:
    - type: document
      description: "Description of the strategic situation: players, choices available, payoffs or outcomes, interaction history, relationship duration"
    - type: none
      description: "Skill can also work from a verbal description of the conflict"
  tools-required: [Read, TodoWrite]
  tools-optional: []
  environment: "Can run from any directory; operates on situation descriptions provided by the user"
discovery:
  problem-patterns:
    - "everyone cooperating is best but no one does"
    - "race to the bottom / destructive competition"
    - "commons overexploitation"
    - "cartel or collusion arrangement that keeps collapsing"
    - "collective action failure"
    - "arms race or advertising war"
    - "defection spiral that no one can stop unilaterally"
    - "coordination vs cooperation confusion"
  anti-patterns:
    - "zero-sum game: one player's gain is another's loss (no cooperation surplus exists)"
    - "single-shot interaction with no future relationship (may need external enforcement)"
    - "parties have different information about payoffs (screening/signaling skill may be more relevant)"
---

# Prisoners' Dilemma Resolver

## When to Use

Use this skill when you face a situation where:
- **All parties would benefit from mutual cooperation, but each has a private incentive to defect** — this is the defining feature of a prisoners' dilemma. Price wars, overfishing, arms races, and advertising spirals are canonical instances.
- **A collective agreement has been tried and keeps collapsing** — parties promise to cooperate, then someone cheats.
- **A commons is being overexploited** — individuals capture private gains while spreading costs across the group.
- **You need to decide between cooperation and competition** — unsure whether a situation calls for restraint or aggressive play.
- **You need to distinguish a cooperation problem from a coordination problem** — cooperation problems (prisoners' dilemma) have dominant strategies to defect; coordination problems have no incentive to deviate once aligned on a convention. The remedies differ sharply.

Preconditions: you have at least one of:
- A description of the conflict, the parties involved, and what each can choose to do
- A rough sense of who gains and who loses from each combination of choices
- Information about how long the relationship has been ongoing and how long it is expected to continue

**Agent:** Before starting, confirm: (1) how many players are involved, (2) whether this is a one-shot or repeated interaction, and (3) whether the user wants only a diagnosis or a full cooperation mechanism design. The mechanism design is the core deliverable.

## Context & Input Gathering

### Input Sufficiency Check

```
User prompt → Identify: who are the players, what can each choose, what are the rough payoffs?
                    ↓
Environment → Are there documents describing the situation, history, or prior agreements?
                    ↓
Gap analysis → Do I have enough to build a payoff table and assess the discount rate?
                    ↓
         Missing critical info? ──YES──→ ASK (one question at a time)
                    │
                    NO
                    ↓
              PROCEED with diagnosis and mechanism design
```

### Required Context (must have — ask if missing)

- **The players:** Who are the parties? How many? Are they identifiable individuals, firms, nations, or anonymous members of a group?
  → Check prompt for: named parties, "we vs. they" language, player count
  → If missing, ask: "Who are the parties in this situation? Are they a small identifiable group or a large anonymous one?"

- **The choices:** What are the "cooperate" and "defect" options for each player?
  → Check prompt for: specific behaviors like pricing, output levels, investment, restraint
  → If missing, ask: "What does cooperation look like in practice — what would each party have to do or give up? And what does defection look like — what is the tempting unilateral advantage?"

- **The payoff structure:** Is defection dominant regardless of what others do? Or does the best response depend on what others choose?
  → Check prompt for: "no matter what they do," "everyone is better off if all cooperate but each has an incentive to cheat," outcomes from prior interactions
  → If missing, ask: "When you consider your best move — is it to defect regardless of what the other party does? Or does your best move depend on what they do?"

- **Relationship horizon:** Is this a one-shot interaction or an ongoing relationship? How long is it expected to continue?
  → Check prompt for: contract duration, relationship history, expected future interactions
  → If missing, ask: "Is this a one-time interaction or an ongoing relationship? If ongoing, is there a foreseeable end date?"

### Observable Context (gather from prompt or documents)

- **Prior defection history:** Have parties cheated before? Does each party know it?
  → Relevant for: assessing whether trust has already been destroyed, choosing repair strategies

- **Observability:** Can each party monitor what others actually do, or only infer from outcomes?
  → Relevant for: mechanism design — hidden defection requires different solutions than visible defection

- **Player composition stability:** Is the group of players stable, or do new entrants frequently disrupt existing arrangements?
  → Relevant for: Ostrom principle 1 (clear membership) and end-game defection risk

### Default Assumptions

- If interaction type is unspecified: assume repeated indefinitely with low probability of each period being the last.
- If payoff magnitudes are unspecified: treat the gain from unilateral defection as moderate — sufficient to tempt but not extreme.
- If number of players is unspecified: assume two-player dilemma; flag multiperson extensions where they change the analysis.
- If observability is unspecified: assume imperfect observability (parties can observe outcomes but not always attribute them to a specific defector).

### Questioning Guidelines

Ask ONE question at a time, most critical first. Show what you already know before asking. State why you need the information.

## Process

Use `TodoWrite` to track steps before beginning.

```
TodoWrite([
  { id: "1", content: "Diagnose: confirm prisoners' dilemma vs coordination problem", status: "pending" },
  { id: "2", content: "Map payoffs: build rough payoff table with cooperation surplus", status: "pending" },
  { id: "3", content: "Assess discount rate: calculate cooperation sustainability threshold", status: "pending" },
  { id: "4", content: "Score mechanism prerequisites: detection, clarity, certainty, size, repetition", status: "pending" },
  { id: "5", content: "Select and design cooperation mechanism from resolution menu", status: "pending" },
  { id: "6", content: "For commons/multiperson: apply Ostrom's design principles checklist", status: "pending" },
  { id: "7", content: "Produce cooperation design plan with implementation steps", status: "pending" }
])
```

---

### Step 1: Diagnose the Game Type

**ACTION:** Determine whether the situation is a genuine prisoners' dilemma, a coordination problem, or something else. This diagnosis determines the entire remedy.

**WHY:** The treatments are fundamentally different. A prisoners' dilemma has a dominant strategy to defect — each player is better off defecting regardless of what others do. A coordination problem has multiple equilibria — once aligned on the same convention (QWERTY, driving on the right, common standards), no one wants to deviate unilaterally. Applying cooperation mechanisms to a coordination problem is wasted effort; what's needed is a focal point or a critical mass to tip behavior.

**Two-question diagnostic:**

1. "If the other party cooperates, am I still better off defecting?" (unilateral defection payoff)
2. "If the other party defects, am I still better off defecting?" (defection-regardless test)

If YES to both → prisoners' dilemma. Defection is dominant. Proceed through this skill.

If NO to question 1 (defection hurts me when others cooperate) → coordination problem. The issue is aligning expectations, not suppressing defection. See coordination notes in References.

If YES to question 1 but NO to question 2 → asymmetric game. One party may have a dominant defection strategy while the other's best response depends on context. Requires separate analysis.

**Also check: is this zero-sum?** In a zero-sum game, every gain by one player comes at exactly equal cost to another. Zero-sum games have no cooperation surplus — there is nothing to gain from mutual restraint. The prisoners' dilemma is NOT zero-sum: both players are better off in the mutual cooperation cell than in the mutual defection cell.

**Confirm the cooperation surplus exists:** There must be a combination of choices where all parties are better off than in the mutual-defection equilibrium. If no such combination exists, this is not a prisoners' dilemma.

Mark Step 1 complete in TodoWrite.

---

### Step 2: Map the Payoffs

**ACTION:** Construct a rough payoff table showing the four cells: (Cooperate, Cooperate), (Cooperate, Defect), (Defect, Cooperate), (Defect, Defect). For multiperson dilemmas, show how collective payoff changes as the number of cooperators rises.

**WHY:** The payoff table makes the temptation and its cost concrete. Without it, the mechanism design is abstract. The payoff structure also determines which mechanisms are viable — specifically, the ratio of the one-period defection gain to the ongoing cooperation gain determines the discount-rate threshold (Step 3).

**Standard prisoners' dilemma payoff ordering (for each player):**
- Defect while other cooperates > Mutual cooperation > Mutual defection > Cooperate while other defects
- In shorthand: T > R > P > S (Temptation > Reward > Punishment > Sucker)

**Quantify if possible:**
- One-period defection gain = T − R (what you gain in the short term by cheating)
- Annual cooperation gain loss = R − P (what you lose each year if cooperation collapses)
- Break-even interest rate (see Step 3): (R − P) / (T − R)

**For multiperson / commons situations:** Replace the 2x2 table with a contribution schedule showing how each party's payoff changes as the number of cooperators increases. Key feature: each defector gains a fixed amount regardless of how many others defect, but spreads a cost across all cooperators. This is the "contribution game" structure.

Mark Step 2 complete in TodoWrite.

---

### Step 3: Calculate the Discount-Rate Threshold

**ACTION:** Determine the maximum interest rate (discount rate) at which sustained cooperation is rational. If the actual interest rate or impatience level is below this threshold, cooperation through repeated play is sustainable without external enforcement.

**WHY:** The key insight of repeated-game analysis is that defection gains a short-term advantage but destroys long-term cooperation value. Whether defection is worth it depends on how much you value the future. If you are very impatient (high discount rate), the future is worth little and defection becomes attractive even knowing the relationship will collapse. If you value the future sufficiently (low discount rate), the prospect of losing ongoing cooperation outweighs the temptation to cheat today.

**Formula:**

Cooperation is self-sustaining if the interest rate r satisfies:

```
r < (R − P) / (T − R)
```

Where:
- R = payoff from mutual cooperation (reward)
- P = payoff from mutual defection (punishment)
- T = payoff from defecting while other cooperates (temptation)
- T − R = one-period gain from cheating
- R − P = annual cost of losing the cooperative relationship

**Worked example (from Ch. 3, Rainbow's End / B.B. Lean pricing):**
- Mutual cooperation: $72,000 each per year (R)
- Mutual defection: $70,000 each per year (P)
- One-period defection gain: $110,000 − $72,000 = $38,000 (T − R)
- Annual cost of collapsed cooperation: $72,000 − $70,000 = $2,000 (R − P)
- Break-even rate: $2,000 / $38,000 = 5.26% per year
- If actual interest rate > 5.26%: cooperation collapses; if < 5.26%: self-sustaining

**What raises the sustainability threshold (makes cooperation easier):**
- Longer shadow of the future: indefinite interactions, no known end date
- Growing relationship value: stakes grow over time, making defection increasingly costly
- Stable player composition: no new entrants who don't share the history
- Frequent interactions: each "period" is shorter, making each defection opportunity less rewarding

**What lowers the sustainability threshold (makes cooperation harder):**
- High discount rates: impatience, financial distress, declining industry
- End-game visibility: clear final period triggers backward-induction unraveling
- Low cooperation surplus: R − P is small relative to T − R
- Business booms: temporary windfalls make defection more tempting right now

**End-game problem:** In a finite repeated game with a known final period, backward induction unravels cooperation all the way to round 1. Solution: eliminate the clear end-game — use indefinite time horizons, rolling contracts, or overlapping agreements so there is no obvious "last round."

Mark Step 3 complete in TodoWrite.

---

### Step 4: Score the Punishment Mechanism Prerequisites

**ACTION:** Evaluate the situation against five prerequisites for an effective punishment-based cooperation mechanism. Each gap identifies a specific design problem to solve.

**WHY:** Punishment is the most common mechanism for sustaining cooperation. But punishment only deters defection if it meets specific structural requirements. A gap in any one of the five areas undermines the entire mechanism. Identifying which prerequisite is weak tells you exactly what the mechanism design needs to fix.

**The five prerequisites:**

**1. Detection** — Can defection be observed, attributed to the right party, and detected quickly?
- Fast, accurate detection allows immediate targeted punishment, reducing the gain from cheating.
- Slow or inaccurate detection gives defectors a long free ride before punishment arrives.
- Ambiguous attribution (cannot tell WHO cheated, only that cheating occurred) forces blunt punishments that hurt cooperators too.
- Design tool: make defection observable by structure — arrange for cheating to immediately surface (price-matching clauses put customers in charge of detection; lunar-calendar bid rotation made cheating immediately visible to rivals).

**2. Clarity** — Are the rules of cooperation and the boundaries of acceptable behavior unambiguous?
- If boundaries are complex, parties may cheat by mistake or fail to make a rational calculation. Tit-for-tat's great strength is its clarity: if you cheat, I cut prices immediately next period. No ambiguity.
- Collusion cartels that fail often fail because what counts as cheating is unclear (price cuts vs. quality upgrades vs. extended credit terms).

**3. Certainty** — Is punishment guaranteed when defection occurs? Is cooperation reliably rewarded?
- Uncertain punishment (WTO-style multi-year adjudication where political considerations override facts) provides weak deterrence.
- Design tool: make punishment automatic and mechanical — most-favored-customer clauses, automatic price-matching policies.

**4. Size** — Is the punishment large enough to deter, but not so large that errors cause catastrophic spirals?
- Minimum deterrent punishment is preferable: set punishment at the level required to make defection unprofitable, no larger. Larger punishments amplify errors.
- When errors are possible (as they always are in practice), forgiveness of occasional defections is optimal — punishment should be as low as is compatible with deterrence.

**5. Repetition** — Is there a sufficient "shadow of the future"? (Covered in Step 3 — confirm result applies here.)
- Self-enforcing cooperation requires future value to exceed the one-period defection gain (the threshold from Step 3).

Mark Step 4 complete in TodoWrite.

---

### Step 5: Select and Design the Cooperation Mechanism

**ACTION:** Choose the appropriate mechanism from the resolution menu below. Mechanisms are ordered from lowest to highest external requirement. Select the lowest-level mechanism that is feasible given the prerequisites scored in Step 4.

**WHY:** Not every situation requires the same intervention. Using heavy external enforcement when self-enforcement would work wastes resources and creates regulatory risk (antitrust exposure for firms). Using self-enforcement when the prerequisites are missing produces predictable failure. Matching mechanism to situation is the core design task.

**Resolution Menu (escalating external requirement):**

---

**Level 1 — Self-enforcing repeated play (tit-for-tat or generous variant)**

*Use when:* discount rate is below threshold (Step 3 result), players interact repeatedly, detection is feasible.

*How it works:* Players cooperate initially, then mirror the other's last move. The credible threat of future retaliation deters current defection.

*Standard tit-for-tat:* Cooperate first; replicate opponent's last move every subsequent period.
- Strengths: clear, nice (never initiates defection), provocable (punishment is immediate), forgiving (restores cooperation after one cooperative move).
- Fatal flaw in noisy environments: any error or misperception triggers an alternating defection spiral (Hatfield-McCoy pattern). Each side punishes what it perceived as defection, but the other side is merely responding to the punishment.

*Generous tit-for-tat (preferred in practice):* Cooperate first; punish sustained defection but forgive isolated defections. The tamarin monkey threshold: tolerate up to 1 defection, punish 2 consecutive defections.
- Advantage: breaks punishment spirals caused by noise or misperception.
- Rule: Do not respond to a single defection. If the other party defects twice in a row, switch to defect until they cooperate twice in a row.

*Implementation checklist:*
- [ ] Define what counts as cooperation and defection in operational terms
- [ ] Set the monitoring interval (how quickly is defection detected and responded to?)
- [ ] Choose forgiveness threshold (1 defection? 2 consecutive? context-dependent)
- [ ] Communicate the strategy clearly to the other party — clarity is a prerequisite

---

**Level 2 — Mutual promises with escrow or simultaneous commitments**

*Use when:* Tit-for-tat is viable but trust is currently depleted; players need a credibility boost to restart cooperation.

*How it works:* Both parties make simultaneous promises and deposit promised payments (or penalties for non-performance) in a neutral escrow account. Neither can renege without forfeiting the escrow. Converts soft promises into hard commitments.

*Use cases:* Restart of cooperation after a defection episode; situations where one party is less patient and needs a structural guarantee; situations where the relationship is new and no reputation exists yet.

---

**Level 3 — Reputation and linkage across multiple interactions**

*Use when:* The dyadic interaction has insufficient cooperation surplus to sustain self-enforcement on its own, but the parties interact in multiple dimensions.

*Reputation mechanism:* Public record of each party's cooperation history creates external cost for defection beyond the bilateral relationship. A firm that cheats on a pricing arrangement faces skepticism from future trading partners, lenders, and employees. Works best when: the reputation is observable to third parties who matter, defection is clearly attributable, and the relationship horizon extends far enough to make the reputation investment worthwhile.

*Linkage mechanism:* Bundle multiple interactions so defecting in one dimension risks the entire relationship. Cooperation surplus across all linked interactions must exceed the temptation to defect in any one. Warning: linkage scales both gains and defection gains proportionally if all dimensions have identical payoff structures — benefit comes only from asymmetries across dimensions.

---

**Level 4 — Third-party intervention and external enforcement**

*Use when:* Self-enforcement is not viable (relationship is too short, discount rate too high, detection too imperfect), and parties cannot credibly commit to punishments themselves.

*Options:*
- **Contract enforcement:** Parties write an explicit agreement specifying cooperation terms and penalties for defection. A court or arbitrator enforces it. Requires: clear specification of what counts as defection (not always possible for tacit understandings like pricing).
- **Third-party mediator/arbitrator:** A neutral third party with sufficient authority and interest in cooperation (e.g., Camp David mediator rewarding Egypt and Israel for cooperating). Third party provides both a focal point for what cooperation means and a punishment mechanism for defection.
- **Regulatory prohibition of defection:** Government makes the defect option illegal (antitrust law prevents cartelization, but also prevents the cooperation problem from arising). Note: this works in both directions — antitrust also prevents self-enforcing collusion among competitors, which is why it must be used selectively.

---

**Level 5 — Ostrom commons governance (multiperson dilemmas)**

*Use when:* The dilemma involves a large group managing a shared resource (fishery, groundwater, pasture, shared infrastructure, open-source contribution, common standards).

Apply all eight Ostrom design principles as a checklist. Each principle maps to a specific prerequisite gap:

1. **Defined boundaries** — Clear rules about who has the right to use the resource. Geographic, professional, or membership criteria. Prevents free-rider entry by outsiders.

2. **Rules match local conditions** — Usage rules (time restrictions, location restrictions, technology limits, quantity quotas) must be calibrated to what is actually detectable and enforceable in this specific context. Quantity quotas work well when quantities are easily observable; they fail for fish because catch size is hard to control exactly. Time-based and location-based rules are often more enforceable.

3. **Graduated sanctions** — Punishment starts low (verbal warning, direct approach to the violator) and escalates only for repeat or severe violations. First-offense fines are low; they ratchet up. This principle exists because light initial punishment maintains community relationships while still deterring escalation, and because the first violation might be a misunderstanding rather than deliberate defection.

4. **Automatic detection embedded in operations** — Design the governance system so monitoring happens as a byproduct of normal operations (rotation systems where the person with the good spot automatically notices if someone else is using it; team harvesting that makes solo overharvesting visible). Dedicated guard systems are costly and generate evasion; embedded detection is cheap and hard to game.

5. **Locally designed and adjusted rules** — The group that uses the resource has the information to design rules that are both effective and legitimate. Top-down management consistently fails because it lacks local knowledge of the resource, the technology, and the community norms. The group should design its own rules through a participatory process.

6. **Conflict resolution mechanisms** — Low-cost dispute resolution must be available to address perceived violations without escalating to punishment spirals. The first response to apparent cheating should be inquiry, not punishment.

7. **Recognition by external authority** — External governments must recognize the community's right to organize and enforce its own rules, not override or undermine local governance.

8. **Nested governance for large systems** — For large-scale resources, governance is organized in multiple layers — local groups handle local problems; coordination bodies handle cross-group issues. Monolithic centralized governance fails at scale; purely local governance fails at the system level.

**Ostrom's warning:** "The dilemma never fully disappears, even in the best operating systems. No amount of monitoring or sanctioning reduces the temptation to zero. Effective governance systems cope better than others — they do not eliminate the problem."

Mark Step 5 complete in TodoWrite.

---

### Step 6: For Multiperson Dilemmas — Assess Contribution Game Structure

**ACTION:** If the dilemma involves more than two players and a public good or shared resource, assess the contribution game structure: each party's dominant strategy is to free-ride; collective optimum requires all to contribute.

**WHY:** Multiperson prisoners' dilemmas have a specific feature that bilateral dilemmas lack: each defector's gain is fixed (they free-ride on others' contributions), but each defector imposes a cost spread across ALL cooperators. In the 4-player contribution game: contributing $1 to the pool raises total benefit by $2 (after doubling), but the contributor receives only $0.50 of the gain ($1.50 goes to others). This makes free-riding dominant regardless of what others do.

**Contribution game checklist:**
- [ ] Identify the collective good and who benefits from it regardless of contribution
- [ ] Identify the private gain from free-riding and the collective cost
- [ ] Assess whether the group is small enough for social sanctions to operate (village-scale vs. anonymous urban setting)
- [ ] Assess whether punishment is available: can group members observe who contributed and impose social costs on defectors?
- [ ] Note: experimental evidence shows people will pay a personal cost to punish free-riders (third-party punishment), which activates the dorsal striatum — the biological basis of cooperative norm enforcement

Mark Step 6 complete in TodoWrite.

---

### Step 7: Produce the Cooperation Design Plan

**ACTION:** Write a structured cooperation design plan covering the full analysis from Steps 1-6.

**WHY:** The plan must be specific and actionable. "Use tit-for-tat" is not useful. "Define price cuts of more than 5% as defection, respond with a 10% price cut effective next catalog cycle, forgive single-period deviations, treat two consecutive deviations as intentional" is useful — it specifies the operational terms the parties need to implement.

**HANDOFF TO HUMAN** — the agent produces the plan; the human negotiates, implements, and monitors.

**Plan format:**

```markdown
# Cooperation Design Plan

## Game Diagnosis
**Type:** [Prisoners' Dilemma / Coordination Problem / Asymmetric / Not applicable]
**Cooperation surplus exists:** [Yes/No — the mutual cooperation cell vs. mutual defection cell]
**Payoff structure:**
| | Other: Cooperate | Other: Defect |
|---|---|---|
| You: Cooperate | R = [value] | S = [value] |
| You: Defect | T = [value] | P = [value] |

**Dominant strategy:** [Defect / Cooperate / Context-dependent]

## Discount-Rate Assessment
**One-period defection gain (T − R):** [value]
**Annual cooperation value at stake (R − P):** [value]
**Break-even interest rate:** [calculation and result]
**Self-enforcement viable?** [Yes if actual rate < break-even / No if above]

## Prerequisite Gaps
| Prerequisite | Status | Gap Description |
|---|---|---|
| Detection | [Strong/Weak/Missing] | [specific gap] |
| Clarity | [Strong/Weak/Missing] | [specific gap] |
| Certainty | [Strong/Weak/Missing] | [specific gap] |
| Size (minimum deterrent) | [Strong/Weak/Missing] | [specific gap] |
| Repetition / shadow of future | [Strong/Weak/Missing] | [specific gap] |

## Recommended Mechanism
**Level selected:** [1–5 from resolution menu]
**Mechanism:** [Name and brief description]
**Rationale:** [Why this level fits the situation]

## Implementation Steps
1. [Operational definition of cooperation and defection in this context]
2. [Monitoring arrangement: how, by whom, with what frequency]
3. [Response rule: specific trigger and specific response]
4. [Forgiveness threshold: when to restore cooperation]
5. [Communication plan: how to make the strategy clear to all parties]
6. [Escalation path: if self-enforcement fails, what is the next level?]

## Risks and Anti-Patterns
- **End-game defection risk:** [Is there a visible final period? How to address?]
- **Punishment spiral risk:** [Is generous tit-for-tat needed? What forgiveness threshold?]
- **Player composition risk:** [Are new entrants expected? How does the mechanism handle them?]
- **Boom/bust defection timing:** [When is defection temptation highest? Special provisions needed?]

## For Commons/Multiperson Situations
Ostrom principle compliance:
- [ ] Defined boundaries
- [ ] Rules match local conditions
- [ ] Graduated sanctions
- [ ] Automatic detection embedded in operations
- [ ] Locally designed rules
- [ ] Conflict resolution mechanisms
- [ ] External recognition
- [ ] Nested governance (for large systems)
```

Mark Step 7 complete in TodoWrite.

## Inputs

- **Situation description:** Who are the players? What choices do they have? What happens to each player under each combination of choices?
- **Relationship context:** How long has the relationship been ongoing? Is it expected to continue indefinitely or does it have a clear end date?
- **Observability:** Can each party monitor what others actually do, or only observe outcomes?
- **Prior cooperation history:** Have parties cooperated or defected in the past? What happened?

## Outputs

- **Cooperation Design Plan** (Markdown) — complete structured plan with diagnosis, payoff table, discount-rate threshold, prerequisite gap assessment, mechanism selection and implementation steps, risk register, and Ostrom compliance checklist where applicable
- **Decision rationale** — for each recommendation, the WHY (which prerequisite is met or missing, why this mechanism level was chosen)

## Key Principles

- **Diagnosis before mechanism** — the correct remedy depends entirely on the game type. Applying cooperation tools to a coordination problem (or vice versa) is worse than doing nothing because it misidentifies the problem and wastes resources. Always confirm whether defection is dominant regardless of others' choices.

- **The cooperation surplus is the prize** — the difference between mutual cooperation and mutual defection payoffs is what parties are fighting over. If this surplus is small relative to the temptation to defect, self-enforcement requires either an extremely low discount rate or an external mechanism. The surplus size determines the viable mechanism space.

- **The future must be sufficiently valuable** — cooperation is self-enforcing only when the prospect of losing ongoing cooperation outweighs the one-period temptation gain. This is not a moral claim; it is an arithmetic one. Calculate the break-even discount rate explicitly. Fuzzy talk about "long-term relationships" is insufficient.

- **Punishment must be targeted, certain, and proportionate — not maximal** — the temptation to set catastrophic punishments ("nuke any country that breaks the tariff agreement") is wrong because errors will occur. When errors occur, catastrophic punishments are either not credible or produce catastrophic outcomes. Set punishment at the minimum level that makes defection unprofitable.

- **Standard tit-for-tat is fragile in noisy environments** — the real world has errors and misperceptions. Tit-for-tat cannot distinguish intentional defection from noise, and its perfect responsiveness creates punishment spirals. Generous tit-for-tat (2-consecutive-defections threshold) preserves punishment credibility while tolerating noise.

- **Cooperation and coordination are different problems** — cooperation problems (prisoners' dilemma) have dominant strategies to defect; the challenge is suppressing that incentive. Coordination problems have no incentive to deviate from the prevailing convention once everyone is on it; the challenge is getting to the right convention and escaping a bad one. Cigarette advertising bans, QWERTY entrenchment, and racial tipping are coordination problems. Arms races, price wars, and overfishing are cooperation problems.

- **Ostrom: no governance system eliminates the dilemma** — the goal is a governance system that manages the dilemma better than alternatives, not one that eliminates it. Perfection is not the standard. Practical improvement — fewer defections, faster detection, lower cost of enforcement — is.

## Examples

**Example 1: Pricing cartel between two mail-order retailers**

Situation: Rainbow's End (RE) and B.B. Lean (BB) both price shirts at $70 when both could price at $80 and each earn $72,000 vs. $70,000 per year. Each firm cuts to $70 because it's the dominant strategy: cutting while the other holds at $80 yields $110,000; holding at $80 while the other cuts yields only $24,000.

Diagnosis: Classic 2-player prisoners' dilemma. T=$110k, R=$72k, P=$70k, S=$24k. Defection dominant for both.

Discount-rate calculation: (R−P)/(T−R) = ($72k−$70k)/($110k−$72k) = $2k/$38k = 5.26%. If prevailing interest rate < 5.26%, tacit cooperation at $80 is self-sustaining.

Mechanism: Level 1 (self-enforcing repeated play). Detection: price lists are publicly observable. Clarity: define "defect" as cutting below $80. Response: immediately match any price cut in next catalog. Forgiveness: if other party restores $80 pricing, match that too. Communication: a "most-favored-customer" clause makes the automatic response policy public, removing ambiguity. Anti-pattern avoided: no explicit agreement is reached (antitrust risk); cooperation is purely tacit.

---

**Example 2: Fishery commons overexploitation**

Situation: New England fishing fleet: each captain has incentive to catch as much as possible before others do; result is collapse of species after species (Atlantic halibut, ocean perch, haddock).

Diagnosis: Multiperson prisoners' dilemma (contribution game). Each additional catch by one captain reduces the stock for all others. Dominant strategy: fish aggressively regardless of what others do.

Discount-rate calculation: Not determinative on its own — relationship continues indefinitely but individual boats can't unilaterally enforce rules against strangers.

Mechanism: Level 5 (Ostrom commons governance). Apply 8-principle checklist:
1. Boundaries: issue licenses to fish specific species in specific zones — clear membership
2. Rules match conditions: seasonal closures and gear restrictions (net size) — more observable than quantity quotas
3. Graduated sanctions: first violation = warning + education; second = fine; third = license suspension
4. Automatic detection: rotating assignment to prime fishing zones creates natural monitoring — the assigned captain notices unauthorized use immediately
5. Local design: fishing community designs the quota and rotation rules with knowledge of local conditions, not federal agency
6. Conflict resolution: fishing association mediation before formal sanction
7. External recognition: state and federal agencies recognize community fishing governance authority
8. Nested governance: local associations handle local waters; interstate compact handles migratory species

---

**Example 3: Coordination problem misdiagnosed as cooperation problem**

Situation: Ivy League colleges keep overspending on athletics even though the relative standings stay the same. "Each school would be better off if we all limited spring training to one day."

Diagnosis check: "If other schools limit training, should I limit training?" → Yes, my performance improves no more than theirs, but I save costs. "If other schools don't limit training, should I limit training?" → No, I'd be at a disadvantage. This is NOT a prisoners' dilemma. Defection is NOT dominant regardless of others' choices. The best response to "others cooperate" is to cooperate; the best response to "others defect" is to defect. This is a coordination problem.

Mechanism: Not Level 1-5 from the resolution menu. Instead: establish a focal point for the cooperative equilibrium through a collective agreement with clear enforcement (the Ivy League agreement limiting spring training to one day). Once the convention is established and everyone expects everyone else to comply, compliance becomes self-sustaining without punishment — because no one wants to be the only school overdoing training when no one else is.

Key distinction that matters: if this were a genuine prisoners' dilemma, the agreement would keep collapsing despite everyone's stated preference for cooperation. In coordination problems, a credible agreement is usually sufficient because there is no dominant strategy to defect — just a fear that others won't cooperate.

## References

- For the Ostrom 8-principle checklist with detailed examples and design notes, see [ostrom-commons-governance.md](references/ostrom-commons-governance.md)
- For the resolution menu with full specification of each mechanism level, see [resolution-menu.md](references/resolution-menu.md)
- For the cooperation vs. coordination distinction with examples, see [cooperation-vs-coordination.md](references/cooperation-vs-coordination.md)
- Source: *The Art of Strategy*, Dixit & Nalebuff, Chapter 3 (pp. 74–105) and Chapter 9 (pp. 254–280)

## License

This skill is licensed under [CC-BY-SA-4.0](https://creativecommons.org/licenses/by-sa/4.0/).
Source: [BookForge](https://github.com/bookforge-ai/bookforge-skills) — The Art of Strategy: A Game Theorist's Guide to Success in Business and Life by Avinash K. Dixit, Barry J. Nalebuff.

## Related BookForge Skills

This skill is standalone. Browse more BookForge skills: [bookforge-skills](https://github.com/bookforge-ai/bookforge-skills)

FILE:references/cooperation-vs-coordination.md
# Cooperation vs. Coordination: Two Different Problems, Two Different Remedies

Source: Dixit & Nalebuff, *The Art of Strategy*, Chapter 9 (pp. 254–280)

---

## The Core Distinction

**Cooperation problem (prisoners' dilemma):**
Defection is the dominant strategy — the best response regardless of what others do. Everyone would be better off if all cooperated, but each individual has an incentive to defect even when everyone else cooperates. The challenge is suppressing the incentive to defect.

**Coordination problem:**
There is no dominant strategy to deviate from the prevailing equilibrium. Once enough people are doing the same thing, no individual benefits from switching alone. The challenges are: (1) getting to a good equilibrium in the first place, (2) escaping a bad equilibrium once established, and (3) preventing tipping to an inferior equilibrium.

**The diagnostic question:**
"If everyone else cooperates, do I still benefit from defecting?"
- YES → Prisoners' dilemma (dominant defection strategy). Use cooperation mechanisms (Levels 1-5).
- NO → Coordination problem. Use focal points, conventions, and tipping mechanisms.

---

## Cooperation Problems: Key Features

**Defining feature:** Defection dominant regardless of others' choices.

**Common manifestations:**
- Price wars between competitors (each firm benefits from cutting prices regardless of what the other does)
- Overfishing (each captain benefits from catching more regardless of what others do)
- Arms races (each party benefits from building arms regardless of what others do)
- Advertising wars (each firm must advertise because rivals do, but collective advertising spend is waste)
- Free-riding on public goods (each individual benefits from not contributing regardless of others' contributions)
- Overworking in relative-performance evaluations (studying more improves rank regardless of what others study)

**Remedy:** Suppress the dominant defection strategy through punishment mechanisms, external enforcement, repeated play, escrow, or governance structures. The cooperation surplus (value of mutual cooperation over mutual defection) is the resource being captured.

**Key diagnostic test:** In the payoff table, check whether the column "defect" dominates the column "cooperate" for all players. If yes, prisoners' dilemma confirmed.

---

## Coordination Problems: Key Features

**Defining feature:** Multiple equilibria, each self-sustaining once adopted. No one wants to deviate unilaterally from the prevailing convention. The problem is not suppressing defection — it is getting to the right equilibrium.

**Three sub-types:**

### Type A: Pure Coordination (no equilibrium is clearly better — just need alignment)
Example: Two people want to meet in New York but didn't specify where. Both know the prominent places (Times Square, Empire State Building). Either meeting point is fine; the only requirement is choosing the same one. Thomas Schelling called this a "focal point" — the natural, salient choice that people gravitate toward without communication.

**Remedy:** Establish a focal point (the most prominent, natural, or historically precedented option) and communicate it. Communication alone is usually sufficient because both parties want the same outcome.

### Type B: Inferior equilibrium lock-in (one equilibrium is better, but history chose the worse one)
Examples: QWERTY keyboard, gasoline engines, light-water nuclear reactors. A historical accident gave one option a head start. The bandwagon effect locked that option in: the more people use it, the more new users choose it, reinforcing its dominance. Even if a better alternative exists (Dvorak keyboard), switching requires coordinated action by enough people to tip the equilibrium.

**Key insight:** History matters. The option that happened to get an early lead may persist even when it is technically inferior, because superior alternatives never get a chance to accumulate the learning and network effects needed to catch up.

**Remedy:** Identify the tipping point (the fraction of users who need to switch to make the new equilibrium self-sustaining). Coordinate a critical mass to switch simultaneously or in a short burst. A short, intense campaign is more effective than gradual pressure because it tips the equilibrium all at once rather than allowing slow reversion. A major employer or standard-setter (government, large firm) switching can provide enough of a toehold to start the bandwagon rolling toward the better equilibrium.

**Anti-remedy:** Gradual voluntary switching without a mechanism for coordination. Each individual's switching costs exceed their individual benefit; only when enough people switch simultaneously does the benefit materialize. Gradual pressure dissipates without reaching the tipping point.

### Type C: Coordination with spillovers (each person's choice imposes costs or benefits on others)
Examples: Highway congestion (more drivers on the Bay Bridge imposes delay on all other drivers), car size arms race (driving a larger car makes lighter cars less safe, inducing others to buy larger cars), neighborhood tipping (racial composition equilibria).

**Key insight:** The Nash equilibrium (where each individual optimizes given others' choices) may not be the social optimum because spillover effects are not priced. Too many people drive on the Bay Bridge because each driver doesn't pay for the delay they impose on others.

**Remedy:** Price the spillover (congestion tolls), regulate the behavior (CAFE fuel economy standards that shift the mix of cars), or use public information to reshape expectations (preventing panic selling of houses by suppressing "For Sale" signs and providing property value insurance, as Oak Park, Illinois did).

---

## Decision Tree for Choosing the Remedy

```
Is defection dominant regardless of others' choices?
        │
       YES → Prisoners' dilemma
        │     → Use resolution menu (Levels 1-5)
        │     → Focus on punishment mechanisms and cooperation prerequisites
        │
       NO → Coordination problem
              │
              Is there only one possible equilibrium, just need alignment?
              │
             YES → Pure coordination
              │     → Find/create a focal point
              │     → Communicate clearly; explicit agreement may be all that's needed
              │
              NO → Are we locked into an inferior equilibrium?
                    │
                   YES → Inferior lock-in (bandwagon problem)
                    │     → Find the tipping point
                    │     → Coordinate a critical mass to switch simultaneously
                    │     → Short intense enforcement > long gradual pressure
                    │
                    NO → Coordination with spillovers
                          → Price the externality (Pigouvian tax/subsidy)
                          → Or regulate to shift the equilibrium
                          → Or redesign the game so choices internalize spillover costs
```

---

## Case Comparisons: Same Surface Appearance, Different Game

| Situation | Prisoners' dilemma or coordination? | Why |
|---|---|---|
| Price war between two retailers | Prisoners' dilemma | Cutting price is dominant regardless of rival's pricing |
| Ivy League athletic spending arms race | Coordination problem | If others limit training, limiting training is best response too |
| Overfishing | Prisoners' dilemma | Catching more is dominant regardless of others' behavior |
| QWERTY keyboard lock-in | Coordination (inferior equilibrium) | Everyone learns QWERTY because everyone else uses it; no one benefits from switching alone |
| Cigarette advertising wars | Prisoners' dilemma | Each company must advertise defensively regardless of rivals |
| Racial neighborhood tipping | Coordination with spillovers | Households' moving decisions affect others' moving incentives |
| SUV arms race / car size | Coordination with spillovers | Each large car makes others less safe, inducing more large cars |
| Tax compliance norms | Coordination (multiple equilibria) | If most evade, evading is best response; if most comply, complying is safer |

---

## The Hotelling Median Voter / Product Homogeneity Pattern

A specific coordination outcome: when firms or political parties compete for customers/voters on a single dimension, both converge to the median position. Neither benefits from deviating to the flank while the other is at the center, because the center captures more votes/customers.

This produces excessive homogeneity — political parties that sound alike, products that are indistinguishable ("cider is too homogeneous" — Harold Hotelling, 1929).

This is NOT a prisoners' dilemma. Deviating to an extreme position when the rival is at the center REDUCES your market share. The self-interest of each party produces homogenization, not the mutual-defection trap of a prisoners' dilemma.

**Remedy:** Zoning regulations on product positioning, mandatory product differentiation, or structural changes that remove the single-dimension competition (adding more dimensions of competition changes the equilibrium structure).

---

## When Remedies Fail Because the Diagnosis Was Wrong

**Applying cooperation mechanisms to a coordination problem:** Parties sign agreements, use escrow, and set up punishment systems — but none of this is needed because no one has an incentive to defect unilaterally anyway. Resources are wasted on enforcement for a problem that would have self-resolved through alignment.

**Applying coordination mechanisms to a prisoners' dilemma:** Establishing a focal point or achieving alignment on a convention does nothing to suppress the dominant defection strategy. Parties agree on what cooperation looks like and then defect anyway because defection is dominant. The agreement collapses because signing it was cheap and defecting on it is profitable.

**The diagnostic test is therefore essential before any remediation effort:**
1. Build the payoff table
2. Check whether defection is dominant
3. Choose the remedy type accordingly

FILE:references/ostrom-commons-governance.md
# Ostrom's 8 Design Principles for Commons Governance

Source: Elinor Ostrom (Indiana University), derived from case studies of successful and unsuccessful attempts to manage common-pool resources — fisheries, groundwater, pastures, forests. Nobel Prize in Economics, 2009.

## The Core Problem

A commons (shared resource) faces a multiperson prisoners' dilemma. Each user has a dominant strategy to overexploit because:
- Each unit they take is a private gain
- The cost of depletion is spread across ALL users
- If others will overexploit anyway, restraint is irrational; if others restrain, you capture extra gain by not restraining

Result without governance: tragedy of the commons — collapse of the resource.

## Ostrom's Warning

> "The dilemma never fully disappears, even in the best operating systems. No amount of monitoring or sanctioning reduces the temptation to zero. Instead of thinking of overcoming or conquering tragedies of the commons, effective governance systems cope better than others."

The goal is not elimination of the dilemma but management of it. Set realistic expectations.

---

## The 8 Design Principles

### Principle 1: Defined Boundaries

**What:** Clear rules identifying who is a member of the governing group — who has the right to use the resource.

**Criteria for membership:** Geography (fishing zone), residence (local watershed), occupation (licensed fishermen), ethnicity or community membership (historical commons), or market mechanisms (auction entry fee or purchased license).

**Why it matters:** Without defined membership, new entrants can continuously dilute the cooperation arrangement. Every new member who does not share the history and social norms of the group weakens the enforcement mechanism. Open access destroys commons governance.

**Design question:** "Can we unambiguously tell whether a person is or is not entitled to use this resource?" If not, the governance system cannot close its borders to free-riders.

**Example:** New England fishing licenses tied to specific vessel/captain combinations and species — outsiders cannot join the pool informally.

---

### Principle 2: Rules Match Local Conditions

**What:** Usage restrictions (time, location, technology, quantity) must be calibrated to what is detectable and controllable in this specific resource context.

**Why it matters:** Top-down governance consistently imposes inappropriate rules because it lacks local knowledge of the resource, the technology for exploiting it, and the practical feasibility of detection. Quantity quotas for fish fail because catch size is difficult to observe exactly and hard for even well-intentioned fishermen to control precisely. Seasonal closures and gear restrictions (net size, mesh size) are more enforceable because violations are visible.

**Design question:** "What can a community member actually observe about compliance or violation during their normal operations?"

**Rule types and their observability:**
| Rule type | Observability | Example |
|---|---|---|
| Quantity quotas | Low (fish are underwater) | Fishing quota — rarely effective alone |
| Gear/technology restrictions | High (equipment is visible) | Net mesh size, hook count limits |
| Time restrictions (seasons) | High (calendar is public) | Open/closed seasons for hunting, fishing |
| Location restrictions | Moderate (rotation visible) | Inshore rotation rights, assigned fishing zones |
| Water/resource flow | High when metered | Irrigation water from storage tank |

**Anti-pattern:** Federal fishery managers imposing quantity-based individual transferable quotas without the monitoring infrastructure to enforce them — compliance rates drop, black market develops.

---

### Principle 3: Graduated Sanctions

**What:** Punishment for violations starts light and escalates only for repeat or severe infractions.

**Scale:**
1. Direct approach to violator + verbal request to resolve the problem
2. Public warning / social shame within the community
3. Small fine (first or second offense)
4. Escalating fines for repeated violations
5. Loss of future rights / suspension
6. Expulsion from the community
7. Criminal prosecution (extreme cases)

**Why it matters:** First violations are often genuine errors, misunderstandings, or emergency situations (a fisherman fighting for survival may take more than their quota — this is the "extreme situation" that warrants forgiveness). Severe punishment for a first violation destroys community relationships unnecessarily and may not be credible. Light initial punishment maintains social cohesion while still deterring casual defection. Punishment credibility increases because light punishments will actually be administered, whereas community members resist invoking catastrophic punishments for minor violations.

**Design question:** "Would community members actually be willing to enforce this sanction against a neighbor? If not, reduce the sanction until they would."

**Connection to tit-for-tat:** Graduated sanctions are the commons governance analogue to generous tit-for-tat — tolerate occasional defections, respond firmly to sustained or repeated defections.

---

### Principle 4: Automatic Detection Embedded in Operations

**What:** Design the governance system so monitoring happens as a natural byproduct of how participants normally use the resource — not as a separate, costly guard function.

**Why it matters:** Dedicated monitoring is expensive, generates adversarial relationships, and creates an evasion game (cheat when the guard isn't watching). Embedded detection is cheaper, harder to time, and creates positive incentives for community members to report violations (because the person who has the good spot has the strongest incentive to report the person illegally using it).

**Design patterns:**
- **Rotation systems:** Assign rights to desirable resource areas on rotation. The current assignee naturally notices unauthorized use and has a strong incentive to report it. No guards needed.
- **Team operations:** Require harvesting from forests or commons to be done in teams. Solo operations are detectable by their absence from the team. Mutual monitoring emerges from the work structure.
- **Information asymmetry exploitation:** Rules on what constitutes permissible use are calibrated so that violations are obvious to any participant but not easily hidden. Fish quantity is hard to observe; gear type is easy.
- **Market-embedded monitoring:** Buyers who receive resources from the commons have a commercial incentive to verify that sellers have legitimate extraction rights (otherwise their own purchase is at legal risk).

**Anti-pattern:** A fishery that relies on self-reported catch data without any independent verification mechanism. Systematic over-reporting is individually rational.

---

### Principle 5: Locally Designed and Adjusted Rules

**What:** The community using the resource should design and modify its own rules through a participatory process, rather than having rules imposed by a central authority.

**Why it matters:** Local users have three critical information advantages over external managers:
1. Knowledge of the resource's actual behavior, seasonal patterns, and vulnerabilities
2. Knowledge of the technology and which restrictions are practically feasible
3. Knowledge of the community norms, relationships, and which sanctions are credible

Top-down governance consistently fails because it lacks all three. Rules that look sensible in a capital city fail in the field because they assume the wrong detection mechanisms, impose unenforceable quantity limits, or apply uniform rules to resources that have highly variable local characteristics.

**Design question:** "Have the people who will live with these rules had genuine input into designing them? Do the rules reflect their knowledge?"

**Also covers:** The right of community members to modify rules over time as conditions change. Static governance fails in dynamic resource environments.

---

### Principle 6: Conflict Resolution Mechanisms

**What:** Low-cost, rapid dispute resolution must be available so that perceived violations can be addressed without escalating to punishment spirals or formal legal proceedings.

**Why it matters:** Not every apparent violation is intentional. Many require investigation before punishment. When conflict resolution is expensive or slow, aggrieved parties take unilateral action (escalation, retaliation), which triggers counter-retaliation — exactly the punishment spiral that destroys cooperation. The first response to apparent cheating should be inquiry and negotiation, not punishment.

**Mechanisms:**
- Community meeting or council to address specific disputes
- Designated mediator with community trust
- Investigation protocol before sanctions are imposed
- Provision for correcting genuine errors without punishment

**Connection to governance:** This principle is the commons analogue to the tit-for-tat problem with noisy environments. In the real world, apparent defection is often noise, not defection. Conflict resolution is the mechanism for distinguishing the two before punishment is applied.

---

### Principle 7: Recognition by External Authority

**What:** External governments must recognize the community's right to organize and enforce its own governance rules rather than overriding, undermining, or ignoring local governance structures.

**Why it matters:** Local governance cannot function if external legal authority can nullify its decisions. A fishing community that designs rules for its commons has no enforcement power if those rules can be overridden by a federal agency, violated without legal consequence, or ignored by outsiders who know local enforcement has no standing.

**Design question:** "Does the relevant government authority recognize this community's governance structure as legitimate and enforceable?"

**Common failure mode:** Federal governments replace community governance with top-down management, destroying local knowledge and community accountability structures, then suffer from the information problems Principle 5 predicts.

---

### Principle 8: Nested Governance for Large Systems

**What:** For large-scale resources that cross community boundaries, governance is organized in multiple nested layers. Local groups govern local issues; coordination bodies handle cross-boundary issues; regional or national bodies handle system-level coordination.

**Why it matters:** Purely local governance fails when the resource crosses local boundaries (migratory fish, river watersheds, transboundary aquifers). Purely central governance fails because it lacks local knowledge (see Principle 5). The solution is hierarchical governance: each level handles the scale of problem it can actually manage with the information it has.

**Example structure:**
- Local fishing association → governs inshore zone, specific species, seasonal patterns
- Regional fishing council → coordinates between local associations for shared species, migratory routes
- National/international body → manages international waters, highly migratory species (tuna, whales)

**Anti-pattern:** A single national agency trying to manage all levels without local governance structures → information failure at local level; a set of purely local associations with no coordination mechanism → defection at the cross-boundary level.

---

## Applying the Checklist

Score each principle as: Met / Partially Met / Not Met / Not Applicable

| Principle | Status | Gap | Proposed fix |
|---|---|---|---|
| 1. Defined boundaries | | | |
| 2. Rules match conditions | | | |
| 3. Graduated sanctions | | | |
| 4. Automatic detection | | | |
| 5. Locally designed | | | |
| 6. Conflict resolution | | | |
| 7. External recognition | | | |
| 8. Nested governance | | | |

Any "Not Met" is a specific gap that the governance design must address. Priority order: Principles 1-4 are typically the most critical for functional governance; Principles 5-8 determine sustainability over time.

FILE:references/resolution-menu.md
# Resolution Menu: Cooperation Mechanism Levels

Each level assumes the prerequisites of levels below it are insufficient or unavailable.
Select the lowest level that is feasible given the situation's prerequisites.

---

## Level 1: Self-Enforcing Repeated Play

**Prerequisite check:**
- [ ] Interest rate / discount rate is below the break-even threshold: r < (R−P)/(T−R)
- [ ] Interaction is expected to continue indefinitely (or at least without a known final period)
- [ ] Detection is feasible (defection can be observed within a reasonable timeframe)
- [ ] Players are identifiable (defection can be attributed to the correct party)

**Mechanism: Tit-for-tat (two variants)**

### Standard Tit-for-Tat
Rule: Cooperate in period 1. In every subsequent period, do what the other player did in the previous period.

Properties (Axelrod's four principles):
- **Clear:** The rule is simple. The opponent does not have to guess what you will do; the link between their behavior and your response is unambiguous.
- **Nice:** Never initiates defection. No first-mover defection advantage.
- **Provocable:** Defection is punished immediately — no free rides.
- **Forgiving:** After one cooperative move by the other party, cooperation is immediately restored. No grudges.

Fatal flaw in noisy environments: Any misperception or error produces an alternating defection spiral. Party A defects by mistake in round 11. Party B retaliates in round 12. Party A (who cooperated in round 11) retaliates against B's retaliation in round 13. Round 14: B retaliates again. The pattern alternates indefinitely until another error accidentally restores cooperation or both defect permanently. Real-world example: Hatfield-McCoy feud — no one remembers what started it, but retaliation is perpetual. Middle East cycles of reprisal follow the same structure.

**Do not use** standard tit-for-tat when:
- Observability is imperfect (outcomes observable but attribution uncertain)
- Communication between parties is limited
- There is a history of prior misperceptions
- The relationship has experienced recent turbulence

### Generous Tit-for-Tat (recommended for most real-world applications)

Rule: Cooperate in period 1. Tolerate up to 1 isolated defection. Punish if the other party defects twice consecutively. Resume cooperation when the other party cooperates twice consecutively.

Source: Cotton-top tamarin monkey experiments — stable cooperation emerged with this exact rule. Consistent with subsequent Axelrod tournament results when error/noise was introduced.

Why 2-consecutive threshold works:
- Single defection may be noise (error, misperception, emergency). Punishing it starts a spiral.
- Two consecutive defections are highly unlikely to both be noise. This is deliberate defection.
- The threshold is clear and simple enough to be mutually understood.

**Operational implementation steps:**
1. Define "cooperate" and "defect" in specific, observable, operational terms (e.g., "cooperate" = maintain price at or above $X; "defect" = undercut by more than $Y)
2. Set monitoring interval (how frequently are outcomes observed?)
3. Establish response lag (how quickly can punishment be implemented?)
4. Communicate the rule explicitly (or create structural clarity through most-favored-customer clauses, price-matching policies, etc.)
5. Document forgiveness rule in advance — both parties must know cooperation will be restored

**Anti-patterns:**
- Defining defection too loosely (ambiguous trigger → false positives → unnecessary spirals)
- Setting punishment too harshly (catastrophic punishment → not credible when errors occur)
- Using tit-for-tat in a market with hidden price cuts (undetectable defection makes the strategy impossible)

---

## Level 2: Mutual Promises with Escrow or Simultaneous Commitments

**When Level 1 fails because:** Discount rate is acceptable but trust has been depleted by prior defection; parties believe the other will defect; no mechanism to make future cooperation credible.

**Mechanism:**
Both parties simultaneously commit to cooperative action AND deposit a performance bond (or penalty payment) in a neutral escrow account. If either party defects, the escrow funds are paid to the other party or forfeited to a third party.

**Design elements:**
- Escrow holder: neutral third party with no stake in the outcome (law firm, bank, arbitration service)
- Trigger definition: what counts as defection must be clearly specified and verifiable by the escrow holder
- Bond size: must exceed the one-period defection gain (T−R); otherwise defection plus forfeiting the bond is still profitable
- Duration: escrow is maintained for the agreed cooperation period; released upon successful completion
- Amendment process: how do the parties modify the agreement if circumstances change?

**Limitation:** Requires explicit agreement on what constitutes defection. Works for discrete, observable commitments. Does not work for tacit cooperation arrangements where defection is continuous and ambiguous.

**Example:** Camp David Accords — US provided explicit economic rewards (external escrow, essentially) to both Egypt and Israel conditional on maintaining the peace agreement. The third party with an interest in cooperation funded the escrow.

---

## Level 3: Reputation and Linkage

### Reputation Mechanism

**When to use:** The direct bilateral relationship has insufficient cooperation value, but the parties care about their reputation with third parties (customers, future partners, lenders, regulators, employees).

**How it works:** Defection in the current relationship damages the party's reputation with observers who will matter in future relationships. This creates an indirect future cost beyond the bilateral punishment.

**Reputation effectiveness conditions:**
- Defection must be observable to relevant third parties (not just the bilateral partner)
- Defection must be clearly attributable (cannot hide behind noise or ambiguity)
- The reputational audience must be substantial — the parties must have future interactions with people who will care
- Reputation recovery is slow — the damaged party must continue in the market long enough for reputation to have positive value

**Design question:** "Who observes this interaction besides the two parties? Do those observers affect the defector's future?" If the answer is no one significant, reputation alone will not sustain cooperation.

### Linkage Mechanism

**When to use:** Multiple dimensions of interaction between the same parties; defection in one dimension can be punished across multiple dimensions; the total cooperation surplus across all linked dimensions exceeds the temptation to defect in any one.

**How it works:** Bundle multiple interactions into a single relationship. "If you defect on pricing, we will also defect on delivery terms, quality standards, and the three other dimensions of our relationship."

**Critical limitation:** If all dimensions have identical payoff structures (same T, R, P, S), then linking them scales both the cooperation gain AND the defection gain proportionally, leaving the break-even interest rate unchanged. The ratio (R−P)/(T−R) is the same for the bundle as for any individual dimension.

Linkage only helps when dimensions are asymmetric — some dimensions have low defection temptation (high R−P relative to T−R) while others have high temptation. Bundling transfers sustainability from the low-temptation dimension to the high-temptation one.

**Multiproduct interaction warning (from Ch. 3):** Cheating on a multiproduct relationship brings the prospect of multiproduct retaliation, but also the temptation of multiproduct cheating simultaneously. These forces cancel if payoff structures are identical across products. Do not assume multiproduct relationships automatically sustain cooperation better than single-product relationships.

---

## Level 4: Third-Party Intervention and External Enforcement

### Explicit Contract Enforcement

**When to use:** Self-enforcement is not viable; parties cannot write credible unilateral punishment threats.

**Requirements:**
- Clear, verifiable specification of what cooperation and defection mean (cannot leave this ambiguous)
- Enforcement authority with jurisdiction and willingness to act
- Evidence mechanism: how will defection be proven?
- Proportionate remedy: what is the contracted consequence for defection?

**Limitation:** Many cooperation arrangements cannot be fully specified in a contract. "Do not undercut on price" is straightforward. "Maintain service quality at the level expected by our customers" is not. Contracts work for discrete, observable, attributable defections.

**Antitrust risk for firms:** Explicit agreements among competitors to cooperate on pricing or output are illegal in most jurisdictions. External enforcement through contract is not available for cartel arrangements — this is precisely why tacit self-enforcement mechanisms (Levels 1-3) are commercially important.

### Third-Party Mediator or Arbitrator

**When to use:** Parties cannot credibly commit to punishments themselves; a neutral third party has both authority and interest in the cooperative outcome.

**What the mediator provides:**
1. Focal point: the mediator defines what cooperation means (resolves the "clarity" prerequisite gap)
2. Verification: the mediator monitors compliance (resolves the "detection" gap)
3. Punishment: the mediator imposes consequences that neither party could credibly threaten unilaterally

**Mediator requirement:** The mediator must have either direct authority to impose consequences OR enough leverage over both parties (economic rewards, reputational influence, legal authority) that their judgment matters.

**Example:** US role at Camp David — neither Egypt nor Israel could credibly reward each other's cooperation. The US, with economic and military assistance to offer both, could.

### Regulatory Prohibition of the Defect Option

**When to use:** The defection option itself can be made illegal or structurally unavailable, eliminating the dominant strategy.

**How it works:** Rather than punishing defection after the fact, the regulation removes the defection choice from the game. Cigarette advertising ban (1968): TV advertising was a prisoners' dilemma for cigarette companies — each had to advertise because others did, even though collective advertising budgets were pure waste. The ban eliminated the defection option and actually benefited the firms by removing the arms race.

**Considerations:**
- Regulatory prohibition that benefits the regulated parties (the firms) may face regulatory capture concerns
- Prohibition must be designed to prevent new forms of defection from emerging (if TV advertising is banned, does billboard advertising become the new defection option?)
- Prohibition works best when the defection option has a clear legal definition

---

## Level 5: Ostrom Commons Governance

See [ostrom-commons-governance.md](ostrom-commons-governance.md) for full treatment.

**When to use:** Multiperson dilemma involving a shared resource; group is identifiable and can be organized; resource is local enough for community governance to be credible.

**Summary of when NOT to use Level 5 alone:**
- Completely anonymous large group (no community relationships)
- Resource spans jurisdictions in ways that prevent coherent local governance
- External authority will not recognize community governance rights (Principle 7 gap)
- Group is too heterogeneous for shared norms to develop

In these cases, Level 4 (external enforcement) may need to be the primary mechanism, with Level 5 principles used where possible to reduce enforcement costs.

---

## Mechanism Selection Matrix

| Situation | Recommended Level | Key reason |
|---|---|---|
| Ongoing bilateral relationship, observable, discount rate low | 1 (generous tit-for-tat) | Self-enforcing, no external requirement |
| Bilateral, trust depleted by prior defection, otherwise viable | 2 (escrow/promises) | Restart credibility |
| Bilateral, low direct cooperation value, strong reputational stakes | 3 (reputation/linkage) | Indirect enforcement |
| One-shot or short relationship, observable, specifiable defection | 4 (contract) | No shadow of future |
| Competitor coordination (antitrust-constrained) | 1 or 3 only (tacit) | Explicit agreements illegal |
| Shared resource, identifiable community, local resource | 5 (Ostrom) | Scale and community fit |
| Large anonymous group, no community structure | 4 (regulation) | Cannot self-govern |

---
name: negotiation-strategist
description: "Apply the complete game-theoretic bargaining framework to any negotiation. Use this skill when a user needs to structure a negotiation, determine who has leverage, calculate the fair split, or decide whether to make a concession or walk away. Triggers include: user is preparing for a salary negotiation, contract renegotiation, partnership deal, M&A term sheet, or labor negotiation and wants to know what number to open with and why; user wants to determine the 'pie' — the true surplus that is actually at stake between the two parties, not the headline dollar figures; user needs to identify and quantify their Best Alternative to a Negotiated Agreement (BATNA) or the other side's BATNA before entering talks; user wants to know how to improve their bargaining position before the negotiation starts (raise your BATNA, lower theirs); user must decide whether to bundle multiple issues together or separate them; user is weighing whether to actually strike, walk out, or threaten to do so, and wants to understand the cost-benefit calculation; user wants to propose a virtual-strike or escrowed-revenue arrangement to eliminate collateral damage while preserving negotiating pressure; user is in an alternating-offer negotiation and wants to calculate the equilibrium split given relative patience levels; user suspects they are negotiating over the wrong number (confusing total value with incremental value above no-deal); user faces brinkmanship — escalating risk of breakdown — and wants to calibrate how far to push. This skill does NOT cover simultaneous-move games (use nash-equilibrium-analyzer), one-shot ultimatum games without iteration, or multi-party coalition bargaining beyond two principal parties."
version: 1.0.0
homepage: https://github.com/bookforge-ai/bookforge-skills/tree/main/books/the-art-of-strategy/skills/negotiation-strategist
metadata: {"openclaw":{"emoji":"📚","homepage":"https://github.com/bookforge-ai/bookforge-skills"}}
status: draft
source-books:
  - id: the-art-of-strategy
    title: "The Art of Strategy"
    authors: ["Avinash K. Dixit", "Barry J. Nalebuff"]
    chapters: [11]
tags: [game-theory, negotiation, bargaining, BATNA, deal-structuring]
depends-on: [backward-reasoning-game-solver]
execution:
  tier: 1
  mode: full
  inputs:
    - type: document
      description: "Description of the negotiation situation: the parties, what is being negotiated, what each side can do if no deal is reached (their outside options or BATNAs), the timeline, any multi-issue dimensions, and the current state of talks"
  tools-required: [Read, Write]
  tools-optional: []
  mcps-required: []
  environment: "Any agent environment; user describes the negotiation context in text; agent produces a structured strategy with explicit numbers, leverage analysis, and recommended moves"
discovery:
  goal: "Measure the pie correctly, identify each party's BATNA, calculate the equilibrium split, diagnose the leverage imbalance, and produce a concrete negotiation strategy with specific opening positions, concession logic, and walk-away thresholds"
  tasks:
    - "Identify both parties and what they are negotiating over"
    - "Establish each party's BATNA: what each gets with zero agreement"
    - "Calculate the pie: total agreement value minus sum of BATNAs"
    - "Apply the equal-split principle: each party gets BATNA plus half the pie"
    - "Assess impatience asymmetry using the delta formula to adjust the split"
    - "Diagnose BATNA improvement opportunities: raise yours, lower theirs"
    - "Evaluate whether to bundle or unbundle multiple issues"
    - "Assess brinkmanship and strike dynamics if talks have or may break down"
    - "Evaluate virtual-strike option if a real disruption would cause collateral damage"
    - "Deliver: pie calculation, each party's fair-split number, BATNA improvement moves, recommended opening position, and walk-away threshold"
  audience: "Business negotiators, HR and labor relations teams, corporate development professionals, lawyers structuring deals, partnership managers, procurement teams, and anyone entering a bilateral negotiation where the stakes are explicit and the goal is an optimal, durable agreement"
  when_to_use:
    - "User needs to calculate what number to open with in a salary or contract negotiation"
    - "User suspects they are being asked to split total value instead of incremental value — the pie-measurement trap"
    - "User wants to know how their BATNA or the other side's BATNA affects who gets what"
    - "User must decide whether to take actions before negotiating that will change the BATNAs"
    - "User is structuring a multi-issue deal and wants to know whether to bundle or separate issues"
    - "User is in or approaching a strike, lockout, or deal-collapse situation and wants a clear cost-benefit view"
    - "User wants to propose a virtual-strike mechanism to preserve negotiating credibility without collateral harm"
---

# Negotiation Strategist

## When to Use

Use this skill any time two parties are negotiating over value that only exists if they reach agreement. The framework applies whether the context is labor-management, commercial contracts, partnership terms, M&A deal points, or international trade negotiations.

The two foundational questions this skill answers:

1. **How much value is actually at stake?** (The pie — not the headline number)
2. **Who gets how much, and why?** (BATNA-based equal split, adjusted for patience and leverage)

## Core Framework

### Step 1: Establish the BATNAs

Before calculating anything, identify what each party gets with no deal.

**BATNA = Best Alternative to a Negotiated Agreement.** It is the best outcome each side can secure on its own without the other party's cooperation.

- Ask: "If talks break down completely, what does each side do and what does that earn them?"
- BATNAs are not aspirations. They are concrete fallback outcomes. If a union can earn $300/day in outside work during a strike, that is their BATNA. If management can operate with replacement workers at $500/day profit, that is theirs.
- BATNAs can be improved strategically (see Step 5).

### Step 2: Calculate the Pie

**The pie is not the total value of the agreement. It is the additional value created by agreement above what both parties would get anyway.**

Formula:
```
Pie = (Agreement Value) - (Party A BATNA) - (Party B BATNA)
```

Example: Hotel earns $1,000/day when open. Union BATNA = $300/day (outside work). Management BATNA = $500/day (scab operation).
```
Pie = $1,000 - $300 - $500 = $200/day
```

The parties are not negotiating over $1,000. They are negotiating over $200. This matters enormously — anchoring on the wrong number leads to systematically wrong expectations.

**The Talmud principle traces this insight to ancient fairness norms:** when two parties dispute a garment each claims, the portion each concedes to the other is split evenly. The logic is exactly the BATNA-based pie calculation applied to physical division of cloth.

### Step 3: Calculate the Equal Split

With equal patience and equal bargaining position:

```
Party A receives: A's BATNA + Pie/2
Party B receives: B's BATNA + Pie/2
```

From the hotel example:
```
Union receives: $300 + $100 = $400/day
Management receives: $500 + $100 = $600/day
```

The equal split of the pie is the baseline. It is "equal" not in the sense of equal total receipts, but in the sense that each party gains the same increment above their no-deal fallback.

**Why this is the right number:** Both parties contribute equally to the existence of the pie. Neither can claim a larger share of the surplus simply because their outside option happens to be higher. The BATNA is already theirs — it is not a bargaining chip to be divided.

### Step 4: Adjust for Impatience (Rubinstein Equilibrium)

When the two parties alternate making offers and delays are costly, impatience breaks the 50/50 split of the pie in favor of the more patient party.

**Impatience factor δ (delta):** The fraction of value that remains after one round of delay. If a dollar next week is worth $0.99 today, then δ = 0.99 (patient). If a dollar next week is worth $0.33 today, then δ = 1/3 (very impatient).

**Rubinstein equilibrium split of the pie:**
```
Proposer's share of pie = 1 / (1 + δ)
Responder's share of pie = δ / (1 + δ)
```

Key cases:
- δ = 1 (no cost to waiting): split is 1/2 and 1/2. Pure 50/50.
- δ = 1/2 (each delay loses half the pie): proposer gets 2/3, responder gets 1/3.
- δ → 0 (extreme impatience, like ultimatum game): proposer gets essentially everything.

**Backward induction logic (from the backward-reasoning-game-solver framework):** The proposer's advantage arises because each round the proposer can claim the portion of pie that would be lost if the responder said no. The responder only gets the δ-discounted value of their next turn. Working backward from the terminal condition produces the δ/(1+δ) formula for the responder's share.

**Practical implication:** If your counterpart is under more time pressure than you are (public media coverage, quarterly earnings, expiring option, cash crunch), you are effectively the more patient party and should capture more than 50% of the pie. Conversely, if your organization faces political pressure to settle quickly, expect to concede more.

### Step 5: Improve the BATNAs — "This Will Hurt You More Than It Hurts Me"

BATNAs are often not fixed. Before or during negotiations, both parties can take actions that shift the BATNA landscape.

**General rule:** You will do better in the negotiation if your BATNA improves relative to your counterpart's BATNA — even if both BATNAs get worse in absolute terms, as long as theirs gets worse by more.

**Calculation:** If an action costs you X but costs the other party Y, and Y > X, the action is worth taking even though it hurts you, because it improves your relative bargaining position by (Y - X)/2 after the pie is re-split.

Example:
- Baseline: Union BATNA $300, Management BATNA $500. Pie = $200. Union share = $400.
- Union intensifies picketing: costs union $100/day in outside income, reduces management's scab profit by $200/day.
- New BATNAs: Union = $200, Management = $300. Pie = $500. New union share = $200 + $250 = $450.
- Net gain to union: +$50/day despite the self-imposed cost.

**MLB 1980 case study:** Players struck during the exhibition season (players received no salary, but owners collected gate revenue from vacationers). Players returned for the regular season but threatened another strike on Memorial Day weekend — when owner revenues spike sharply. The players had no salary at stake during the exhibition season, so their cost was low. Owner revenue loss was highest precisely when the strike threat loomed. The players identified the timing that maximized the asymmetry between their cost and the owners' cost.

**BATNA improvement tactics:**
- **Raise your BATNA:** Develop credible outside alternatives before negotiating. Competing offers, alternative suppliers, internal capability development, coalition formation.
- **Lower their BATNA:** Actions that reduce the other side's ability to walk away or operate without you. Intensified competition for their customers, public commitments that make their fallback position more visible or more costly.
- **Both simultaneously:** If both BATNAs drop but theirs drops more, you gain.

### Step 6: Multi-Issue Bundling

When multiple issues are on the table, the choice to bundle or unbundle them strategically affects outcomes.

**Bundle when:**
- You value different issues differently than your counterpart does.
- Bundling allows a package trade where you concede on issues you care less about in exchange for wins on issues you care more about.
- Example: A company values group health coverage at $1,000/worker while an individual worker would pay $2,000 for the same coverage. The company can offer health coverage instead of an equivalent wage increase — both parties prefer the bundle.
- Broad negotiations (like GATT/WTO trade rounds) succeed more than narrow sector-by-sector talks for exactly this reason: the larger the bundled package, the more room for issue-swapping.

**Unbundle when:**
- Your counterpart is trying to use strength on one issue to extract concessions on an issue where you are strong.
- Bundling a security alliance with a trade dispute allows one party to threaten the security arrangement to extract economic concessions. The weaker party on one dimension should insist on separating the games.
- Example: Japan insisted the U.S.-Japan military alliance and trade disputes be negotiated separately to prevent the U.S. from leveraging security threats to extract trade concessions.

### Step 7: Brinkmanship and Strikes

Strikes and breakdowns happen even when both parties would prefer agreement, because:

1. **Asymmetric information:** Each side must guess the other's cost of waiting. Since a lower cost of waiting is advantageous, each side has incentive to claim its costs are low. Claims without proof are not credible. The only credible proof is actually incurring the cost.

2. **Signaling through pain:** A strike is a costly signal. By actually striking, the union demonstrates that its cost of striking is lower than management believed. The signal is credible precisely because it hurts.

3. **Brinkmanship mechanics:** Rather than an all-or-nothing strike threat (not credible when much time remains), the effective form is gradual escalation — tempers rising, talks souring, increasing probability of breakdown each day. The party that fears breakdown less has the stronger position. Brinkmanship is a weapon for the stronger party.

**When strikes occur despite the theory predicting they should not:** Both sides must have a common view of the eventual outcome for agreement to happen immediately. Disagreement about who will concede — caused by private information or genuine strategic ambiguity — causes both sides to hold out, incurring real costs. The strike ends when one side's resolve is tested enough that the other side's belief updates.

**Implication for preparation:** Before talks begin, invest in understanding the other side's true cost of delay — their cash position, external pressures, political constraints. The better your model of their impatience, the less likely you are to miscalibrate and accidentally trigger a breakdown.

### Step 8: Virtual Strikes

When a real strike or lockout would cause large collateral damage — to customers, third parties, public reputation, or the long-term enterprise — consider proposing a virtual strike arrangement.

**Mechanism:**
- Workers continue working as normal.
- The employer continues producing as normal.
- During the virtual strike period, neither side gets paid: workers forfeit wages; employer forfeits all revenue (paid to a third party — government, charity, or customers as free product).
- The BATNAs are unchanged: both sides feel the same financial pain as in a real strike.
- No third parties are harmed.

**Why it works:** The bargaining-theoretic logic of a strike is purely about pain imposition — demonstrating that your cost of not having a deal is lower than the other side believes. A virtual strike replicates this exactly without the collateral harm of service disruption, consumer inconvenience, or reputational damage.

**Historical precedent:** WWII-era Jenkins valve plant (Bridgeport, CT); 1960 Miami bus strike (customers rode free); 1999 Meridiana Airlines pilot strike (Italy's first virtual strike — flights operated, Meridiana donated all ticket revenue to charities). In all cases management forfeited gross revenue, not just profits, because profit measurement is too easy to manipulate.

**When to propose it:** Propose before the real strike becomes imminent — ideally as a contingency clause in the contract: "If negotiations fail at the next renewal, the default dispute mechanism is a virtual strike." Agreeing in advance avoids the game-theoretic problem of appearing weak by proposing it in the heat of a breakdown.

**Limitation:** Public relations benefit of virtual strikes may paradoxically make them harder to implement — some employers prefer the reputational damage of a real strike over the reputational windfall a virtual strike gives workers.

## Worked Example: The Triangle Airfare Negotiation

Two companies (Houston and San Francisco) share a New York lawyer. The lawyer flies a triangle route NY-Houston-SF-NY ($2,818) instead of two round trips ($3,818). The savings = $1,000.

**Wrong approaches:**
- Split the triangle fare 50/50 → Houston pays $1,409. But Houston's standalone round-trip is $1,332. Houston would refuse.
- Allocate by leg or mileage → leads to ad hoc results that depend on route geometry, not fairness.

**Right approach — measure the pie:**
- Houston BATNA: $1,332 (its own round-trip).
- SF BATNA: $2,486 (its own round-trip).
- Total BATNA sum: $3,818.
- Agreement value: $2,818.
- Pie = $3,818 - $2,818 = $1,000.
- Equal split: each saves $500 above their BATNA.
- Houston pays: $1,332 - $500 = **$832**
- SF pays: $2,486 - $500 = **$1,986**

This is the uniquely fair and stable outcome: each party gets an equal share of the value they jointly created by cooperating.

## Quick Reference

| Concept | Formula |
|---|---|
| Pie | Agreement value - Party A BATNA - Party B BATNA |
| Equal split (Party A) | A BATNA + Pie/2 |
| Equal split (Party B) | B BATNA + Pie/2 |
| Proposer share (Rubinstein) | 1 / (1 + δ) |
| Responder share (Rubinstein) | δ / (1 + δ) |
| Patient limit (δ → 1) | 50/50 |
| Ultimatum limit (δ → 0) | 100/0 |

## Structural Self-Check

Before finalizing any negotiation strategy, verify:

- [ ] Have I identified each party's BATNA concretely, not aspirationally?
- [ ] Have I calculated the pie as the increment above BATNAs, not the total headline figure?
- [ ] Have I accounted for asymmetric patience and used the δ/(1+δ) formula if one party is more impatient?
- [ ] Have I considered pre-negotiation BATNA improvement moves (raise mine, lower theirs)?
- [ ] Have I identified all issues on the table and decided whether bundling benefits me?
- [ ] If a strike or breakdown is possible, have I calculated who bears more relative cost?
- [ ] If collateral damage is large, have I considered proposing a virtual strike mechanism?
- [ ] Is my opening position anchored to the correct number (my BATNA + half the pie), not a naive split of total value?

## References

- `references/rubinstein-bargaining-math.md` — Full derivation of the δ/(1+δ) split from backward induction, with worked numerical examples for δ = 0.99, 0.5, and 1/3
- `references/batna-improvement-case-studies.md` — MLB 1980 exhibition season strike, hotel union/management 101-day season, detailed numbers
- `references/virtual-strike-mechanics.md` — Meridiana 1999 case, Jenkins valve plant, Miami bus strike, proposal language for contingency clauses
- `references/multi-issue-bundling-guide.md` — GATT/WTO bundling logic, Japan-US security/trade separation, health benefits vs. wages example

## License

This skill is licensed under [CC-BY-SA-4.0](https://creativecommons.org/licenses/by-sa/4.0/).
Source: [BookForge](https://github.com/bookforge-ai/bookforge-skills) — The Art of Strategy by Avinash K. Dixit, Barry J. Nalebuff.

## Related BookForge Skills

Install related skills from ClawhHub:
- `clawhub install bookforge-backward-reasoning-game-solver`

Or install the full book set from GitHub: [bookforge-skills](https://github.com/bookforge-ai/bookforge-skills)

FILE:references/batna-improvement-case-studies.md
# BATNA Improvement: Case Studies

Source: The Art of Strategy, Ch. 11 (pp. 312–317)

## Principle

Your BATNA and your counterpart's BATNA determine the pie and each party's share of it. Since share = BATNA + Pie/2, anything that raises your BATNA or lowers theirs improves your outcome. Even moves that hurt both parties are worthwhile if they hurt the other party more (asymmetric damage).

## Case Study 1: Hotel Union/Management — 101-Day Season

### Setup
- Hotel season: 101 days. Hotel earns $1,000/day operating.
- Baseline: Union BATNA = $0 (no outside income during negotiations). Management BATNA = $0 (no operation during strike).
- Pie = $1,000 - $0 - $0 = $1,000/day.
- Equal split: each gets $500/day.

### Introducing Outside Income (Union BATNA)
- Union members can earn $300/day in outside activities during negotiations.
- Union BATNA = $300/day. Management BATNA = $0.
- Pie = $1,000 - $300 - $0 = $700/day.
- Union gets: $300 + $350 = $650/day. Management gets: $350/day.

### Introducing Scab Operation (Management BATNA)
- Management can operate with replacement workers at $500/day profit (less efficient, some guest reluctance to cross picket lines).
- If union BATNA = $0 and management BATNA = $500:
- Pie = $1,000 - $0 - $500 = $500/day.
- Management gets: $500 + $250 = $750/day. Union gets: $250/day.

### Combined BATNAs
- Union BATNA = $300/day (outside work). Management BATNA = $500/day (scab operation).
- Pie = $1,000 - $300 - $500 = $200/day.
- Union gets: $300 + $100 = $400/day. Management gets: $500 + $100 = $600/day.

### BATNA Improvement Move: Union Intensifies Picketing
- Union foregoes $100/day of outside income to intensify picketing.
- Picketing reduces management's scab profit by $200/day.
- New BATNAs: Union = $200/day. Management = $300/day.
- New pie = $1,000 - $200 - $300 = $500/day.
- Union gets: $200 + $250 = $450/day. Management gets: $300 + $250 = $550/day.
- Union's net gain: $450 - $400 = +$50/day, despite the $100/day cost. The asymmetry ($200 damage to management vs. $100 cost to union) creates a $50 net gain.

**Lesson:** When evaluating a costly move before or during negotiations, compute (opponent's BATNA damage - your cost) / 2. If positive, the move improves your bargaining position even if it hurts you in absolute terms.

## Case Study 2: MLB 1980 — Exhibition Season Strike Timing

### Structure
- Players negotiated during the exhibition season (March–April), then returned for the regular season.
- Threatened a second strike starting Memorial Day weekend.

### BATNA Asymmetry by Timing

**Exhibition season:**
- Players' cost: no salary during exhibition season (players not paid for exhibition games).
- Owners' cost: owners earned revenue from vacationers and local fans at exhibition games.
- Player cost = 0. Owner cost = positive.
- Asymmetry: owners lose more than players during exhibition season → players hold leverage.

**Memorial Day weekend:**
- Owners' revenue from gate and television rises sharply starting Memorial Day.
- Players' salary is flat throughout the season.
- Striking at Memorial Day maximizes owner loss relative to player cost.

### Outcome
- Owners conceded just before the second threatened strike.
- The first strike (exhibition season) actually occurred — owners called the bluff.
- Key insight: the players chose the timing to maximize relative damage. They did not strike when their own costs were highest; they struck when the owner's costs were highest relative to theirs.

### Lesson for Practice
When choosing the timing of a threat or action, identify the calendar points where the ratio of (cost to opponent) / (cost to yourself) is maximized. Seasonal businesses, quarterly reporting cycles, elections, product launches, and contract renewals all create timing asymmetries.

## General Framework for BATNA Improvement Analysis

Before entering any negotiation, build a two-column table:

| Action | Cost to You | Cost to Opponent | Net Gain to You |
|--------|------------|-----------------|-----------------|
| Develop outside offer | Low (time) | Moderate (competitive pressure) | (Opponent cost - Your cost)/2 |
| Intensify competition for their clients | Low | High | High/2 |
| Public commitment to walk away | Medium (reputational) | High (credibility signal) | Positive |
| Coalition with others against them | Low | High | High/2 |

Select actions where the third column is positive and significant.

## BATNA Improvement vs. BATNA Bluffing

**Improving** your BATNA means actually creating better outside alternatives (competing offers, internal capability, coalition partners). This is durable and credible.

**Claiming** a higher BATNA than you have is not credible — it will be tested. The other side will often call a bluff precisely because they know each side has incentive to exaggerate its outside options. The only credible proof of a low cost of breakdown is to actually incur it (the signaling role of strikes).

Invest in genuine BATNA improvement before negotiations begin. Do not rely on claims that cannot be independently verified.

FILE:references/multi-issue-bundling-guide.md
# Multi-Issue Bundling in Negotiation

Source: The Art of Strategy, Ch. 11 (pp. 320–321)

## Core Insight

When multiple issues are on the table, the decision to bundle them together or negotiate them separately is itself a strategic choice with large consequences. The right choice depends on whether differential valuations create room for mutually beneficial trades.

## When to Bundle

**Bundle multiple issues when the parties value them differently.**

If Party A values Issue X highly and cares less about Issue Y, while Party B values Issue Y highly and cares less about Issue X, then bundling the two issues into one negotiation creates room for a trade where both parties end up better off than they would by negotiating each issue separately.

### Example: Health Benefits vs. Wages

- A company can provide group health coverage for $1,000/worker/year.
- An individual worker would pay $2,000 for the same coverage on the open market.
- The "exchange rate" is 2:1 — a dollar of health benefit is worth $2 of wages to the worker, but costs only $1 for the company to provide.

By bundling health benefits into the wage negotiation:
- The company offers health coverage instead of a $1,500 wage increase.
- The worker is better off (values the coverage at $2,000, which exceeds the $1,500 wage alternative).
- The company is better off (pays $1,000 instead of $1,500).
- Total surplus from the bundle is $500/worker greater than the wage-only negotiation.

**Rule:** When one party can provide something at lower cost than the other party can obtain it on the outside, bundling that item into the negotiation creates joint surplus that can be split.

### Example: GATT and WTO Trade Rounds

Broad, multi-country, multi-sector trade liberalization rounds (GATT/WTO) have consistently achieved more liberalization than narrow sector-by-sector or bilateral negotiations. The reason:

- Each country has different sensitivities and different export strengths.
- In a narrow negotiation, each country must give up something it protects in exchange for access to one specific market.
- In a broad round, Country A can concede on agriculture (where it is weak) in exchange for gains in manufactured goods (where it is strong), while Country B makes the opposite trade.
- The bundling of many issues creates more opportunities for issue-swapping where each party concedes on issues it cares less about.

**Rule:** Larger bundles create more opportunities for mutually beneficial trades among parties with heterogeneous valuations. Advocate for broad agendas in multi-party negotiations where you have uneven strengths.

## When to Unbundle

**Unbundle issues when your counterpart is trying to use leverage in one game to extract concessions in another.**

If Party B is strong on Issue X and uses the threat of deterioration on Issue X to force concessions on Issue Y (where Party A is strong), Party A should insist on separating the negotiations.

### Example: U.S.-Japan Security and Trade

- The United States had leverage in military/security arrangements with Japan.
- The U.S. could theoretically threaten to withdraw military protection unless Japan made trade concessions.
- But the U.S. has no real interest in withdrawing — it is purely a threat.
- Japan insisted these issues be negotiated separately to prevent the U.S. from using security threats as leverage in trade negotiations.

**Rule:** If your counterpart has strong positions on issues where you are vulnerable, separating those issues from the issues where you are strong prevents cross-contamination of leverage. Insist on single-issue negotiations when the bundled agenda would give the other side unfair leverage.

### When You Are the Weaker Party on Some Issues

- If you are weaker on Issue X but stronger on Issue Y, and your counterpart wants to bundle, consider the following:
  - Bundling will allow your counterpart to use their strength on X to dominate the entire package.
  - You may do better by isolating Issue Y (where you have strength) and negotiating it separately.
  - Push for separate tracks or sequential negotiations that start with issues where you have leverage.

## The Linkage Decision Matrix

| Situation | Bundle or Unbundle? | Reason |
|-----------|--------------------|----|
| You value issues differently from counterpart | Bundle | Enables mutually beneficial trades |
| Your counterpart is strong on one issue, you on another, and strength doesn't transfer | Bundle | Allows issue-swapping |
| Counterpart wants to use strength on Issue X to dominate Issue Y where you are strong | Unbundle | Prevents leverage transfer |
| You are uniformly weak across issues | Unbundle | Limits the scope of damage; isolate each battleground |
| Multi-party negotiation with heterogeneous valuations | Bundle | More parties = more trade opportunities |
| Two-party negotiation with one-sided leverage | Depends | Bundle if it creates joint surplus; unbundle if it transfers leverage to your disadvantage |

## Linkage as a Threat Mechanism

Bundling can also be used as a coercive threat rather than a cooperative opportunity. A party may offer to link two issues to create a new threat: "Unless you concede on X, we will withdraw cooperation on Y (where you need us)."

This is a strategic use of bundling as leverage, not a search for mutual gain. The target party's defense is to insist on separating the games — formally or publicly — so that the threat has no legal, contractual, or reputational basis.

In international relations, this manifests as "issue linkage" diplomacy. Countries with more issue areas to link (larger powers) generally benefit more from this strategy than smaller countries, which is why smaller countries often advocate for rules-based, single-issue negotiations in multilateral forums like the WTO.

FILE:references/rubinstein-bargaining-math.md
# Rubinstein Bargaining: Mathematical Derivation

Source: The Art of Strategy, Ch. 11 Appendix (pp. 326–330)

## Setup

Two parties alternate making proposals for how to divide a pie of size 1. The pie is described as (X, 1-X) where X is the proposer's share. As soon as one side accepts the other's proposal, the game ends. Delay is costly: a dollar received next period is worth δ times a dollar received today.

δ (delta) is the patience parameter:
- δ close to 1: parties are patient, delay is cheap
- δ close to 0: parties are very impatient, delay destroys most of the value
- δ = 0: ultimatum game (entire pie disappears if offer is not accepted immediately)

## Finding the Equilibrium

Let L = the minimum share you will ever accept (your "lowest acceptable").

**Step 1:** If you turn down today's offer and wait to make a counteroffer tomorrow, the other side knows you will never accept less than L. So the most they can ever hope for is 1 - L. But that is what they get in one period. What they get today is δ(1 - L).

Since the best you can offer them tomorrow is δ(1 - L), they will accept it. That means by rejecting today and countering tomorrow, you can secure 1 - δ(1 - L).

**Step 2:** You should therefore never accept less than 1 - δ(1 - L) today.

Setting L = 1 - δ(1 - L) and solving:

```
L = 1 - δ(1 - L)
L = 1 - δ + δL
L - δL = 1 - δ
L(1 - δ) = 1 - δ
```

Wait — that gives L = 1 regardless of δ, which is wrong. The correct setup is:

The minimum you accept must satisfy: L ≥ δ(1 - δ(1 - L))

Working through the algebra as shown in the text:

```
L ≥ δ(1 - L) [the other side accepts δ(1-L) tomorrow]

Setting L = δ(1 - δ(1 - L)):
L = δ - δ²(1 - L)
L = δ - δ² + δ²L
L - δ²L = δ - δ²
L(1 - δ²) = δ(1 - δ)
L = δ(1 - δ) / (1 - δ²)
L = δ(1 - δ) / [(1 - δ)(1 + δ)]
L = δ / (1 + δ)
```

So L = δ/(1 + δ). This is the least you will ever accept when you are the responder (i.e., the person receiving an offer).

By symmetry, this is also the most you will ever offer the other side (since they will use the same logic). Therefore:

**Proposer's share = 1 - L = 1 - δ/(1 + δ) = 1/(1 + δ)**

**Responder's share = δ/(1 + δ)**

## Numerical Examples

### δ = 0.99 (weekly offers, 1% weekly discount — very patient)

```
Proposer gets: 1 / (1 + 0.99) = 1 / 1.99 ≈ 0.503
Responder gets: 0.99 / 1.99 ≈ 0.497
Split: approximately 50.3 / 49.7
```

Interpretation: With very short intervals between offers, the proposer's first-mover advantage nearly vanishes. The split is almost exactly 50/50.

### δ = 0.5 (each delay loses half the pie — moderately impatient)

```
Proposer gets: 1 / (1 + 0.5) = 1 / 1.5 = 2/3
Responder gets: 0.5 / 1.5 = 1/3
Split: 67 / 33
```

Interpretation: The proposer claims the half that would disappear if the responder said no (since δ = 1/2, half is lost per round), plus half of the remainder. Each round, the proposer collects twice as much as the responder.

### δ = 1/3 (impatient — two-thirds of value lost each delay)

```
Proposer gets: 1 / (1 + 1/3) = 1 / (4/3) = 3/4
Responder gets: (1/3) / (4/3) = 1/4
Split: 75 / 25
```

### δ → 1 (limit case: perfectly patient)

```
Proposer gets: 1/(1+1) = 1/2
Responder gets: 1/(1+1) = 1/2
Split: 50/50
```

### δ → 0 (limit case: ultimatum game)

```
Proposer gets: 1/(1+0) = 1
Responder gets: 0/(1+0) = 0
Split: 100/0
```

This matches the ultimatum game result from Chapter 2: if the pie disappears entirely if the responder says no, the proposer can demand everything.

## Applying to Real BATNAs

The Rubinstein formula applies to the pie (surplus above BATNAs), not to total value. The full split is:

```
Party A (proposer) receives: A_BATNA + Pie × [1/(1+δ)]
Party B (responder) receives: B_BATNA + Pie × [δ/(1+δ)]
```

If both parties have the same δ and alternate making offers, use δ = 1 as the practical approximation (50/50 pie split) unless there is a clear impatience asymmetry.

## Differing Impatience

If the two parties have different discount factors δ_A and δ_B, and the time between offers shrinks to zero, the pie is split in the ratio of their waiting costs. If one party is twice as impatient as the other, it gets one-third of the pie (half as much as the more patient party).

Practical implication: institutional structures that force impatience (electoral cycles, quarterly reporting, public media pressure) systematically reduce bargaining power. The U.S. government's impatience in international negotiations is repeatedly exploited by more patient counterparts.

FILE:references/virtual-strike-mechanics.md
# Virtual Strike Mechanics

Source: The Art of Strategy, Ch. 11 (pp. 321–323)

## The Problem with Real Strikes

A traditional strike (or lockout) imposes costs on:
1. Workers (lost wages)
2. Management (lost profits)
3. Customers (lost service)
4. Third parties (supply chain disruption, economic spillover)

The dock-worker lockout of 2002: the dispute involved ~$20 million in productivity enhancements. The lockout disrupted the U.S. economy by more than $10 billion. Collateral damage was 500 times the size of the dispute.

**The strike's logic is purely signaling:** each party needs to prove its cost of not having a deal is lower than the other believes. The collateral damage to third parties is pure waste — it serves no bargaining purpose.

## The Virtual Strike Mechanism

**Mechanism:**
- Workers continue working as normal (no service disruption).
- Employer continues producing as normal (no lost output for customers).
- All revenue during the virtual strike period is forfeited by the employer to a third party: government, charity, or customers (as free product/service).
- Workers receive no wages during the virtual strike period.

**Result:**
- Workers feel the same financial pain as in a real strike (no pay).
- Employer feels the same financial pain as in a real strike (no revenue — in practice, gross revenue is used rather than profit because profit is too easy to manipulate).
- Customers are unharmed.
- The BATNAs are unchanged relative to a real strike.
- The bargaining incentives are unchanged.

## Historical Precedents

### Jenkins Company Valve Plant — Bridgeport, CT (World War II)
- The U.S. Navy used a virtual strike to settle a labor dispute.
- Workers continued working; management forfeited profits.
- First documented industrial use of the mechanism.

### Miami Bus Strike (1960)
- Bus drivers staged a virtual strike.
- Buses ran on their normal schedule.
- Passengers rode for free — the revenue that would normally go to the transit authority was effectively given to riders.
- "Customers got a free ride, literally."

### Meridiana Airlines — Italy (1999)
- Italy's first virtual strike in aviation.
- Pilots and flight attendants worked their normal flights, unpaid.
- Meridiana donated all ticket revenue from virtually-struck flights to charities.
- Flights were not disrupted. No stranded passengers.
- The virtual strike worked as predicted — the threat was credible and management settled.

### Italy Transport Union (2000)
- 300 pilots participated in a virtual strike.
- The union forfeited 100 million lire (strike payments to charity).
- The union chose a medical device for a children's hospital as the recipient.
- Public relations benefit: instead of destroying consumer goodwill (as in NHL 2004-5), the virtual strike generated a positive story.

### NHL Lockout (2004–5)
- Management imposed a lockout rather than accepting virtual strike terms.
- The entire season was cancelled. No Stanley Cup awarded.
- Arena attendance took years to recover.
- This is the counterfactual: real disruption permanently destroys consumer demand and enterprise value. Virtual strikes avoid this.

## Implementation Design

### Revenue vs. Profit Forfeiture

In all four documented cases, management agreed to forfeit gross revenue, not profit. The reason: profit is a residual after costs, and management can shift costs or accounting choices to minimize reported profit during the dispute period. Gross revenue is harder to manipulate.

Workers forfeit wages. Employer forfeits gross revenue. The escrowed funds go to the third party.

### Third-Party Recipient Options
- **Government (tax authority):** Neutral, accepted by both parties, legally clean.
- **Charity:** Creates positive public relations; may be preferred by labor as a signal of strength.
- **Customers (free product/service):** Most consumer-visible option; aligns customer interests with quick resolution.

### Timing of Agreement

The right time to agree to a virtual strike mechanism is **before the real strike becomes imminent** — ideally as a contingency clause in the labor contract:

> "If contract negotiations fail to reach agreement by [date], both parties agree to enter a virtual strike arrangement beginning [date+n], under which [revenue escrow terms], until agreement is reached."

Agreeing in advance avoids the game-theoretic problem where proposing a virtual strike during a real dispute signals weakness or loss of resolve.

## Limitations and Complications

1. **Public relations benefit may backfire:** If a virtual strike generates positive PR for workers (charity donations, customer goodwill), management may prefer the reputational damage of a real strike — it is a known threat rather than an unknown one.

2. **Customer inconvenience as a strategic weapon:** Some real strikes are deliberately designed to inconvenience customers so they pressure management to settle. A virtual strike removes this pressure. Workers who rely on customer pressure as part of their strategy may not want to go virtual.

3. **Revenue measurement disputes:** Even gross revenue can be contested in industries with complex pricing, barter, intercompany transactions, or subscription models. The escrow mechanism must specify exactly what counts.

4. **Worker motivation during virtual strike:** Workers who work for nothing during a virtual strike must believe settlement is imminent and that the sacrifice is worth it. If the dispute drags on, worker participation may erode.

## Decision Rule: When to Propose a Virtual Strike

Propose a virtual strike mechanism when:
- The collateral damage to customers or third parties is large relative to the size of the dispute.
- The enterprise's long-term customer relationships or brand value are at risk from disruption.
- Both parties genuinely want to settle but need a credible signaling mechanism.
- You are the party whose BATNAs are better protected under a virtual arrangement than a real one.

Do not propose a virtual strike when:
- Customer pressure is a key part of your negotiating strategy.
- Management will use the virtual strike's PR benefit to undercut your leverage.
- The dispute is so bitter that neither party wants to cooperate even on the mechanism.

---
name: nash-equilibrium-analyzer
description: Find Nash equilibria in simultaneous-move games by constructing payoff matrices, eliminating dominated strategies (Rules 2-3), mapping best responses (Rule 4), and calculating mixed strategy proportions using the indifference principle (Rule 5). Use this skill when two or more players choose actions simultaneously without seeing each other's moves — pricing decisions, product launches, competitive bids, penalty kicks, resource allocation conflicts — and you need to identify stable strategy configurations, calculate exact mixing proportions for zero-sum conflicts, or select among multiple equilibria using focal-point analysis.
version: 1.0.0
homepage: https://github.com/bookforge-ai/bookforge-skills/tree/main/books/the-art-of-strategy/skills/nash-equilibrium-analyzer
metadata: {"openclaw":{"emoji":"📚","homepage":"https://github.com/bookforge-ai/bookforge-skills"}}
status: draft
source-books:
  - id: the-art-of-strategy
    title: "The Art of Strategy: A Game Theorist's Guide to Success in Business and Life"
    authors: ["Avinash K. Dixit", "Barry J. Nalebuff"]
    chapters: [4, 5]
tags: [game-theory, decision-making, equilibrium-analysis, competitive-strategy, mixed-strategy]
depends-on: []
execution:
  tier: 1
  mode: full
  inputs:
    - type: document
      description: "Description of a strategic situation with simultaneous choices: players, available strategies, and payoffs (or enough information to estimate them)."
  tools-required: [Read, Write]
  tools-optional: []
  mcps-required: []
  environment: "Works from a strategic situation description. Output: payoff matrix, equilibrium analysis, and strategy recommendation."
discovery:
  goal: "Identify Nash equilibria — strategy combinations where no player can improve by changing course alone — and prescribe which to play, including exact mixing proportions when pure strategies leave you exploitable."
  tasks:
    - "Build or validate the payoff matrix for all strategy combinations"
    - "Check for dominant strategies and eliminate dominated ones (Rules 2-3)"
    - "Map best responses to find cells where both players are simultaneously best-responding (Rule 4)"
    - "Calculate mixed strategy proportions using the indifference principle when no pure equilibrium exists (Rule 5)"
    - "Apply focal-point analysis when multiple equilibria exist"
  audience: "strategists, managers, product managers, negotiators, business professionals facing competitive simultaneous-choice situations"
  when_to_use: "When players choose actions simultaneously without observing each other's moves and you need to predict stable outcomes or prescribe optimal play"
  environment: "Strategic situation description with identifiable players, strategies, and payoffs. The agent constructs the matrix; the user need not supply pre-built tables."
  quality: placeholder
---

# Nash Equilibrium Analyzer

## When to Use

A simultaneous-move game is one where each player chooses without observing the other's choice — there is no first mover, no look-forward-and-reason-backward structure. Both choose at the same moment (or in sealed envelopes, or with mutual concealment). Apply this skill when:

- Players are deciding simultaneously: pricing, product launch timing, bidding, competitive positioning
- You face a repeating conflict where being predictable makes you exploitable: penalty kicks, advertising schedules, audit strategies, military feints
- Multiple outcomes each look like equilibria and you need to select one
- You want to know whether randomizing (mixing) beats committing to one action

**What this skill does not cover:** Games with sequential moves (one player chooses, then the other responds) use backward induction instead. If players observe each other's choices before responding, use the sequential-game framework.

**The four-rule sequence:**
1. Build the payoff matrix
2. Find and use any dominant strategy (Rule 2)
3. Eliminate dominated and never-best-response strategies successively (Rule 3)
4. Search remaining cells for mutual best responses — Nash equilibrium (Rule 4)
5. If no pure-strategy equilibrium, compute mixing proportions using the indifference principle (Rule 5)

---

## Context and Input Gathering

### Required Information

Before building the matrix, gather:

- **Players:** Who are the decision-makers? List each one.
- **Strategies:** What choices does each player have? List all options.
- **Payoffs:** For each combination of choices, what does each player get? Payoffs can be profits, success rates, scores, or any numerical representation of outcomes.

If payoffs are not precisely known, estimate them directionally (High/Medium/Low) or ask the user to rank outcomes — ordinal payoffs often suffice to find equilibria.

### Observable Context

If a situation description is provided, look for:
- Statements about what each player "prefers" or "would choose" — these indicate payoff ranking
- Outcome descriptions that depend on both players' choices simultaneously — confirms simultaneous structure
- Any statement that one strategy "always beats" another regardless of what others do — signals a dominant strategy
- Situations where interests are exactly opposed ("every dollar I gain, you lose") — signals a zero-sum game requiring Rule 5

### Sufficiency Check

You can proceed when:
1. You can name every player
2. You can list every strategy option per player
3. You can estimate payoffs for each combination (even approximate rankings)

If payoff information is missing, ask: "What outcome does Player X get if they choose A while Player Y chooses B?" Repeat until the matrix is complete or estimates are sufficient.

---

## Execution

### Step 1 — Build the Payoff Matrix

Construct a table with rows for one player's strategies and columns for the other's. Each cell contains both players' payoffs.

**Convention:** Row player's payoff in the southwest corner of each cell; column player's payoff in the northeast corner. (In zero-sum games, show only the row player's payoff since the column player's is always the complement.)

**Why build it explicitly:** The matrix makes all 2n combinations visible at once. Without it, players reason about their options in sequence, miss interactions, and reach wrong conclusions. The matrix converts circular "what if they..." thinking into simultaneous inspection.

For games with many strategies (more than 4-5 per player), build the matrix in a spreadsheet and note that software can compute equilibria directly. The manual method below applies to small games; the concepts transfer to large ones.

**Example — Pricing game (2 firms, 2 prices):**

|  | Rival: Low | Rival: High |
|---|---|---|
| **You: Low** | 40, 40 | 60, 20 |
| **You: High** | 20, 60 | 80, 80 |

---

### Step 2 — Check for Dominant Strategies (Rule 2)

A **dominant strategy** is one that gives you a higher payoff than every other option you have, regardless of what your opponent does.

**How to check:** For each of your strategies, compare its payoff across every column (every opponent strategy). If one row always produces a higher payoff than all other rows, it is dominant.

**If a dominant strategy exists:** Play it. No further analysis is needed for that player. Also expect rational opponents to play their dominant strategies.

**Why dominance matters:** A dominant strategy is your best choice no matter what — it eliminates the need to guess what the other player will do. If both players have dominant strategies, the Nash equilibrium is simply the cell where both dominant strategies intersect.

**Anti-pattern — ignoring dominance:** Players who do not check for dominance first waste effort reasoning about contingencies that do not apply. Always check dominance before best-response mapping.

---

### Step 3 — Eliminate Dominated and Never-Best-Response Strategies (Rule 3)

When no strategy is globally dominant, look for strategies to eliminate:

**Type A — Dominated strategy:** Strategy A is dominated by strategy B if B gives a higher payoff than A regardless of what the opponent does. A dominated strategy will never be played by a rational player; eliminate it.

**Type B — Never-best-response strategy:** A strategy that is never the best response to any opponent strategy — even if it is not dominated. Eliminate it; no rational player will use a strategy that is never the best they can do.

**Successive elimination:** After eliminating a strategy, re-examine the reduced game. Strategies that were not dominated in the original game may become dominated once other strategies are removed. Repeat until no further elimination is possible.

**Why this helps:** Successive elimination reduces a large, complex matrix to a smaller, tractable one. In some games it narrows the outcome to a single cell — finding the Nash equilibrium without explicitly checking all mutual best responses.

**Example — 5x5 pricing game reduced to 3x3:**
If prices of $42 and $38 are never best responses to any rival price, eliminate them. In the remaining 3x3 game, a dominant strategy of $40 may emerge for both firms, identifying the Nash equilibrium directly.

**Stopping rule:** If successive elimination produces a unique outcome, that is the Nash equilibrium. If it produces a smaller game but not a unique outcome, proceed to Step 4.

---

### Step 4 — Map Best Responses and Find Nash Equilibria (Rule 4)

A **best response** is the strategy that maximizes your payoff given a specific belief about what your opponent will do.

**How to find best responses:**
1. Fix one player's strategy (e.g., assume Rival plays Low).
2. For the other player, find which strategy produces the highest payoff given that assumption. Mark it (bold, underline, or highlight).
3. Repeat for every possible fixed strategy of the opponent.
4. Swap roles and repeat for the other player.

**Identify Nash equilibria:** A Nash equilibrium is any cell where **both** players' payoffs are marked as best responses. In such a cell, each player is already doing the best they can given the other's choice — neither has any incentive to deviate unilaterally.

**Definition check:** A configuration of strategies is a Nash equilibrium if and only if:
- Each player is choosing a best response to what they believe the other players are doing, AND
- Those beliefs are correct (each player is actually doing what the other expects)

**Why Nash equilibrium is the right solution concept:** Any outcome that is not a Nash equilibrium has at least one player who could improve by switching. That player has an incentive to deviate, making the outcome unstable. Nash equilibrium is the resting point of rational strategic reasoning.

**Multiple equilibria:** Games often have more than one Nash equilibrium. When this occurs, proceed to Step 5 (if zero-sum, compute mixing) or Step 6 (coordination games, use focal-point analysis).

---

### Step 5 — Calculate Mixed Strategy Proportions (Rule 5)

When there is no Nash equilibrium in pure strategies — that is, when best-response arrows cycle (L beats R, R beats L, L beats R...) — the equilibrium exists in **mixed strategies**: deliberate randomization over pure strategies.

**Why mixing is necessary:** In zero-sum (pure conflict) games, any predictable pure strategy can be exploited. A penalty kicker who always shoots left will face a goalie who always covers left. Randomization removes exploitability by making the opponent indifferent between their choices.

**The indifference principle (Rule 5):** Choose your mixing proportions so that your opponent gets the same expected payoff from any of their pure strategies. When your opponent is indifferent, they cannot exploit you by choosing one strategy over another.

**Two-strategy formula:**

For a 2x2 game where row player mixes strategy A (probability p) and strategy B (probability 1-p):

Set the column player's expected payoff from their Left strategy equal to their expected payoff from their Right strategy:

```
payoff(column Left | row A) × p + payoff(column Left | row B) × (1-p)
= payoff(column Right | row A) × p + payoff(column Right | row B) × (1-p)
```

Solve for p. This is the row player's equilibrium mixing proportion.

**Worked example — Penalty kick (kicker payoffs, goalie minimizes):**

|  | Goalie: Left | Goalie: Right |
|---|---|---|
| **Kicker: Left** | 58 | 95 |
| **Kicker: Right** | 93 | 70 |

To find the kicker's equilibrium mix (proportion p = Left):
- Against goalie's Left: 58p + 93(1-p) = 93 - 35p
- Against goalie's Right: 95p + 70(1-p) = 70 + 25p
- Set equal: 93 - 35p = 70 + 25p → 23 = 60p → p = 23/60 ≈ 0.383

Kicker should kick Left 38.3% of the time, Right 61.7%.

To find the goalie's equilibrium mix (proportion y = Left):
- Against kicker's Left: 58y + 95(1-y) = 95 - 37y
- Against kicker's Right: 93y + 70(1-y) = 70 + 23y
- Set equal: 95 - 37y = 70 + 23y → 25 = 60y → y = 25/60 ≈ 0.417

Goalie should cover Left 41.7% of the time, Right 58.3%.

**Three-strategy system of equations:**

For a 3-strategy zero-sum game, let player mix with probabilities p (strategy 1), q (strategy 2), (1-p-q) (strategy 3). Compute the opponent's expected payoff for each of their three strategies. Set all three equal (indifference across all three). Solve the system of two equations (the third is redundant since probabilities sum to 1).

**Graphical method (minimax V-shape):** Plot the row player's mixture proportion (x) on the horizontal axis. For each column player pure strategy, draw a line showing the row player's payoff against that pure strategy. The upper envelope (the maximum of these lines) forms an inverted-V shape. The kicker's best mixture is at the apex — the proportion that maximizes the minimum payoff. This is the maximin point, equal to the minimax by von Neumann's theorem.

**Critical constraint — only for zero-sum games:** Rule 5 (mix to make opponent indifferent) applies to zero-sum or pure-conflict games. In coordination games (where interests overlap), mixing produces the worst expected outcome — players get trapped in miscoordinated outcomes more often than the pure strategy equilibria. Never randomize independently in coordination games.

**Why your mix is determined by opponent's payoffs:** In a non-zero-sum game with a mixed equilibrium, your equilibrium mixture is calculated to keep your opponent indifferent — so it depends on your opponent's payoffs, not your own. Counterintuitively, if your own payoffs change, your equilibrium mixture is unaffected; only your opponent's equilibrium mixture changes.

---

### Step 6 — Handle Multiple Equilibria with Focal-Point Analysis

When a game has multiple Nash equilibria (common in coordination games), Nash equilibrium theory alone does not select among them. Use focal-point analysis.

**Definition:** A focal point (Schelling point) is an equilibrium that is "obvious" to all players without communication — because of mathematical salience, cultural convention, symmetry, precedent, or shared experience.

**Focal-point sources — check in order:**
1. **Mathematical salience:** Is one equilibrium uniquely simple? (Equal split = focal. Round numbers = focal. First in a ranked list = focal.)
2. **Cultural or historical convention:** Do the players share a background that makes one option stand out? (Geographic division, industry norms, social customs.)
3. **Symmetry or fairness:** Does one equilibrium treat players symmetrically? Equal splits are focal partly because of their fairness appearance.
4. **Prior interaction:** Have these players coordinated before? The outcome of past play is often the focal point for future play.
5. **Communication:** If players can talk before choosing, use that channel — any agreement, even a vague one, can create a focal point.

**When no focal point exists:** If players lack shared context to converge expectations, equilibrium selection may fail. This is a genuine limitation — Nash equilibrium predicts what will happen only when beliefs converge. Multiple equilibria without a focal point is a coordination failure risk, not an analytic error.

**Conflict games with multiple equilibria (Battle of Sexes, Chicken):** When players have different preferences over the multiple equilibria (unlike pure coordination games), the equilibrium selection problem is harder. Options:
- Pre-commitment: one player credibly binds themselves to their preferred equilibrium (covered in commitment skills)
- Alternation over repeated plays: agree to rotate between equilibria
- Third-party coordination: use an external signal both players observe to coordinate

---

## Output

Deliver the following in writing:

1. **Payoff matrix** — formatted table showing all strategy combinations and payoffs
2. **Equilibrium identification** — which cells are Nash equilibria, with the best-response markings shown
3. **Analysis path taken** — which rules applied (dominance, elimination, best-response, mixing)
4. **Mixing proportions** (if applicable) — exact fractions/percentages with the indifference calculation shown
5. **Strategy recommendation** — which equilibrium to target and why, including focal-point reasoning if multiple equilibria exist
6. **Anti-patterns flagged** — if the situation is a coordination game, warn against uncoordinated mixing

---

## Key Principles

**Nash equilibrium is a self-consistent belief system:** Each player chooses the best action given correct beliefs about others. Any outcome where at least one player could improve by switching is unstable.

**Every finite game has at least one Nash equilibrium** — possibly in mixed strategies. The only games without equilibria are exotic theoretical constructs.

**Mixing purpose:** Mix to prevent exploitation, not to be unpredictable for its own sake. The correct mix is the one that makes your opponent indifferent — calculated from their payoffs, not your preferences.

**Do not mix in coordination games:** When interests overlap and there are multiple pure-strategy equilibria, mixing independently leads to miscoordination. Use focal points or explicit coordination instead.

**Improving your weakness changes the equilibrium mix:** In a zero-sum game, if you improve your performance in one option, your opponent uses that option less in their mix — and you use it less too. The value of improving a weakness is that you do not have to use it as often; the opponent can no longer exploit it as easily.

---

## Examples

### Example 1 — Pricing game (unique pure equilibrium via successive elimination)

**Situation:** Two retailers are setting prices simultaneously in a range of $38–$42.

**Analysis:**
1. Build 5×5 payoff matrix
2. $42 is dominated by $41 for both firms (higher profit regardless of rival's price); eliminate
3. $38 is dominated by $39; eliminate
4. In the resulting 3×3 game, $40 is dominant for both firms
5. **Nash equilibrium:** Both price at $40 (40,000 profit each)

**Insight:** The $40 equilibrium is less profitable than the collusion price ($80 = 72,000 each), but neither firm can unilaterally raise price without losing customers. The equilibrium is stable even if both wish for a different outcome.

### Example 2 — Penalty kick (no pure equilibrium, mixed strategy required)

**Situation:** Kicker vs. goalie, simultaneous choice of Left/Right. Payoffs (kicker success %):

|  | Goalie: Left | Goalie: Right |
|---|---|---|
| **Kicker: Left** | 58 | 95 |
| **Kicker: Right** | 93 | 70 |

**Analysis:**
1. No pure equilibrium — best-response arrows cycle
2. Rule 5 applies (zero-sum game)
3. Kicker indifference equation → p = 38.3% Left, 61.7% Right
4. Goalie indifference equation → y = 41.7% Left, 58.3% Right
5. Equilibrium success rate: 79.6% (minimax = maximin by von Neumann's theorem)

**Recommendation:** Kicker randomizes Left 38.3% using an objective device (page numbers, watch second hand). Goalie randomizes Left 41.7%. Any predictable pattern invites exploitation.

### Example 3 — Coordination game with multiple equilibria (focal-point selection)

**Situation:** Two division managers must independently choose which cloud platform to deploy on (AWS or Azure). Payoffs: both benefit equally from coordinating (3 each), neither benefits from mismatching (0 each). Two Nash equilibria: (AWS, AWS) and (Azure, Azure).

**Analysis:**
1. Both equilibria are Nash — each is a mutual best response
2. No dominant strategy; game is pure coordination
3. **Do not mix** — independent randomization at 50/50 produces miscoordination 50% of the time, yielding expected payoff 1.5 < 3
4. Focal-point check: Does the company already use one platform? Is there an industry default? Is one option mathematically simpler? → Use whichever has the strongest salience as the focal point
5. If no focal point exists, establish one through explicit pre-game communication

---

## References

Detailed worked calculations and additional examples are in:
- `references/mixed-strategy-calculation.md` — Full algebraic and graphical walkthrough of 2×2 and 3×3 mixed strategy problems
- `references/game-type-classifier.md` — How to distinguish zero-sum, coordination, and mixed-motive games
- `references/focal-point-examples.md` — Schelling's classic examples and modern applications

## License

This skill is licensed under [CC-BY-SA-4.0](https://creativecommons.org/licenses/by-sa/4.0/).
Source: [BookForge](https://github.com/bookforge-ai/bookforge-skills) — The Art of Strategy: A Game Theorist's Guide to Success in Business and Life by Avinash K. Dixit, Barry J. Nalebuff.

## Related BookForge Skills

This skill is standalone. Browse more BookForge skills: [bookforge-skills](https://github.com/bookforge-ai/bookforge-skills)

FILE:references/focal-point-examples.md
# Focal Point Examples and Selection Guide

Thomas Schelling's focal point concept and its application to equilibrium selection.

---

## What a Focal Point Is

A focal point (Schelling point) is a Nash equilibrium that stands out as "obvious" to all players without communication. Its salience is not a property of the game's payoffs — it comes from the players' shared context: cultural background, mathematical properties, historical convention, or the physical structure of the problem.

**The critical requirement:** A focal point must be mutually obvious. It must be obvious to Player A that it is obvious to Player B that it is obvious to Player A... and so on without limit. An equilibrium that seems obvious to one player but not the other fails as a focal point.

---

## Sources of Salience

### 1. Mathematical Salience

Numbers that stand out on purely formal grounds:

- **1** — the first positive integer, the smallest, the "unit"
- **0** — the empty set, zero contribution
- **Equal split** — the symmetric, mathematically neutral division
- **Round numbers** — 100, 50, 10, 1000
- **Powers of 2** — 2, 4, 8, 16 (salient in computing contexts)

**Application:** In the "pick a number 0-100 where the winner is closest to half the other's number" game, the Nash equilibrium is 0. But the focal point for first-time players is often 50 (a round number) or 25 (half of 50). The Nash equilibrium is only reached through iterated reasoning.

**Business application:** A 50/50 equity split in a partnership is focal even when the parties' contributions are unequal, because equal split is mathematically neutral and appears fair.

### 2. Cultural and Linguistic Convention

Shared associations from the players' common background:

- Geographic conventions: East/West Mississippi division, North/South of a border
- Linguistic: alphabetical ordering, "first" choice in a list
- Industry norms: standard contract terms, market-clearing price conventions
- Social conventions: noon as meeting time, front door as entrance

**Schelling's experiment:** Yale and Boston students asked to meet a stranger in New York City without prior communication, overwhelmingly chose noon as the time and Grand Central Station or the Empire State Building as the location — even with no coordination. The shared cultural reference created convergent expectations.

**Business application:** When two teams must split responsibility across nine cities, teams with shared cultural background converge on geographic East/West split. Teams without shared background fail to coordinate.

### 3. Symmetry and Fairness

Outcomes that treat players symmetrically are focal when:
- Players cannot readily determine who is "more deserving"
- The relationship is ongoing and fairness matters for reputation
- Breaking symmetry requires justification neither party can provide

**Application:** In CEO compensation, "above average" is a convergent aspiration. Because every firm wants above-average CEOs and everyone wants to pay them above average, the focal point escalates. To shift the focal point, a different comparison standard must be established (e.g., public service prestige).

**Application:** Wage bargaining often converges on percentage increases (cost of living, inflation) because a percentage treats all workers symmetrically.

### 4. Prior Interaction and Precedent

The most powerful focal points come from prior play:
- Previous agreements or contracts set expectations for renewal
- Historical prices are focal for re-negotiation
- Past technology choices create path dependency

**Application:** In contract renewal negotiations, the prior contract's terms are the focal point. Deviation requires explicit justification; the prior terms are the default.

### 5. Physical or Structural Salience

The physical setup of the problem creates natural meeting points:
- "Lost and Found" window as meeting place in a department store
- Railway platforms' designated "meeting point" (Treffpunkt in German stations)
- Front door vs. side entrance

---

## When Focal Points Fail

Not all games with multiple equilibria have focal points:

1. **Players lack shared context** — The Boston/San Francisco student example showed that non-American students failed to find the geographic East/West split that was obvious to Americans.

2. **Too many equilibria** — In the city-division game with 9 cities, there are 2^9 = 512 Nash equilibria. Mathematical salience (equal split of 4-5) is blocked by odd numbers.

3. **Competing salient features** — If one player finds one feature salient (alphabetical order) and another finds a different feature salient (geographic proximity), expectations diverge.

4. **New situations** — Established conventions do not apply; no new one has formed.

**When no focal point exists:** Explicitly acknowledge this. Coordination failure is a real risk. Options:
- Establish a focal point through explicit prior communication before the game is played
- Use an external randomizing device with publicly agreed-upon rules
- Delegate selection to a neutral third party

---

## Focal Points as Strategic Tools

Players can sometimes create focal points strategically:

**Making a convention obvious:** Companies that "name" their standard practice create a focal point for industry norms. Once enough competitors adopt a standard, it becomes the coordination equilibrium.

**Prior announcement:** Even a one-sided announcement ("we will always price at $X") can create a focal point if the other party observes it and has reason to believe it.

**Note:** Creating a focal point is distinct from making a commitment. A focal point works through shared expectations; a commitment works through changing payoffs or eliminating options. Commitments are covered in the commitment-strategy-designer skill.

---

## Quick Reference

| Source of salience | Check | Example |
|---|---|---|
| Mathematical | Is one equilibrium uniquely simple, symmetric, or extreme? | Equal split, smallest integer |
| Cultural | Do players share background that makes one outcome obvious? | Geographic convention, industry norm |
| Physical structure | Does the game's setting highlight one meeting point? | Lost and Found, front door |
| Prior interaction | Have these players coordinated before? | Prior contract, past price |
| Communication | Can players talk before choosing? | Any agreement, even vague, becomes focal |
| No focal point | None of the above — acknowledge coordination failure risk | Need explicit pre-game coordination |

FILE:references/game-type-classifier.md
# Game Type Classifier

Identifies the type of simultaneous-move game to ensure the correct analysis method is applied.

---

## Why Classification Matters

The correct strategy depends on the game type:

| Game Type | Interests | Pure Equilibria? | Mixing? |
|---|---|---|---|
| Zero-sum / pure conflict | Exactly opposed | Sometimes | Yes — Rule 5 applies |
| Coordination | Aligned | Usually multiple | No — focal point instead |
| Mixed-motive | Partly aligned, partly opposed | Often multiple | Possible but fragile |
| Prisoners' dilemma | Individually rational, collectively bad | One (bad) | No |

---

## Zero-Sum (Pure Conflict) Games

**Signature:** What one player gains, the other loses exactly. Payoffs in each cell sum to a constant (often 0 or 100).

**Examples:** Penalty kicks, military feints, inspection/evasion, competitive espionage, zero-sum market share battles.

**Identifying test:** In each cell of the payoff matrix, check if the two payoffs sum to the same number (or to 0 if negative payoffs are used). If yes, zero-sum.

**Analysis approach:**
1. Show only row player's payoffs (column player's = constant minus row's)
2. Check for dominant strategies first
3. If no pure equilibrium, use Rule 5 (indifference principle)
4. Mixing is optimal and necessary if the game has no pure Nash equilibrium

---

## Coordination Games

**Signature:** Players benefit from matching each other's choices. Miscoordination produces the worst outcomes. Both players prefer any coordination to no coordination.

**Examples:** Meeting point selection, platform adoption, technology standards, project methodology alignment, driving on the same side of the road.

**Identifying test:** The best cells are on the diagonal (where players match), and these are all better than off-diagonal cells for both players.

**Classic structures:**
- **Pure coordination:** Players have identical preferences over the equilibria (stag hunt, meeting game)
- **Battle of sexes:** Players agree that coordinating beats mismatching, but disagree about which coordinated outcome is better

**Analysis approach:**
1. Find all Nash equilibria (typically multiple)
2. Do NOT apply Rule 5 (mixing) — uncoordinated mixing leads to miscoordination
3. Apply focal-point analysis to select an equilibrium
4. If repeated interaction, consider alternation or explicit agreement

**Anti-pattern warning:** Applying the indifference principle to a coordination game produces a mixed strategy equilibrium, but this equilibrium has lower expected payoff than either pure equilibrium. Players who independently follow this mixed strategy will end up miscoordinated more than the pure equilibria would predict.

---

## Mixed-Motive Games (Non-Zero-Sum with Conflict)

**Signature:** Players have some aligned interests (prefer coordination to chaos) but different preferences about which outcome to coordinate on.

**Examples:** Chicken game, competitive promotions (Coke/Pepsi), labor negotiations, technology format wars.

**Identifying test:** There are multiple Nash equilibria, and different players prefer different equilibria.

**Analysis approach:**
1. Find all pure Nash equilibria
2. A mixed Nash equilibrium exists but typically has lower expected payoffs than the pure equilibria
3. Consider commitment devices to secure the preferred equilibrium (see commitment-strategy-designer skill)
4. If the game is repeated, cooperation solutions may dominate

---

## Prisoners' Dilemma Structure

**Signature:** Each player has a dominant strategy that leads to a collectively bad outcome. The Nash equilibrium is worse for both than a coordinated alternative.

**Identifying test:** One strategy strictly dominates all others for each player, and the resulting Nash equilibrium is Pareto-dominated (both players would prefer a different outcome if they could coordinate).

**Analysis approach:**
1. Apply Rule 2 (dominant strategy) — the equilibrium is determined
2. Note that the equilibrium is stable even though both parties would prefer mutual cooperation
3. In repeated interactions, cooperation may be sustained (separate repeated-games skill)

---

## Quick Classification Protocol

1. Do payoffs in each cell sum to a constant? → **Zero-sum** → use Rule 5 if no pure equilibrium
2. Are all on-diagonal outcomes better for both players than off-diagonal? → **Coordination game** → focal-point analysis, no mixing
3. Does one strategy dominate for each player, but the result is worse than mutual cooperation? → **Prisoners' dilemma** → dominant strategy equilibrium, note the trap
4. None of the above → **Mixed-motive** → find all equilibria, consider commitment or repetition

FILE:references/mixed-strategy-calculation.md
# Mixed Strategy Calculation Reference

Full algebraic and graphical methods for computing Nash equilibrium mixing proportions.

---

## The Indifference Principle

The key insight: in a mixed strategy Nash equilibrium, each player's mixing proportions are chosen to make their **opponent** indifferent between their pure strategies. If your opponent strictly preferred one strategy over another against your mix, they would abandon mixing and play that pure strategy — and you should have responded differently. So equilibrium requires opponent indifference.

**Consequence:** Your equilibrium mix is determined by your opponent's payoffs. If your own payoffs change, your equilibrium mix is unaffected — only your opponent's mix changes.

---

## 2x2 Algebraic Method

**Setup:** Row player mixes strategy A (probability p) and strategy B (probability 1−p).

**Step 1:** Write the column player's expected payoff for each of their strategies as a function of p.

If column player's payoffs are (a, b) when column plays Left against row's A and B respectively, and (c, d) when column plays Right:
- Expected payoff from Left: a·p + b·(1−p)
- Expected payoff from Right: c·p + d·(1−p)

**Step 2:** Set the two expressions equal and solve for p.

a·p + b·(1−p) = c·p + d·(1−p)

**Step 3:** Verify p is between 0 and 1. If not, a pure strategy equilibrium exists (one strategy dominates).

**Step 4:** Repeat with column player's proportion y to find row player's equilibrium mix.

---

## Worked Example: Penalty Kick

Payoff table (kicker success percentages; goalie minimizes):

|  | Goalie: Left | Goalie: Right |
|---|---|---|
| **Kicker: Left** | 58 | 95 |
| **Kicker: Right** | 93 | 70 |

**Finding kicker's mix (p = proportion Left):**

Set goalie's worst-case equal. Goalie prefers to minimize kicker's success. When kicker uses proportion p of Left:
- If goalie plays Left: 58p + 93(1−p) = 93 − 35p
- If goalie plays Right: 95p + 70(1−p) = 70 + 25p

Set equal: 93 − 35p = 70 + 25p → 23 = 60p → **p = 23/60 ≈ 0.383**

Kicker: Left 38.3%, Right 61.7%. Equilibrium success rate: 93 − 35(0.383) = 79.6%

**Finding goalie's mix (y = proportion Left):**

When goalie uses proportion y of Left:
- If kicker plays Left: 58y + 95(1−y) = 95 − 37y
- If kicker plays Right: 93y + 70(1−y) = 70 + 23y

Set equal: 95 − 37y = 70 + 23y → 25 = 60y → **y = 25/60 ≈ 0.417**

Goalie: Left 41.7%, Right 58.3%. Kicker's success rate: 70 + 23(0.417) = 79.6%

**Von Neumann's minimax theorem:** The kicker's maximum of minima (79.6%) equals the goalie's minimum of maxima (79.6%). This equality is guaranteed for zero-sum games — compute the best mix for one player and you know the equilibrium value for both.

---

## 3x3 System of Equations

For three strategies per player, let row player mix with p (strategy 1), q (strategy 2), (1−p−q) (strategy 3).

**Step 1:** Write column player's expected payoff for each of their three strategies as a function of p and q.

**Step 2:** Set all three equal. This gives two independent equations (the third is implied by the first two plus the constraint p + q + (1−p−q) = 1).

**Step 3:** Solve the system for p and q. Check that p > 0, q > 0, (1−p−q) > 0.

**Example: Janken step game** (win with Paper = +5 steps, Scissors = +2, Rock = +1; losses are negatives):

Row player's payoff from each of column's strategies, expressed in terms of Takashi's p (Paper) and q (Scissors):

- Yuichi's Rock payoff: −5p + 1q + 0(1−p−q) = −5p + q
- Yuichi's Paper payoff: 0p − 2q + 5(1−p−q) = −5p − 7q + 5
- Yuichi's Scissors payoff: 2p + 0q − 1(1−p−q) = 3p + q − 1

Set Rock = Paper: −5p + q = −5p − 7q + 5 → 8q = 5 → q = 5/8

Set Rock = Scissors: −5p + q = 3p + q − 1 → −8p = −1 → p = 1/8

Therefore: p(Paper) = 1/8, p(Scissors) = 5/8, p(Rock) = 2/8

---

## Graphical Method (Minimax V-Shape)

For a 2x2 zero-sum game, plot the row player's success rate against their mixing proportion p (horizontal axis, 0 to 1).

**Two lines:** One showing success against column's Left pure strategy, one showing success against column's Right pure strategy.

- Line for column's Left: starts at payoff(row Right vs col Left) when p=0, ends at payoff(row Left vs col Left) when p=1. This line may rise or fall depending on payoff values.
- Line for column's Right: starts at payoff(row Right vs col Right) when p=0, ends at payoff(row Left vs col Right) when p=1.

**The goalie (column player) will always play the pure strategy that minimizes the kicker's success** — they choose the lower of the two lines at each p.

**The resulting envelope** (lower of the two lines at each p) forms an inverted-V. The kicker maximizes their guaranteed success by choosing p at the apex — where the two lines cross.

**The intersection point p = 0.383** gives success rate 79.6%.

**For the goalie's mix:** Draw the same plot from the goalie's perspective (y on horizontal axis). Two lines show kicker's success against goalie's Left vs. Right mixtures. The kicker always picks the higher line. The goalie minimizes by choosing y at the bottom of the V (where lines cross). This is y = 0.417, same success rate 79.6%.

---

## When Pure Strategy Equilibria Exist in Zero-Sum Games

Not all zero-sum games require mixing. If one strategy dominates all others for the row player, play it; the game is solved. More generally, if the row player's maximin (best of worst cases) equals the column player's minimax (worst of best cases) in pure strategies, a pure strategy equilibrium exists.

Example: If kicker success when going Left is 38 (vs goalie Left) and 65 (vs goalie Right), but Right gives 93 and 70 respectively — Right dominates Left for the kicker. No mixing needed; the kicker should always go Right.

The mixing method will confirm this: solving the indifference equation gives p < 0 or p > 1, indicating the pure strategy is optimal.

---
name: information-asymmetry-strategist
description: "Diagnose and resolve information asymmetry in strategic interactions using four mechanisms: signaling, screening, signal jamming, and countersignaling. Use this skill when a user needs to credibly communicate private information to an uninformed counterpart; when a user needs to elicit honest information from a better-informed counterpart without being able to verify their claims; when a user suspects they are on the receiving end of signal jamming or adverse selection and wants to see through it; when a user is designing a pricing scheme, contract structure, hiring process, or product menu and needs to induce self-selection among different customer or candidate types; when a user wants to know whether to signal their quality, countersignal by staying silent, or jam an opponent's signals; when a user needs to apply Bayes' rule to update beliefs after observing an opponent's action in a mixed-strategy game; when a user faces Akerlof-style market collapse risk and wants to identify signaling or screening remedies; when a user is designing a menu of options (e.g., airline fare classes, insurance deductibles, product tiers) and needs to check participation constraints and incentive compatibility constraints. This skill handles both directions of information asymmetry: the informed party communicating outward (signaling, countersignaling, jamming) and the uninformed party extracting information inward (screening, adverse selection management). It does NOT cover moral hazard or principal-agent problems after a contract is signed, nor does it handle simultaneous-move games without information asymmetry (use the Nash equilibrium skill for those)."
version: 1.0.0
homepage: https://github.com/bookforge-ai/bookforge-skills/tree/main/books/the-art-of-strategy/skills/information-asymmetry-strategist
metadata: {"openclaw":{"emoji":"📚","homepage":"https://github.com/bookforge-ai/bookforge-skills"}}
status: draft
source-books:
  - id: the-art-of-strategy
    title: "The Art of Strategy"
    authors: ["Avinash K. Dixit", "Barry J. Nalebuff"]
    chapters: [8]
tags: [game-theory, information-economics, signaling, screening, adverse-selection]
depends-on: []
execution:
  tier: 1
  mode: full
  inputs:
    - type: document
      description: "Description of the strategic situation: who holds private information, what they want the other side to believe or infer, what actions are available, and the cost structure for each type of player"
  tools-required: [Read, Write]
  tools-optional: []
  mcps-required: []
  environment: "Any agent environment; user describes the situation in text or structured form"
discovery:
  goal: "Identify which information asymmetry mechanism applies to the user's situation, design credible signals or effective screening menus, diagnose adverse selection and prescribe remedies, apply Bayes' rule to update beliefs from observed actions"
  tasks:
    - "Classify the information structure: who knows what, who benefits from revealing vs. concealing it"
    - "Identify the active mechanism: signaling, screening, signal jamming, or countersignaling"
    - "For signaling: verify the cost-difference property holds (signal is credible only if cost to the wrong type exceeds cost to the right type by more than the informational value)"
    - "For screening: design a menu that satisfies both the participation constraint (target type remains willing to transact) and the incentive compatibility constraint (wrong type prefers their intended option)"
    - "For adverse selection: diagnose whether bad types are being systematically attracted and prescribe a remedy (signal, screen, or positive selection)"
    - "Apply Bayes' rule to update beliefs from observed actions in mixed-strategy or semi-separating equilibria"
    - "Identify the equilibrium type: separating, pooling, or semi-separating"
    - "Check for countersignaling conditions: determine whether top types should refrain from signaling entirely"
    - "Deliver structured recommendations: which mechanism to use, specific design of the signal or menu, predicted equilibrium, and informational externality costs"
  audience: "Business strategists, product managers, hiring managers, marketers, negotiators, policy designers, and anyone who must communicate credibly or elicit honest information from strategic counterparts"
  when_to_use:
    - "User needs to convince a skeptical counterpart of their quality, commitment, or type without being able to simply say so"
    - "User is designing a contract, pricing tier, or application process to attract a specific customer or candidate type while discouraging others"
    - "User suspects they are in an adverse selection situation where bad types are systematically attracted to their offer"
    - "User wants to interpret an opponent's action probabilistically using Bayes' rule after observing a signal in a game with mixed strategies"
    - "User is unsure whether to signal their ability/quality or stay silent and countersignal"
  quality:
    correctness: null
    depth: null
    actionability: null
    specificity: null
---

# Information Asymmetry Strategist

## When to Use

Use this skill when one player in a strategic interaction knows something important that another player does not — and both sides are trying to act strategically given this imbalance.

The core principle: **actions speak louder than words.** When interests are not fully aligned, claims cannot be trusted. But actions — especially costly ones — carry information precisely because they are not free to fake. The framework maps four ways to exploit or manage this: signaling, screening, signal jamming, and countersignaling.

This skill applies when:
- One party has private information (about quality, type, intentions, or capability) that affects both parties' payoffs
- That party has an incentive to reveal, conceal, or manipulate what the other infers
- The uninformed party wants to extract honest information or protect itself from strategic misrepresentation

This skill does NOT apply to:
- Post-contract moral hazard (hidden actions after agreement — different problem)
- Simultaneous-move games with symmetric information (use the Nash equilibrium skill)
- Pure negotiation where information is symmetric (use the negotiation BATNA framework)

---

## Context and Input Gathering

### Required (ask if missing)

- **Who holds the private information?** Is it the seller, candidate, counterpart, or the user themselves?
  -> Ask: "Who knows something relevant that the other side cannot directly observe?"

- **What is the private information?** Quality of a product, level of risk, genuine intention, capability, or type?
  -> Ask: "What exactly does the informed party know that the uninformed party wants to know?"

- **What is the direction of benefit?** Does the informed party want to reveal this information (to get a better deal) or conceal it (to avoid disadvantage)?
  -> Ask: "Does the informed party benefit from the other side knowing the truth, or from keeping it hidden?"

- **What actions are available?** What can the informed party do that the uninformed party can observe?
  -> Ask: "What observable actions can the informed party take? What can they offer, invest, display, or commit to?"

- **Cost asymmetry between types:** Does the potential signal cost more for one type than another?
  -> Ask: "Would this action cost more — in money, time, risk, or inconvenience — for someone who does NOT have the private information?"

### Useful (gather if present)

- Base rates: what proportion of the population are the "good" vs. "bad" type?
- The magnitude of payoff differences between types (needed for Bayes' rule calculations)
- Whether the interaction is one-shot or repeated (reputation effects change signaling economics)
- What screening options the uninformed party controls (contract terms, product versions, timing constraints)

---

## Execution

### Step 1 — Classify the Information Structure

**Why:** The four mechanisms work in opposite directions — applying the wrong one wastes effort and may backfire. A credible signal requires cost differences that may not exist. Screening requires control over the menu that the uninformed party may not have. Getting the classification right in thirty seconds prevents misdirected analysis.

**1a. Who holds private information?**

- Informed party = seller, applicant, counterpart → they have private knowledge about themselves
- Uninformed party = buyer, employer, user → they want to elicit or infer that knowledge

**1b. Who acts first?**

- Informed party acts first to reveal information → **Signaling** (seller offers warranty; applicant gets MBA)
- Uninformed party designs menu to force self-selection → **Screening** (employer requires MBA; insurer offers deductible tiers)
- Informed party acts to suppress or obscure information → **Signal Jamming** (poker bluffing; corporate obfuscation)
- Informed party refrains from signaling to communicate top status → **Countersignaling** (old money doesn't flaunt; top mathematicians don't follow convention)

Note: signaling and screening often produce similar equilibria via different initiative paths. The same credential (MBA) can be a screening device (firm requires it) or a signaling device (candidate volunteers it). The key principle — cost difference between types — is the same either way.

---

### Step 2 — Apply the Cost-Difference Property (For Signaling and Screening)

**Why:** This is the decisive test for whether any signal or screen will actually work. A signal that is equally cheap for the wrong type to mimic provides no information — it will be mimicked, and the equilibrium collapses. Every credible signal in every domain — warranties, education, tattoos, gang initiation rites — works because it satisfies this property.

**The cost-difference property:** A signal is credible if and only if it costs more for the wrong type to send than for the right type. The cost difference must exceed the informational value of the signal.

**Formally:** For a signal to separate types in equilibrium:
- Cost of signal for the "false" type > Cost of signal for the "true" type
- The gap must be large enough that the false type's expected gain from mimicking (the price premium or employment benefit from being mistaken for the true type) does not outweigh the cost

**Worked diagnostic:**

| Claimed signal | True type cost | False type cost | Cost difference | Signal credible? |
|---|---|---|---|---|
| Warranty | Low (cheap to honor good car) | High (costly repairs on bad car) | Large | Yes |
| Mechanic inspection offer | Zero (no commitment) | Zero (can walk away if bad) | None | No |
| MBA degree (talented) | $200K, certain pass | $200K, 50% fail risk | Extra $100K expected | Yes (if wage premium > $40K/yr) |
| Clean car for sale | Minimal | Minimal | None | No (pooling equilibrium) |
| Tattoo with partner's name | Low (committed person) | High (non-committed person) | Large | Yes |

**When to use each type of signal:**

- **Financial guarantees / warranties:** Credible when repair/fulfillment cost differs sharply by quality
- **Costly credentials / degrees:** Credible when completion rate or opportunity cost differs sharply by underlying ability
- **Irreversible commitments:** Credible when the commitment destroys future options that only a non-serious party would value
- **In-kind benefits (not cash):** Credible for self-selection when secondary value (resale of a wheelchair) differs sharply between true and false claimants

---

### Step 3 — Identify the Equilibrium Type

**Why:** The outcome of a signaling game depends on the mix of types in the population and the size of cost differences. Knowing which equilibrium type you are in determines both how to interpret the other side's actions and which interventions will shift the equilibrium.

**Separating equilibrium:** Different types take different observable actions. The action reliably identifies the type. This is the desired outcome for signaling and screening. Requires: the cost-difference property holds strongly enough that the wrong type will not mimic.

**Pooling equilibrium:** All types take the same action. The action conveys no information. This occurs when: cost differences are small, the proportion of wrong types is small (so even wrong types benefit from mimicking), or the signal is free to imitate. A clean car at sale time is a pooling signal when everyone cleans their car before selling.

**Semi-separating equilibrium:** Some of the wrong type mimic, some do not. The action is informative but not perfectly so. Occurs when cost differences are modest relative to the gain from mimicking. Requires Bayes' rule to interpret the resulting probabilistic signal. A dirty car would be a sure indicator of carelessness; a clean car is likely (but not certain) to indicate care.

**Diagnostic for equilibrium type:**

1. Is the cost difference large relative to the gain from mimicking? → Separating
2. Is the proportion of wrong types small? → Pooling (everyone mimics; signal is uninformative)
3. Is the cost difference small but positive? → Semi-separating (mixed-strategy mimicking, use Bayes' rule)

---

### Step 4 — Apply Bayes' Rule to Update Beliefs from Observed Actions

**Why:** In semi-separating equilibria (and when opponents play mixed strategies), observed actions are informative but not decisive. Bayes' rule is the correct tool for updating from "what I believed before" to "what I should believe now given what I observed." Applying it prevents both overconfidence (acting as if a signal is perfectly revealing) and underconfidence (ignoring a signal entirely).

**Bayes' Rule for type inference:**

P(true type | observed action) = P(action | true type) × P(true type) / P(action)

Where: P(action) = P(action | true type) × P(true type) + P(action | false type) × P(false type)

**Worked example (poker bluffing):** A rival raises 2/3 of the time with a good hand and 1/3 of the time with a poor hand. Prior probability of good hand = 1/2.

- P(raise | good hand) = 2/3
- P(raise | poor hand) = 1/3
- P(raise) = (1/2)(2/3) + (1/2)(1/3) = 1/3 + 1/6 = 1/2
- P(good hand | raise) = (1/3) / (1/2) = **2/3**

After observing a raise, update from 1/2 prior to 2/3 posterior. The raise is informative but not conclusive — there is still a 1/3 chance of a bluff. Decisions after the raise should incorporate this updated probability, not the original 50/50.

**Using Bayes' rule diagnostically:**

- If P(action | true type) >> P(action | false type): the action is highly informative; posterior shifts strongly toward true type
- If P(action | true type) ≈ P(action | false type): the action is barely informative; posterior barely moves from prior
- If you observe "fold" or another action unique to one type: posterior collapses to certainty (P = 0 or 1)

---

### Step 5 — Design or Evaluate a Screening Menu

**Why:** When the uninformed party controls the transaction structure (employer, insurer, seller offering product tiers), they can design a menu that induces each type to self-select into the option designed for them. This is more powerful than waiting to receive signals. But a menu that ignores the two constraints below will fail: either the target type opts out entirely, or the wrong type defects to the other option.

**Two binding constraints in screening design:**

**Participation constraint (PC):** The option designed for type T must offer that type at least as much value as opting out entirely. If the price exceeds the type's maximum willingness to pay (reservation price), they do not participate.

- For a screen to work: price ≤ reservation price of the target type
- Binding PC = you are extracting the maximum surplus from that type consistent with their participation

**Incentive compatibility constraint (ICC):** The option designed for type T must give type T at least as much surplus as the option designed for the other type.

- For a screen to work: surplus in own option ≥ surplus in the other option
- Binding ICC = the other type would be exactly indifferent; any worse and they defect

**Screening design procedure:**

1. Identify the two (or more) types and their reservation prices for each product/service version
2. Set the price for the low type at or below their reservation price (binding PC)
3. Calculate the consumer surplus the low-type option gives to the high type
4. Set the price for the high-type option so the high type's surplus equals or exceeds what they get from the low-type option (binding ICC)
5. Verify the resulting profit structure; compare against the alternative of serving only the high type at their full reservation price
6. If the proportion of low types is very small, consider whether it is more profitable to ignore them entirely (violate their PC) and charge high types their full reservation price

**Airline pricing illustration (PITS example):**

| Service | Cost | Tourist reservation price | Business reservation price |
|---|---|---|---|
| Economy | 100 | 140 | 225 |
| First | 150 | 175 | 300 |

- Tourist PC: Economy price ≤ 140 → set at 140 (binding)
- Business ICC: surplus from First class ≥ surplus from Economy at 140 → 300 − First price ≥ 225 − 140 = 85 → First price ≤ 215
- Set First class at 215; business travelers' surplus = 300 − 215 = 85 = their surplus from Economy at 140 (225 − 140 = 85) → they are indifferent, but choose First if there is any tiebreaker
- Profit per 100 passengers: (140−100)×70 + (215−150)×30 = 2800 + 1950 = 4750 vs. 7300 with perfect discrimination → informational externality cost = 2550

**The informational externality:** The cost of screening vs. perfect price discrimination equals the ICC rent multiplied by the number of high-type customers (85 × 30 = 2550). This cost exists because the low types, by existing, force the seller to give the high types a surplus to keep them from defecting.

---

### Step 6 — Diagnose and Remedy Adverse Selection

**Why:** Adverse selection — the systematic over-representation of bad types in a transaction — is the most common and damaging consequence of information asymmetry. It can cause entire markets to collapse (Akerlof's lemons). Correctly diagnosing adverse selection is the first step; the remedies are specific to the mechanism.

**Adverse selection diagnosis checklist:**

1. Is there information asymmetry where the transacting party knows their own type but the counterpart does not?
2. Does the current price or offer structure attract bad types more than good types (or exclusively bad types)?
3. Has the good type's willingness to transact fallen as the price has been adjusted to reflect the increasing proportion of bad types?
4. Has the market thinned, collapsed, or settled at a "lemons" price?

**If yes to 2-4: adverse selection is active.**

**Akerlof lemons mechanism (used car market):**
- If sellers know quality but buyers do not, and half the cars are lemons ($1K seller min, $1.5K buyer max) and half are peaches ($3K seller min, $4K buyer max):
- Buyers bid the expected value = (1/2)(1.5K) + (1/2)(4K) = $2.75K
- Peach owners will not sell at $2.75K (below their $3K floor) → only lemons offered
- Buyers, inferring this, bid only $1.5K → market for peaches collapses entirely
- The Groucho Marx effect: "Any car willing to sell at this price is not a car you would want to buy"

**Remedies by mechanism:**

| Situation | Remedy | Mechanism |
|---|---|---|
| Seller has quality information | Seller offers warranty, money-back guarantee, or long-term service contract | Signaling |
| Buyer has access to screening levers | Buyer requires certification, trial period, or deductible structure | Screening |
| Insurer cannot identify risk types | Offer deductible tiers: low-risk types prefer high deductible (lower premium); high-risk types prefer low deductible (higher premium) | Screening |
| Employer cannot observe talent | Require costly credential differentially costly to untalented | Screening |
| Bad types attracted to standard offer | Redesign offer to be unattractive to bad types (Capital One balance transfer: unattractive to maxpayers and deadbeats, attractive only to revolvers) | Positive selection |

**Positive selection (Capital One example):** Adverse selection can be reversed by designing offers that are attractive only to the profitable type. A balance transfer offer attracts revolvers (who have balances to transfer and intend to repay) while being irrelevant to maxpayers (no balance to transfer) and unattractive to deadbeats (who plan to default regardless). Any customer who accepts the offer is one you want.

---

### Step 7 — Evaluate Countersignaling (Should You Stay Silent?)

**Why:** The intuition that "you should always signal if you can" is wrong in some conditions. When there are three or more types (e.g., gold digger, question mark, and true love; or weak, average, and expert), and the uninformed party can distinguish the top type from others through alternative signals or background knowledge, the top type may benefit from NOT signaling. Signaling in this context reveals that you feel the need to distinguish yourself from the middle type — which is exactly what only the middle type needs to do.

**Countersignaling conditions (all three required):**

1. There are at least three types (bottom, middle, top), not just two
2. The top type is distinguishable from the bottom type even without the signal (through other observable attributes or base rate knowledge)
3. The middle type, by signaling, reveals that they are not the top type (because top types do not need to signal)

**Equilibrium with countersignaling:**
- Bottom type: does not signal (cannot afford it, or chooses not to play)
- Middle type: signals (to distinguish from the bottom type)
- Top type: does not signal (to distinguish from the middle type)

**Result:** Signaling is a signal of being middle-tier. The top type signals by the very absence of signaling.

**Examples:**
- Old money does not flaunt wealth; the nouveau riche does. Old money is identifiable through other means; flaunting reveals one is new money trying to be distinguished from those with no money.
- Highly accomplished faculty at top PhD programs use first names only in voicemails; faculty at lesser institutions use their title "Professor Dr." to distinguish themselves from those without doctorates.
- Expert negotiators do not demonstrate expertise through elaborate terminology; novices who have just learned the jargon use it constantly.

**Decision rule for countersignaling:**

- Are you clearly distinguishable from the bottom type even without the signal? → If yes, weigh countersignaling
- Would signaling group you with the middle type in the observer's inference? → If yes, countersignaling is better
- Is the observer sophisticated enough to run this three-way inference? → If no, you may need to signal anyway

---

### Step 8 — Signal Jamming (Obscuring Your Type)

**Why:** Sometimes your optimal strategy is not to communicate your type but to prevent the other side from inferring it. Mixed strategies in poker serve this purpose. Jamming preserves strategic uncertainty — keeping the opponent guessing maintains option value and prevents exploitation.

**When signal jamming applies:**
- Your interests are opposed to the other party (zero-sum or strongly competitive context)
- The other party would benefit from knowing your type and change their strategy accordingly
- You can randomize your actions credibly (or the cost of randomization is low)

**Signal jamming mechanics:**
- The "tight" poker player who never bluffs is exploitable: opponents know large bets mean good hands and fold accordingly, keeping pots small
- The "loose" player who always bluffs is also exploitable: opponents always call
- Optimal play mixes bluffing and legitimate raises at an equilibrium frequency derived from payoffs

**Key rule:** When interests are fully opposed, ignore what the other side says entirely. Do not assume their statement is true; do not assume its opposite is true. Play the equilibrium strategy as if no statement were made. The statement carries zero information content when interests are completely opposed.

**Signal jamming in business contexts:**
- Randomizing project launch timing to prevent competitors from inferring R&D readiness
- Mixing price promotions irregularly to prevent customers from timing purchases around predictable discounts
- Providing employees with information that is accurate in aggregate but not attributable to specific decisions, obscuring the decision rule

---

### Step 9 — Deliver Structured Recommendations

Structure your output as:

**Information asymmetry type:** [Who knows what; which direction it cuts]

**Active mechanism:** [Signaling / Screening / Signal Jamming / Countersignaling — and why]

**Cost-difference assessment:** [Does the cost-difference property hold? What is the cost to the true type vs. false type of the recommended signal or screen?]

**Equilibrium prediction:** [Separating / Pooling / Semi-separating — and why given cost differences and population proportions]

**Recommended action:**
- For signaling: specific action to take, why it satisfies the cost-difference property, and how the other party will update their beliefs
- For screening: specific menu design with prices/conditions, participation constraint check, incentive compatibility constraint check
- For adverse selection: diagnosis of which type is being over-attracted and which remedy (signal, screen, or positive selection redesign) applies
- For countersignaling: whether you are in a three-type situation, whether staying silent signals top-tier status

**Informational externality cost:** [What does correcting the asymmetry cost in total, and who bears it?]

**Failure modes:** [Under what conditions will this signal be mimicked, this screen be gamed, or this equilibrium collapse to pooling?]

---

## Key Principles

**Actions speak louder than words.** Verbal claims are cheap and will be made regardless of truth when interests diverge. Only costly, observable actions carry information — and only when the cost differs between types.

**The cost-difference property is the universal test.** A signal is credible if and only if the cost to the wrong type exceeds the cost to the right type by at least the value of the information conveyed. No other test is needed; no other test is sufficient.

**The informed party and uninformed party can both initiate.** The warranty can be offered (signaling) or demanded (screening). The MBA can be volunteered (signaling) or required (screening). The underlying equilibrium is similar; which party initiates depends on institutional context.

**Adverse selection is the systematic consequence of information asymmetry.** Without remediation, bad types crowd out good types until markets thin or collapse. The lemons market is not a special case — it is the generic outcome when information asymmetry is left unmanaged.

**Informational externalities are unavoidable.** The cost of signaling (the extra wages paid to talented workers to distinguish them from the untalented) is paid by the untalented's mere existence. This cost cannot be eliminated by any one party; it is the price of the information asymmetry.

**Countersignaling is rational for the top type in a three-tier world.** When you are clearly distinguishable from the bottom tier, signaling reveals you as middle-tier. The absence of a signal — for a top type — is itself the strongest signal.

**In purely opposed interactions, ignore the other side's statements.** When interests are fully opposed, verbal statements are strategically determined to mislead. Play the equilibrium strategy; update beliefs only from actions observed, using Bayes' rule.

---

## Examples

### Example 1: Hyundai Warranty (Signaling)

**Situation:** In 1999, Hyundai had improved quality but U.S. consumers did not believe it. Verbal claims of quality were worthless — any manufacturer can claim quality.

**Mechanism:** Signaling. Hyundai is the informed party (knows its own quality). Consumers are uninformed.

**Cost-difference analysis:** A 10-year / 100,000-mile warranty is cheap to offer if you genuinely build reliable cars (few claims). It is catastrophically expensive if your cars break down (massive warranty repair bills). The cost difference between a truly improved Hyundai and a still-defective manufacturer is large.

**Equilibrium:** Separating. A manufacturer who knew its cars were still defective would not offer this warranty — the expected repair costs would exceed the price premium gained. The signal works because only a confident manufacturer can afford to make it.

**Outcome:** Consumers rationally updated their quality beliefs. Hyundai's U.S. market share grew substantially.

---

### Example 2: Capital One Balance Transfer (Positive Selection)

**Situation:** Standard credit card offers suffer adverse selection — they attract maxpayers (no profit: merchant fees barely cover billing costs) and deadbeats (loss: default) while revolvers (most profitable: pay interest over time) are indistinguishable from the others.

**Mechanism:** Positive selection via targeted screening. The balance transfer offer is the screening device.

**Why it works:** Maxpayers have no outstanding balance — the offer is irrelevant to them. Deadbeats have no intention of repaying — bringing a balance they plan to default on provides no benefit. Revolvers, who have real outstanding balances and plan to repay, find a lower interest rate genuinely attractive. The offer self-selects only the profitable type.

**Key insight:** Capital One did not need to identify who the revolvers were. The nature of the offer caused them to identify themselves. This is the reversal of Groucho Marx: any customer who accepts this offer is exactly one you want.

---

### Example 3: Airline Fare Classes (Screening for Price Discrimination)

**Situation:** PITS airline serves 30 business travelers (reservation price: $300 first, $225 economy) and 70 tourists (reservation price: $175 first, $140 economy). PITS cannot tell them apart.

**Mechanism:** Screening. PITS designs two service tiers to induce self-selection.

**Constraint analysis:**
- Tourist PC: economy price ≤ 140 (tourists' max)
- Business ICC: first-class surplus ≥ economy surplus for business travelers
  → 300 − p_first ≥ 225 − 140 = 85 → p_first ≤ 215

**Optimal prices:** Economy = $140, First = $215
**Profit:** (140−100)×70 + (215−150)×30 = 2800 + 1950 = $4,750 per 100 passengers

**vs. perfect discrimination:** $7,300 — the $2,550 gap is the informational externality cost paid to business travelers as ICC rent (85 × 30).

**Failure mode:** If business travelers represent 50% of passengers, it may be more profitable to exclude tourists entirely and charge $300 to business travelers only (profit = (300−150)×50 = $7,500).

---

## References

- `references/cost-difference-property.md` — Formal derivation of the credibility condition, worked examples across domains, common failure cases
- `references/screening-menu-design.md` — Detailed procedure for designing participation-compatible, incentive-compatible menus with two and three types; algebraic examples
- `references/bayes-rule-inference.md` — Bayes' rule applied to type inference: worked poker example, generalized formula, updating in semi-separating equilibria
- `references/adverse-selection-remedies.md` — Akerlof lemons mechanism, insurer adverse selection, Capital One case, bureaucratic delay as screening, in-kind benefits
- `references/countersignaling-conditions.md` — Feltovich/Harbaugh/To three-type model, prenuptial example, voicemail study data, decision rule for when to countersignal

## License

This skill is licensed under [CC-BY-SA-4.0](https://creativecommons.org/licenses/by-sa/4.0/).
Source: [BookForge](https://github.com/bookforge-ai/bookforge-skills) — The Art of Strategy by Avinash K. Dixit, Barry J. Nalebuff.

## Related BookForge Skills

This skill is standalone. Browse more BookForge skills: [bookforge-skills](https://github.com/bookforge-ai/bookforge-skills)

FILE:references/adverse-selection-remedies.md
# Adverse Selection: Mechanism, Diagnosis, and Remedies

## What Adverse Selection Is

Adverse selection occurs when information asymmetry causes a transaction to attract the wrong types — systematically drawing in parties whose participation imposes costs on the uninformed counterpart. The term originated in insurance, where policies at a given price attract sicker, riskier, or more claim-prone customers than the average population. It generalizes to any market with private-type information.

**The Groucho Marx effect:** "Any club willing to accept me as a member is not one I would want to join." Applied to markets: any seller willing to sell at the average price is not one whose car (or service, or counterparty) you actually want.

## The Akerlof Lemons Mechanism (Full Walkthrough)

### Setup

Two car types:
- Lemons (bad quality): seller's minimum price = $1,000; buyer's maximum willingness to pay = $1,500
- Peaches (good quality): seller's minimum price = $3,000; buyer's maximum willingness to pay = $4,000
- Proportion: 50% lemons, 50% peaches

### With Symmetric Information (Baseline)

All transactions happen at prices between each type's floor and ceiling. Lemons trade at $1,000–$1,500. Peaches trade at $3,000–$4,000. Both markets function.

### With Asymmetric Information (Sellers Know, Buyers Don't)

**Step 1 — Buyers compute expected value:**
Expected willingness to pay = 1/2 × $1,500 + 1/2 × $4,000 = **$2,750**

**Step 2 — Peach sellers respond:**
A peach seller's floor is $3,000. No peach seller will sell at $2,750. Peaches are withdrawn from the market.

**Step 3 — Buyer inference updates:**
Buyers observe that only lemons remain. Update expected value to $1,500. Offer only $1,500.

**Step 4 — Equilibrium:**
Only lemons trade. Peach market collapses entirely — not because the good product is undesirable or overpriced, but because the buyer cannot distinguish it from lemons.

**The efficiency loss:** Buyers are willing to pay $4,000 for peaches. Sellers are willing to sell for $3,000. There is $1,000 of surplus available that is never realized. Information asymmetry causes a market failure even when all parties are acting rationally.

## Adverse Selection in Insurance

Insurance policies at a fixed premium attract disproportionately high-risk individuals:
- People with mortality rates above the break-even rate find the policy valuable (expected claims > premiums)
- People with mortality rates below the break-even rate may still buy (risk aversion, family protection), but high-risk individuals disproportionately buy larger policies

When the insurer raises premiums to cover losses, lower-risk individuals exit first (for them, the policy is now overpriced). The remaining pool is even riskier. Premiums must rise again. The process continues until only the highest-risk individuals remain — or the market collapses.

**Diagnosis check:** Are you selling an undifferentiated product where willingness to buy is higher for the "bad" type? If yes, adverse selection is active.

## The Credit Card Three-Type Taxonomy

Three customer types for credit cards:
- **Maxpayers:** Pay balance in full each month. Revenue source: merchant fees (~1-2% of transactions). Cost: billing, fraud, and the small risk of job loss/divorce leading to default. Issuers barely break even on these customers.
- **Revolvers:** Carry a balance from month to month and pay interest. Most profitable customer. Interest revenue at 15-25% APR exceeds costs substantially.
- **Deadbeats:** Carry balances but default. Pure cost to issuers.

**Standard offer adverse selection:** At a given APR, deadbeats and maxpayers both apply. Deadbeats apply because they intend to take money and not return it; maxpayers apply because the card is useful for transactions. The most valuable customers (revolvers) may apply but are mixed in with both unprofitable types.

**Capital One's positive selection solution:** The balance transfer offer is structured so that:
- Maxpayers have no balance to transfer → offer is irrelevant to them
- Deadbeats plan to default → a lower interest rate on a balance they will not repay does not change their expected behavior
- Revolvers have real balances, pay interest, and genuinely benefit from a lower rate → offer is attractive

The offer self-selects only the profitable type without requiring the issuer to identify revolvers directly. This is positive selection: designing the offer's attractiveness to be type-specific.

## Remedies for Adverse Selection

### Remedy 1: Signaling (Informed Party Acts)

The high-quality seller credibly communicates their type before the transaction. Requires the cost-difference property (see cost-difference-property.md).

**Examples:**
- Hyundai's 10-year / 100,000-mile warranty (1999): credibly communicated improved quality when reputation had not yet been established
- Financial co-investment (John's classified ad network): offering to co-invest in acquisitions signals genuine confidence in the deal's quality
- Trial periods with performance guarantees: if quality is truly high, the risk of offering a trial is small; if quality is low, the expected cost is high

### Remedy 2: Screening (Uninformed Party Designs Menu)

The uninformed party creates a menu of options designed so that different types self-select into different options. Requires satisfying participation constraints (PCs) and incentive compatibility constraints (ICCs). See screening-menu-design.md.

**Examples:**
- Insurance deductibles: high-deductible plans are chosen by low-risk individuals (they rarely claim, so a lower premium with more exposure is attractive). High-risk individuals prefer low-deductible plans (they claim frequently, so paying a higher premium for full coverage is worth it). Self-selection reveals risk type.
- Airline fare classes: restricted fares require advance purchase, non-refundability, and Saturday-night stay. These restrictions are cheap for leisure travelers (who plan ahead and stay over weekends) and expensive for business travelers (who need flexibility). Self-selection reveals price sensitivity.
- In-kind benefits (wheelchairs, not cash): benefits structured as specific goods that have high value only to genuine claimants are self-screening. The secondary market value is too low for false claimants to make fraud worthwhile.

### Remedy 3: Bureaucratic Friction as Screening Device

Requiring applicants to expend time and effort — filling out forms, waiting in offices, attending multiple appointments — is differentially costly across types:
- Healthy workers who can earn income will find the time too costly to spend
- Genuinely injured workers who cannot work can afford to spend the time

The apparent inefficiency of bureaucratic delay serves a real purpose: it screens false claimants without requiring direct verification of their claimed condition. Cash benefits create fraud incentives; in-kind benefits requiring effort reduce them.

**Important caveat:** This analysis does not justify bureaucratic inefficiency generally. It justifies specifically designed friction that is differentially costly between genuine and fraudulent claimants. Randomly slow processes that impose equal costs on all types are simply wasteful.

### Remedy 4: Positive Selection (Offer Design)

Design the offer so that its attractiveness is higher for the profitable type than for the unprofitable type — ideally making it attractive only to the desired type.

**Design principle:** Identify what the desired type uniquely needs or values. Structure the offer to deliver exactly that. Ensure the offer is unattractive or irrelevant to the undesired types.

**Testing the design:** Ask for each unwanted type: "Why would they NOT accept this offer?" If you cannot answer that question, the offer will attract them and adverse selection remains.

## Diagnosing Adverse Selection in Your Situation

Run through these questions:

1. **Do the transacting counterparts know something material about themselves that you cannot verify?** (Quality, risk, intention, capability)
2. **Does your current offer structure appeal more to the type that is costly to serve?** Check: who self-selects in vs. who you would want.
3. **Is the average quality of counterparts who accept your offer lower than the average quality of all potential counterparts?** Measure: claims rates, default rates, refund rates, churn rates, performance.
4. **Have you tried raising your standards (price, requirements) only to find the pool got worse, not better?** Classic adverse selection response.

If yes to 2 and 3: adverse selection is active. Proceed to remedy design.

FILE:references/bayes-rule-inference.md
# Bayes' Rule for Type Inference in Strategic Games

## Purpose

When a player observes an action taken by an opponent who is playing a mixed strategy (or in a semi-separating equilibrium), that action is informative but not conclusive. Bayes' rule is the mathematically correct procedure for updating your probability estimate of the opponent's type given what you observed.

## The Formula

**P(type T | action A) = P(A | type T) × P(type T) / P(A)**

Where:

**P(A)** = total probability of observing action A
= P(A | type T) × P(type T) + P(A | type not-T) × P(type not-T)

In words: the probability that the opponent is type T, given you observed action A, equals the fraction of all cases where action A occurs that are attributable to type T.

## Worked Example: Poker Bluffing

### Setup

Your rival plays as follows:
- With a **good hand**: raises 2/3 of the time, calls 1/3 of the time, folds never
- With a **poor hand**: raises 1/3 of the time, calls never, folds 2/3 of the time

Prior: you believe good and poor hands are equally likely (P(good) = 1/2, P(poor) = 1/2).

### Probability Table

| Action | P(action \| good hand) | P(action \| poor hand) |
|--------|----------------------|----------------------|
| Raise  | 2/3                  | 1/3                  |
| Call   | 1/3                  | 0                    |
| Fold   | 0                    | 2/3                  |

### Applying Bayes' Rule to "Raise"

**P(raise)** = P(raise | good) × P(good) + P(raise | poor) × P(poor)
= (2/3)(1/2) + (1/3)(1/2)
= 1/3 + 1/6 = **1/2**

**P(good | raise)** = P(raise | good) × P(good) / P(raise)
= (2/3 × 1/2) / (1/2)
= (1/3) / (1/2) = **2/3**

**Interpretation:** After observing a raise, update from 50% prior to 67% posterior. The raise is informative — it shifts the probability of a good hand upward — but it is not conclusive. There is still a 1 in 3 chance the raise is a bluff.

### Applying Bayes' Rule to "Fold" and "Call"

**Fold:**
P(fold) = (0)(1/2) + (2/3)(1/2) = 1/3
P(good | fold) = (0 × 1/2) / (1/3) = 0

Observing a fold gives certainty of a poor hand. Posterior = 0% good.

**Call:**
P(call) = (1/3)(1/2) + (0)(1/2) = 1/6
P(good | call) = (1/3 × 1/2) / (1/6) = (1/6) / (1/6) = 1

Observing a call gives certainty of a good hand. Posterior = 100% good.

### Summary Table

| Observed action | P(good hand) before | P(good hand) after | Inference |
|---|---|---|---|
| Raise | 50% | 67% | Informative but inconclusive |
| Call | 50% | 100% | Conclusive — good hand |
| Fold | 50% | 0% | Conclusive — poor hand |

## Semi-Separating Equilibrium and Bayes' Rule

In a semi-separating equilibrium, some wrong-type players mimic the signal and some do not. The signal is informative but not perfectly separating. Bayes' rule determines how much posterior belief should shift.

**Key inputs needed:**
1. Prior probability of each type (base rates)
2. Probability each type takes the observed action (the mixing probabilities)

**When cost differences are small:** Wrong-type players mix between signaling and not signaling. The equilibrium mixing probabilities are determined by the condition that the wrong type is indifferent between mimicking and not mimicking. The posterior belief after observing the signal is determined by Bayes' rule given those mixing probabilities.

## Practical Guidance

**When to apply Bayes' rule:**
- You observe an opponent's action and want to update your belief about their type
- The opponent is known to play mixed strategies (not a pure type-revealing signal)
- You are in a pooling equilibrium and want to know how much information you can extract despite the pool

**When Bayes' rule gives extreme results:**
- If only one type ever takes a particular action (P(action | wrong type) = 0), observing that action gives certainty about type. No calculation needed — the posterior collapses to 1.
- If both types take an action with equal probability, observing it gives no information. The posterior equals the prior.

**Common error:** Ignoring base rates. If 90% of the population are the wrong type and only 10% are the right type, even a highly informative signal (P(action | right type) = 0.9 vs. P(action | wrong type) = 0.1) still leaves substantial uncertainty. With equal base rates (50/50), the same signal likelihoods yield a strong posterior update.

## Business Applications

**Interpreting a competitor's price cut:** If competitors with high-cost structures cut price rarely (1/10 of the time) and low-cost competitors cut price frequently (7/10 of the time), and you have a 50/50 prior about which type of competitor you face:
P(low-cost | price cut) = (0.7 × 0.5) / [(0.7 × 0.5) + (0.1 × 0.5)] = 0.35 / 0.40 = **87.5%**

A price cut is strong evidence of a low-cost competitor. Adjust your response accordingly.

**Interpreting a candidate's salary history:** If candidates with strong outside options disclose salary 80% of the time and candidates with weak outside options disclose 20% of the time, observing disclosure raises the probability of a strong outside option significantly. Non-disclosure has the opposite implication (Sherlock Holmes: the dog that did not bark).

FILE:references/cost-difference-property.md
# Cost-Difference Property: Credibility Condition for Signals

## Core Statement

A signal is credible if and only if it is more costly for the wrong type to send than for the right type to send, and the cost difference exceeds the informational value (the benefit the wrong type would gain from being mistaken for the right type).

Formally, for a signal S:

**Credibility condition:** Cost(S | wrong type) − Cost(S | right type) > Value of being mistaken for right type

If this inequality fails, the wrong type will mimic the signal, the equilibrium collapses to pooling, and the signal conveys no information.

## Why Verbal Claims Fail This Test

A verbal claim "I am a high-quality seller" costs zero for both the true high-quality seller and the false claimant. The cost difference is zero. The false type will always make the same claim. Result: verbal claims are uninformative when interests are misaligned.

Exception: when interests are aligned (ordering a steak medium-rare, asking for directions), there is no reason to lie, and verbal communication is reliable. The G.H. Hardy principle — "If the Archbishop of Canterbury says he believes in God, that is just his job; if he says he doesn't, you can believe him" — captures this: statements that go against the speaker's interests are credible precisely because they are costly to make.

## Domain Examples

### Warranties (Product Quality)

**Signal:** Offering a warranty on a used car or consumer product.

**True type (high quality):** Repair costs expected to be low. Warranty obligation is cheap to honor. Net cost of offering warranty = small.

**False type (low quality):** Repair costs expected to be high. Warranty obligation is expensive to honor. Net cost of offering warranty = large.

**Cost difference:** Large. Credibility condition satisfied when the price premium from having a warranty exceeds the expected repair cost difference between types.

**Worked numbers:** If the warranty price premium is $800, expected repair cost for a good car is $500, and expected repair cost for a bad car is $2,000:
- Good car seller: net benefit of warranty = $800 − $500 = +$300
- Bad car seller: net benefit of warranty = $800 − $2,000 = −$1,200
- Bad car seller will not offer the warranty. Signal separates.

**Note:** The warranty itself must be enforceable. A private seller may not be able to credibly offer a warranty (no reputation, may move away) even if the car is genuinely good. A dealer with a long-term reputation can make the warranty credible. Credibility of the signal requires credibility of the commitment.

### Education as Signal (Spence Model)

**Signal:** MBA degree.

**True type (talented manager):** Certain to complete the degree. Cost = tuition + foregone salary = ~$200K. Expected payoff from MBA = wage premium of $40K/yr over 5 years = $200K. Break-even.

**False type (untalented manager):** 50% chance of completion. Expected cost = $200K × probability of attempt = $200K but with 50% chance of no return. Expected payoff from MBA = 50% × $40K/yr × 5 years = $100K. Net: −$100K in expectation. Not worth it.

**Cost difference:** Effective cost to false type is higher because of the failure risk. The signal works even though the dollar cost of tuition is the same, because the relevant cost includes the risk-adjusted cost of failure.

**Range for the signal to work:**
- Wage premium must be > $40K/yr (enough to attract true type)
- Wage premium must be < $130K/yr (low enough that false type's 50% chance is still not worth it)

**Informational externality:** The talented must invest $200K to distinguish themselves from the untalented. This cost would not exist in a world with no untalented applicants. The untalented impose a negative externality on the talented by their mere existence.

### Tattoo (Commitment Signal)

**Signal:** Tattoo with a partner's name.

**True type (committed person):** Small cost — a fitting tribute; the tattoo has positive sentimental value and no significant future cost.

**False type (non-committed person):** High cost — the tattoo is an embarrassing artifact for the next relationship; a permanent liability.

**Cost difference:** Large. The signal credibly separates committed from non-committed because only committed persons find it worthwhile.

**Key insight:** The signal does not need to be expensive in absolute terms; it needs to be differentially costly. Even a small absolute cost can serve as a credible signal if it is genuinely negative for the false type and genuinely neutral or positive for the true type.

### Mechanic Inspection Offer (Non-Credible Signal)

**Signal:** "You can have a mechanic inspect this car before buying."

**True type (good car):** Cost = zero. Mechanic finds nothing and confirms quality.

**False type (bad car):** Cost = zero. The owner is no worse off than before the offer — the buyer walks away if a defect is found, but the owner retains the car. There is no penalty for making this offer.

**Cost difference:** Zero. The signal is uninformative. Any owner — good or bad — can make this offer at no cost. Observing the offer reveals nothing.

**Practical lesson:** Offers to permit inspection are not credible signals. Credible commitments that guarantee compensation (warranties) are. Inspect the car anyway, but do not update your belief about quality based on the inspection offer alone.

### Gang Initiation Rites and Organizational Commitment Devices

**Signal:** Committing crimes, getting identifying tattoos, or participating in rituals as conditions of group membership.

**True type (genuine member):** Small cost — actions align with their values and future plans within the group.

**False type (undercover law enforcement or observer):** High cost — the actions create criminal liability, documented evidence, and personal moral barriers.

**Cost difference:** Large. Initiation requirements credibly screen out infiltrators because the false type's cost of compliance is prohibitively high.

**Counterpoint (bureaucratic delay as screening device):** Requiring claimants to spend hours in a waiting room to access benefits is costly for healthy workers (who forego earnings) and low-cost for injured workers (who cannot work). The apparent inefficiency of bureaucratic delay serves a genuine informational purpose. In-kind benefits (wheelchairs, not cash) serve similarly — the secondary market value of a wheelchair is low for someone who does not need one, making false claiming less attractive.

## Common Failure Modes

**Failure 1: Signal is cheap for both types.** Cleaning a car before sale is nearly costless for careless owners as well as careful ones. The signal is mimicked; pooling equilibrium.

**Failure 2: Signal cost is high for both types but differences are insufficient.** If the wage premium from an MBA exceeds $130K/yr, even untalented applicants find the 50% completion risk worthwhile. The screen collapses.

**Failure 3: Signaling becomes a rat race.** If the more able get a little more education, the less able mimic, so the more able need even more education. True abilities remain unchanged; the only beneficiaries are those providing the signal service (universities). Individual parties cannot stop this arms race unilaterally; public policy intervention is required.

**Failure 4: Enforcement of the signal fails.** A warranty offered by a private seller moving away is not credible even if the car is genuinely good. The signal's credibility depends on the ability to enforce its terms.

---
name: incentive-scheme-designer
description: "Design and diagnose incentive contracts for situations where effort is unobservable (moral hazard). Use this skill when a user needs to motivate a contractor, employee, co-founder, or agent whose actions cannot be directly monitored; when a user is deciding between a fixed salary, piece-rate, equity share, bonus, or fine structure; when a user needs to set the bonus level so that high-quality effort is in the agent's self-interest; when a user must satisfy both the participation constraint (agent accepts the deal) and the incentive compatibility constraint (agent exerts the desired effort); when a user wants to diagnose why an existing incentive scheme is failing — through sandbagging, gaming, effort diversion, or lack of effort; when a user is deciding between carrots and sticks and needs to understand when each is preferred; when a user suspects financial incentives are crowding out intrinsic motivation (Gneezy/Rustichini effect); when a user manages people performing multiple tasks and needs to know whether to bundle or separate them based on complementarity vs. substitutability; when a user needs to understand efficiency wages and when above-market pay is the cheapest way to deter shirking. This skill covers the full principal-agent problem after a contract is signed. It does NOT cover pre-contract adverse selection (who to hire) — use the information-asymmetry-strategist skill for that."
version: 1.0.0
homepage: https://github.com/bookforge-ai/bookforge-skills/tree/main/books/the-art-of-strategy/skills/incentive-scheme-designer
metadata: {"openclaw":{"emoji":"📚","homepage":"https://github.com/bookforge-ai/bookforge-skills"}}
status: draft
source-books:
  - id: the-art-of-strategy
    title: "The Art of Strategy"
    authors: ["Avinash K. Dixit", "Barry J. Nalebuff"]
    chapters: [13]
tags: [game-theory, incentive-design, moral-hazard, principal-agent, compensation]
depends-on: [information-asymmetry-strategist]
execution:
  tier: 1
  mode: full
  inputs:
    - type: document
      description: "Description of the principal-agent situation: what the agent is hired to do, what outcomes are observable, what effort is unobservable, what the market wage is, the cost of effort to the agent, and the probability that high vs. low effort produces a good outcome"
  tools-required: [Read, Write]
  tools-optional: []
  mcps-required: []
  environment: "Any agent environment; user describes the situation in text or structured form"
discovery:
  goal: "Design an incentive contract that induces the desired effort level while satisfying the agent's participation constraint, select the optimal scheme type (linear vs. nonlinear, carrot vs. stick), diagnose multi-task distortions, assess career concern substitutes, and identify when financial incentives will backfire"
  tasks:
    - "Establish the observable outcome, the unobservable effort levels, and the probability that each effort level produces a good outcome"
    - "Compute the minimum bonus needed to make high effort incentive-compatible: B >= cost_of_effort / (p_H - p_L)"
    - "Set base pay to satisfy the participation constraint: base = market_wage - p_H x B (may be negative — a fine)"
    - "Evaluate whether the fine/equity solution is feasible given legal constraints and agent capital"
    - "Compare linear (piece-rate) vs. nonlinear (quota-bonus) schemes: robustness vs. threshold power"
    - "Diagnose carrot vs. stick: equivalent mathematically, but sticks preferred with reliable monitoring; carrots preferred with imperfect monitoring or where punishment destroys participation"
    - "Assess multi-task structure: are tasks complements (bundle, use strong incentives for all) or substitutes (separate, equalize or weaken incentives)"
    - "Check career concern strength: early-career agents may need weaker monetary incentives; near-retirement agents need stronger ones"
    - "Flag intrinsic motivation risk: if the agent currently works for non-monetary reasons, introducing small financial incentives may destroy more motivation than they create"
    - "Design efficiency wage when direct performance measurement is noisy but termination threat is credible"
    - "Deliver structured contract recommendations with explicit numbers wherever inputs allow"
  audience: "Managers, founders, product leads, contract designers, compensation consultants, policy designers, and anyone structuring agreements where effort quality is hard to observe directly"
  when_to_use:
    - "User is designing a contract where they cannot observe how hard or well the other party works"
    - "User suspects an existing incentive scheme is being gamed, sandbagged, or producing effort on the wrong tasks"
    - "User is deciding how large a bonus, equity stake, or penalty to set"
    - "User needs to motivate someone who performs multiple tasks and is worried about effort being diverted away from less-measured tasks"
    - "User is considering whether to add financial incentives to a role currently governed by intrinsic motivation or professional norms"
  quality:
    correctness: null
    depth: null
    actionability: null
    specificity: null
---

# Incentive Scheme Designer

## When to Use

Use this skill when you have a principal-agent problem: you are hiring, contracting with, or managing someone whose effort quality matters to you but cannot be directly observed. You can only observe outcomes — and outcomes depend on effort plus luck. The goal is to design a payment scheme that makes the agent's self-interest align with your interests, without paying more than necessary.

The core challenge: **you cannot pay for effort because you cannot see it.** You can only pay for outcomes. But outcomes are imperfect proxies for effort. Tying pay too tightly to outcomes imposes risk on the agent (who may rationally demand a risk premium). Tying pay too loosely means the agent has little reason to work hard. Every incentive scheme is a tradeoff between incentive power and risk imposition.

This skill builds on the information-asymmetry-strategist framework. The difference: adverse selection is about who signs the contract (pre-contract hidden types). Moral hazard is about what they do after signing (post-contract hidden actions). The mechanisms for dealing with them overlap — both use participation constraints and incentive compatibility — but the design logic is different.

This skill does NOT apply to:
- Selecting among candidates of unknown quality (pre-contract adverse selection — use information-asymmetry-strategist)
- Symmetric-information situations where effort is directly verifiable
- Pure negotiation over price without ongoing effort (use the negotiation skill)

---

## Context and Input Gathering

### Required (ask if missing)

- **What is the observable outcome?** What measurable result can you tie pay to?
  -> Ask: "What can you actually observe at the end — revenue, errors found, project success or failure?"

- **What are the effort levels?** What is the difference between routine/low effort and the desired high effort?
  -> Ask: "What does the agent do differently when working hard vs. coasting? What does this cost them — in stress, hours, skill, or discomfort?"

- **What is the probability difference?** How much does high effort raise the chance of a good outcome vs. low effort?
  -> Ask: "If they work hard, what is the probability of success? If they coast, what is the probability?"

- **What is the market wage?** What could the agent earn elsewhere at the same effort level?
  -> Ask: "What is their outside option — what would they earn doing a comparable job?"

- **What is the agent's cost of effort?** The subjective cost in money terms of exerting high rather than low effort.
  -> Ask: "What pay increment would the agent require to voluntarily choose high effort over low effort on their own, with no monitoring?"

### Useful (gather if present)

- Whether fines or negative base pay are legally and practically enforceable
- Whether the agent has capital to invest (enabling equity sharing or fine solutions)
- How many tasks the agent performs simultaneously and whether they are substitutes or complements
- How early or late the agent is in their career (career concern strength)
- Whether the role currently attracts intrinsically motivated people or mercenaries
- Whether the agent will be monitored repeatedly over time (repeated relationship effects)

---

## Execution

### Step 1 — Establish the Probability Structure

**Why:** Everything else in incentive design depends on this. The bonus formula, the base pay, and the efficiency of the scheme all flow from the probability difference (p_H - p_L). If this gap is large, a small bonus can induce high effort. If the gap is small (luck dominates), you need a very large bonus to matter — which imposes high risk on the agent and is expensive.

**Define:**
- p_H = probability of good outcome under high effort
- p_L = probability of good outcome under low effort
- V = value to you of a good outcome (vs. bad outcome)

**Critical check — does effort actually matter?**

If p_H - p_L is small (say, 0.05), outcome is mostly noise. Outcome-based pay will be weakly linked to effort and impose high risk on the agent for little incentive gain. In this case:
- Use fixed salary or efficiency wages instead of performance pay
- Invest in improving monitoring if possible
- Accept that strong incentives may not be feasible

If p_H - p_L is large (say, 0.30 or more), outcome-based pay is powerful and relatively efficient.

---

### Step 2 — Compute the Minimum Incentive Bonus

**Why:** The bonus must be large enough that, in the agent's own calculation, the expected gain from high effort exceeds its cost. Setting it below this threshold gives the agent no rational reason to exert high effort. Setting it far above this threshold overpays for incentive power you do not need — or imposes unnecessary risk.

**The bonus formula:**

```
B >= cost_of_effort / (p_H - p_L)
```

Where B is the bonus paid for a good outcome (beyond the base pay).

**Derivation logic:** The agent chooses high effort if and only if the expected pay gain from doing so covers the cost. The expected gain from high effort = (p_H - p_L) x B. Setting this equal to cost_of_effort gives the minimum B.

**Worked example (Wizard 1.0 programmer):**
- High effort: p_H = 0.80, cost of effort above market = $20,000
- Low effort: p_L = 0.60
- Minimum bonus: B >= $20,000 / (0.80 - 0.60) = $20,000 / 0.20 = **$100,000**

The bonus for success must be at least $100,000 to make high effort worthwhile.

**Worked example (proofreader):**
- Effort cost: $X (subjective cost of careful reading)
- Probability of finding errors with high effort: p_H
- Probability of finding errors with low effort: p_L
- The piece rate per error found is the continuous analog: the per-error payment must cover the marginal effort cost per marginal error found

---

### Step 3 — Set Base Pay to Satisfy the Participation Constraint

**Why:** The agent must be willing to take the job. If the expected payment under the optimal incentive scheme falls below the market wage, the agent will not accept. You must therefore set base pay so that expected total compensation equals the market wage. This may require negative base pay (a fine) — meaning the agent pays you if the bad outcome occurs.

**The participation constraint:**

```
p_H x (base + B) + (1 - p_H) x base >= market_wage
=> base + p_H x B >= market_wage
=> base >= market_wage - p_H x B
```

Simplified (binding participation constraint at minimum cost to you):

```
base = market_wage - p_H x B
```

Note: base pay can be negative. A negative base pay is a fine paid by the agent in the event of failure. This is mathematically equivalent to equity sharing.

**Worked example (Wizard 1.0):**
- B = $100,000, p_H = 0.80, market wage = $70,000
- base = $70,000 - (0.80 x $100,000) = $70,000 - $80,000 = **-$10,000**

The agent pays a $10,000 fine if the project fails, and receives $90,000 if it succeeds. Expected compensation: (0.80 x $90,000) + (0.20 x -$10,000) = $72,000 - $2,000 = $70,000. Participation satisfied exactly.

**When negative base pay is not feasible:**

If fines are legally unenforceable or the agent lacks capital, the minimum effective bonus is $100,000 on success and $0 on failure. The agent's expected compensation rises to $80,000 (0.80 x $100,000), exceeding the $70,000 market wage. You pay a $10,000 "feasibility premium." This is the cost of moral hazard when the first-best fine solution is unavailable.

---

### Step 4 — Evaluate Scheme Type: Linear vs. Nonlinear

**Why:** The shape of the incentive scheme determines both its power and its vulnerability to gaming. Linear schemes (piece-rate) are robust but apply uniform incentive pressure everywhere on the performance scale. Nonlinear schemes (quota-bonus) concentrate incentive power near the threshold — which is powerful when the threshold is correctly set but creates perverse incentives both below and above it.

**Linear (piece-rate) schemes:**
- Pay proportional to output: every unit of performance earns the same increment
- Strengths: robust to changing circumstances; no inflection points to exploit; agent faces constant marginal incentive
- Weaknesses: may provide insufficient incentive at any particular level; salary cost scales linearly with output
- Best for: ongoing work where effort is continuous and circumstances change (annual sales, error detection)

**Nonlinear (quota-bonus) schemes:**
- Pay a low fixed amount below a threshold; pay a high fixed amount above it
- Strengths: enormous incentive power near the threshold; the spread can be very large without high average cost
- Weaknesses: zero incentive power far from the threshold; invites three failure modes:

| Failure mode | Mechanism | Example |
|---|---|---|
| Below-threshold disengagement | Agent far below quota stops trying; reaching quota is hopeless | Salesperson with bad Q1 coasts through Q2 |
| Above-threshold sandbagging | Agent who hits quota early stops; no reward for exceeding it | Salesperson meets June quota; holds orders until next year |
| Inter-period manipulation | Agent shifts work across periods to hit thresholds optimally | Enron booking false revenues; agent delays customer orders |

**Hybrid approach (in practice):** Combine a base commission rate (linear) with quota bonuses at 100%, 150%, and 200% of target. The linear commission ensures constant baseline incentives; the quota bonuses provide threshold bursts. This captures quota power while limiting the flat-incentive zones.

**Decision rule:**
- Stable task, well-understood probability structure → nonlinear (quota) fine-tuned to the right threshold
- Changing circumstances, multi-period work, or risk of threshold manipulation → linear or hybrid
- Circumstances changed after contract set (making quota unreachable or already met) → linear component prevents incentive collapse

---

### Step 5 — Choose: Carrot or Stick

**Why:** Any incentive spread can be structured as a bonus above the base (carrot) or a fine below the base (stick). The mathematics are identical — what changes is the reference point, the agent's likely behavioral response, and the feasibility constraints. Choosing the wrong framing wastes the incentive power you are paying for.

**Mathematical equivalence:**

Both schemes below have average payment = 100 and spread = 50:
- Carrot: base = 99, bonus = 50 if exceptional performance (2% chance with desired effort) → expected = 99 + (0.02 x 50) = 100
- Stick: base = 101, fine = 51 if exceptionally poor performance (2% chance with desired effort) → expected = 101 - (0.02 x 51) = 100

Same incentive power, same expected cost. The difference is behavioral and practical.

**When to use sticks:**
- Monitoring is reliable: the failure event (shirking coming to light) is observable with reasonable accuracy
- The agent has assets or stakes that can be forfeited
- The agent has outside alternatives that are poor (the threat of job loss plus fine is credible and powerful)
- Example: Stalin's approach would have worked if punishment were reliably tied to effort rather than arbitrary. It failed not because sticks don't work, but because the punishment wasn't tied to actual shirking — people were punished whether they worked hard or not.

**When to use carrots:**
- Monitoring is imperfect: false positives (punishing diligent workers) damage trust and destroy participation willingness
- Legal constraints prohibit negative base pay or fines on employees
- The agent has strong outside options (they will simply leave rather than accept punishment risk)
- Workforce morale is important: being fired or fined publicly destroys motivation for the broader team
- Example: The Dixit proofreading contract — $600 base plus $1/error found. A pure fine structure (pay nothing without errors; fine for missed errors) would cause the student to reject the work.

**Efficiency wages as a special case of the stick:**

When you cannot measure performance well enough to set a bonus formula but can detect gross shirking occasionally, set the wage above market by enough that the threat of losing it deters shirking:

```
X > cost_of_effort / (detection_probability x (1 + 1/interest_rate))
```

The efficiency premium X must exceed the one-time gain from shirking (cost saved) discounted by the probability of detection and the present value of the perpetual wage premium at risk. This works when:
- Shirking is detectable with meaningful probability (even if not always)
- The relationship is ongoing (the premium is valuable precisely because it recurs)
- Firing and blacklisting is credible (word spreads; bad reputation follows the agent)

---

### Step 6 — Handle Multiple Tasks

**Why:** Most real agents perform multiple tasks simultaneously. Adding strong incentives to one task can either help or hurt performance on others, depending on whether the tasks are substitutes or complements. Ignoring this is one of the most common sources of incentive scheme failure.

**Substitutes:** Effort on task A reduces the marginal productivity of effort on task B (because they share a limited effort budget, compete for time, or are cognitively exhausting in the same way).
- Example: A farmhand in corn and dairy. More corn time = more tired = less productive in dairy.
- Example: Teaching and research at a university that keeps them separate by design (French model).
- Example: A customer service rep measured on call speed who also does quality follow-up — strong speed incentives divert effort from quality.

**Complements:** Effort on task A raises the marginal productivity of effort on task B.
- Example: A beekeeper who also tends apple orchards. More bee-keeping makes orchard more productive through pollination.
- Example: Research and teaching at a US research university — research deepens teaching; teaching sharpens research questions.
- Example: A salesperson who does both prospecting and closing — better prospecting leads to better clients who are easier to close.

**Design rules by task relationship:**

| Relationship | Rule | Reason |
|---|---|---|
| Substitutes | Equalize incentive strength across tasks; use weaker incentives for both | Strong incentive on A diverts effort from B; net effect may be negative |
| Complements | Use strong incentives for all tasks | Effort on A amplifies B; synergies compound; no diversion problem |
| Unknown | Start with equal moderate incentives; observe for diversion patterns | Asymmetric incentives are risky when relationship is unclear |

**Organizational design implication:** Group complementary tasks together under one person (or division) and use strong incentives. Separate substitute tasks into different people or divisions with independent incentive schemes. The failure to follow this — mixing substitute tasks with shared incentives — is a structural source of incentive weakness.

**Heathrow Airport example:** Check-in, security, shopping, and boarding are all complements — they form one end-to-end process. Splitting them across BAA, police, and a regulator with opposing incentives (BAA profits from shops; regulator prices landing fees to reduce congestion) created predictable dysfunction. Each owner's incentives partially cancel the others'.

**Multiple owners / bosses:** The incentive strength is inversely proportional to the number of owners with conflicting objectives. When agent answers to N bosses with opposed interests, each boss's incentive partially cancels the others'. In the extreme (fully opposed principals), incentive strength approaches zero. This explains incentive weakness in public sector agencies, international bodies, and joint ventures with misaligned partners.

---

### Step 7 — Account for Career Concerns and Repeated Relationships

**Why:** Financial incentives are not the only incentive mechanism. Early-career agents are often powerfully motivated by reputation, promotion prospects, and future earnings — mechanisms that substitute for or supplement direct monetary incentives. Ignoring career concerns leads to over-paying for monetary incentives early in a career and under-paying for them late.

**Career concerns:**
- Strong when: agent is early-career, relationship is ongoing, agent has future earning potential in this or related fields
- Weak when: agent is near retirement, agent plans to leave the field, relationship is one-shot
- Effect: career concerns substitute for monetary incentives early in career. A junior employee working on a visible project has strong incentive to perform because the outcome is part of their professional record, even with a modest salary bonus.
- Implication: reduce monetary incentive intensity for early-career employees in visible roles; increase it near retirement or for contractors with no ongoing relationship

**Repeated relationships:**
- When the same agent works on multiple projects over time, each outcome gives you more information about their underlying effort level
- By the law of large numbers, average output over many projects is a more accurate indicator of average effort than any single outcome
- This allows stronger incentives over time: persistent poor outcomes become attributable to effort rather than luck
- Also: the employer can credibly threaten to "believe the bad luck story once" but not repeatedly — this itself disciplines the agent

**Design implication:** An incentive scheme that looks too weak for a one-shot interaction may be adequate for a long-term relationship where career concerns and repeated-game reputation effects provide supplementary discipline.

---

### Step 8 — Check for Intrinsic Motivation Risk

**Why:** Adding financial incentives to a role currently governed by intrinsic motivation can backfire. When money enters the picture, it becomes the primary frame for evaluating the task. A small payment that is insulting relative to effort converts a volunteer activity into a poorly-paid job. The result can be performance worse than no incentive at all (the Gneezy/Rustichini finding).

**The Gneezy/Rustichini experiment:**
- Group 1: no payment → 28 correct answers on average (intrinsic motivation intact)
- Group 2: 3 cents per correct answer → 23 correct answers (worst performance — money frame introduced, amount too small)
- Group 3: 30 cents per correct answer → 34 correct answers (incentive large enough to dominate)
- Group 4: 90 cents per correct answer → 34 correct answers (same)

**Key finding:** Small payments destroy intrinsic motivation and produce the worst outcome. The prescription: pay significantly or do not pay at all. There is no safe small-payment middle ground when intrinsic motivation is present.

**Signals that intrinsic motivation may be present:**
- Agents voluntarily seek out the work (academics, doctors, nonprofit workers, open-source contributors)
- Current compensation is below market wage, yet quality is high
- Workers describe their role in mission-oriented rather than transactional terms
- Quality is highest when financial monitoring is lowest

**Decision rule:**
- Strong intrinsic motivation + you cannot offer a substantial financial incentive → do not introduce financial incentives; use non-monetary recognition, career concerns, and mission framing
- Strong intrinsic motivation + you can offer a substantial financial incentive → proceed, but monitor for crowding out (e.g., quality may fall on less-measured dimensions)
- Weak or absent intrinsic motivation → financial incentives are the primary tool; calibrate using the bonus formula in Step 2

---

### Step 9 — Deliver Structured Contract Recommendations

Structure your output as:

**Scheme type:** [Fixed salary / Linear piece-rate / Quota-bonus / Equity share / Efficiency wage — and why given the situation]

**Bonus calculation:**
- p_H = [value], p_L = [value], cost_of_effort = [value]
- Minimum bonus: B = cost_of_effort / (p_H - p_L) = [value]

**Base pay calculation:**
- market_wage = [value]
- base = market_wage - p_H x B = [value]
- If negative and infeasible: constrained solution = [0 base + B on success only]; feasibility premium = [value]

**Carrot vs. stick recommendation:** [With rationale: monitoring reliability, legal constraints, agent alternatives]

**Multi-task assessment:** [Are tasks substitutes or complements? Are incentives equalized or differentiated? Should tasks be bundled or separated?]

**Career concern adjustment:** [Is the agent early-career? Should monetary incentive be reduced accordingly?]

**Intrinsic motivation check:** [Is there risk of crowding out? Is the payment level above the threshold to avoid the Gneezy/Rustichini trap?]

**Failure modes to watch for:** [Sandbagging, gaming, effort diversion, threshold exploitation — specific to this scheme]

**Your expected profit:** [Revenue - average compensation, with and without moral hazard constraint]

---

## Key Principles

**Pay for outcomes, not effort — because you can only observe outcomes.** The fundamental constraint is observability. Build incentive schemes around what you can actually measure. Do not assume you will be able to infer effort from outcomes when noise is high.

**The bonus formula is the anchor.** B = cost_of_effort / (p_H - p_L) is the minimum bonus to induce high effort. Everything else — base pay, scheme shape, carrot vs. stick — is calibration around this anchor. Compute it first.

**Participation and incentive compatibility are both binding.** A scheme that induces effort but pays below market wage will be rejected. A scheme that pays market wage but provides no effort incentive will lead to shirking. Both constraints must be satisfied simultaneously.

**Carrots and sticks are mathematically equivalent; choose based on monitoring reliability and legal context.** Stalin's economy failed not because sticks cannot work, but because punishment was not reliably tied to shirking. With arbitrary punishment, the incentive disappears — workers are penalized whether they work hard or not.

**Multiple tasks destroy incentives unless tasks are complements.** Any time you measure one task well and another poorly, effort flows to the measured task. Fix this by grouping complementary tasks together, giving substitute tasks to different people, and equalizing incentive strength across tasks you cannot avoid bundling.

**Small payments are worse than no payments when intrinsic motivation is present.** Introduce financial incentives only if you can make them substantial. The threshold is not zero — it is large enough that the monetary frame dominates rather than merely corrupts the intrinsic one.

**Efficiency wages work when termination is credible and the relationship is ongoing.** The wage premium deters shirking by making the job worth keeping. The threat is only credible if you will actually fire; the deterrence is only valuable if the relationship is long-term.

---

## Examples

### Example 1: Real Estate Agent Commission Problem

**Situation:** A real estate agent earns a 6% commission on house sales. The seller is asking $500,000.

**The alignment problem:** On a $20,000 price increase (e.g., negotiating harder or waiting for a better offer), the agent gains only 6% x $20,000 = $1,200. But the agent's opportunity cost — the time spent on this house rather than moving to the next listing — is far higher. The agent's optimal strategy is to close quickly at a good-enough price, not maximize price for the seller.

**Why 6% linear commission fails:** The agent's stake in the marginal price improvement ($1,200) is far smaller than the seller's stake ($20,000 - $1,200 = $18,800 net). The incentives are structurally misaligned. More commission rate helps, but the agent would need nearly 100% commission to fully align interests.

**Better scheme:** Progressive commission that increases sharply above a reserve price (e.g., 6% up to $500K, then 30% on every dollar above $500K). This concentrates the agent's incentive exactly where the seller wants effort — on maximizing the price above the baseline.

**Failure mode to avoid:** A poorly set reserve price creates the same quota problem as any nonlinear scheme — if $500K is too high and the market won't support it, the agent disengages.

---

### Example 2: Software Programmer Equity Sharing (Wizard 1.0)

**Situation:** You want to develop a chess game (Wizard 1.0). Value = $200,000 if successful. p_H = 0.80, p_L = 0.60. Market wage for routine effort: $50,000. You want high-quality effort (cost = $20,000 above market).

**Step 2 — Minimum bonus:** B >= $20,000 / (0.80 - 0.60) = **$100,000**

**Step 3 — Base pay:** base = $70,000 - (0.80 x $100,000) = **-$10,000** (fine if project fails)

**Fine/equity scheme:** Pay $90,000 on success, -$10,000 (fine) on failure. Average pay = $70,000. Your average profit = $160,000 - $70,000 = $90,000.

**Constrained equity scheme (if fine is unenforceable):** Give programmer 50% equity (worth $100,000 on success, $0 on failure) in exchange for labor only. Average pay = $80,000. Your average profit = $80,000. You pay a $10,000 feasibility premium for moral hazard.

**Risk premium issue:** If the programmer is risk-averse, she values the $100,000 gamble at less than its $80,000 expected value. She needs additional compensation for bearing risk. The optimal solution is a compromise: less than full incentive, some fixed base to absorb risk — but this reduces incentive power and requires accepting some effort below the ideal.

---

### Example 3: Multi-Task Failure — Teaching vs. Research

**Situation:** A professor teaches and conducts research. University wants high effort on both.

**Tasks:** Teaching and research are complements in US research universities — better research informs better teaching; regular teaching keeps researchers grounded. They share time and attention but each makes the other more productive.

**Correct incentive design:** Use strong incentives for both: tenure and promotion contingent on both teaching evaluations and publication record. Bundle the tasks; apply strong incentives across the bundle.

**Incorrect design (French model):** Separate research into specialized institutes, teaching into pure teaching universities. Treats them as substitutes. Results in weaker incentives (research institutes have no teaching to cross-pollinate; teaching universities have no research to inform). US model comparatively succeeds because the complement structure is respected.

**Test:** If you observe a professor excelling at research and neglecting teaching, the tasks may actually be substitutes for that individual (research crowds out teaching energy). The fix is equalization: weight both dimensions more equally in the incentive scheme rather than rewarding only research publications.

---

## References

- `references/bonus-formula-derivation.md` — Full derivation of B >= cost / (p_H - p_L), participation constraint algebra, worked numerical examples, equity-share equivalence
- `references/carrot-stick-analysis.md` — Mathematical equivalence proof, behavioral differences, Stalin case study, CEO compensation analysis, monitoring reliability as the selection criterion
- `references/nonlinear-schemes-and-gaming.md` — Quota-bonus mechanics, sandbagging patterns, Enron-style manipulation, hybrid linear-nonlinear design, real estate commission analysis
- `references/multi-task-incentive-design.md` — Substitute vs. complement taxonomy, Heathrow Airport organizational failure, research-teaching complementarity, multiple-owners incentive dilution formula
- `references/efficiency-wages-and-intrinsic-motivation.md` — Efficiency wage formula derivation, Gneezy/Rustichini experiment results, intrinsic motivation crowding out, career concern substitution table

## License

This skill is licensed under [CC-BY-SA-4.0](https://creativecommons.org/licenses/by-sa/4.0/).
Source: [BookForge](https://github.com/bookforge-ai/bookforge-skills) — The Art of Strategy by Avinash K. Dixit, Barry J. Nalebuff.

## Related BookForge Skills

Install related skills from ClawhHub:
- `clawhub install bookforge-information-asymmetry-strategist`

Or install the full book set from GitHub: [bookforge-skills](https://github.com/bookforge-ai/bookforge-skills)

FILE:references/bonus-formula-derivation.md
# Bonus Formula Derivation

## Setup

A principal (employer) hires an agent (worker) to perform a task. The agent can exert high effort or low effort. The principal cannot observe effort; only the outcome is observable.

**Variables:**
- p_H = probability of good outcome under high effort
- p_L = probability of good outcome under low effort
- p_H > p_L (high effort genuinely improves the probability)
- C = agent's cost of high effort (in monetary equivalent)
- W = market wage (agent's outside option)
- B = bonus paid for good outcome (payment increment above base)
- base = base pay (paid regardless of outcome)

## Incentive Compatibility Constraint

The agent chooses high effort if and only if expected pay from high effort >= expected pay from low effort:

```
p_H x (base + B) + (1 - p_H) x base >= p_L x (base + B) + (1 - p_L) x base + C

=> p_H x B >= p_L x B + C

=> (p_H - p_L) x B >= C

=> B >= C / (p_H - p_L)
```

The minimum bonus to induce high effort is: **B* = C / (p_H - p_L)**

**Intuition:** The expected bonus gain from switching to high effort is (p_H - p_L) x B. This must exceed the cost C of doing so. The smaller the probability gap, the larger B must be to compensate.

## Participation Constraint

The agent accepts the contract only if expected pay >= market wage:

Under high effort (which we want to induce):

```
p_H x (base + B) + (1 - p_H) x base >= W

=> base + p_H x B >= W

=> base >= W - p_H x B
```

The minimum base pay (binding participation constraint):

**base* = W - p_H x B**

With B = B* = C / (p_H - p_L):

```
base* = W - p_H x C / (p_H - p_L)
```

## When Base Pay Is Negative

base* is negative when p_H x B > W, i.e., when the expected bonus payment already exceeds the market wage. The agent must pay the principal a fine in the bad outcome. This is a fine structure.

**Example (Wizard 1.0):**
- p_H = 0.80, p_L = 0.60, C = $20,000, W = $70,000
- B* = $20,000 / 0.20 = $100,000
- base* = $70,000 - 0.80 x $100,000 = $70,000 - $80,000 = -$10,000

Payment structure: +$90,000 on success, -$10,000 on failure.

Expected pay = (0.80 x $90,000) + (0.20 x -$10,000) = $72,000 - $2,000 = $70,000 = W. Exactly satisfies participation.

## Equity-Share Equivalence

In the fine/bonus scheme, the programmer effectively owns a fraction of the firm:
- Pay $10,000 to acquire 50% stake (firm value: $200,000 on success, $0 on failure)
- Her net: -$10,000 + 50% x $200,000 = $90,000 on success; -$10,000 on failure

Equity share s that induces high effort:

```
(p_H - p_L) x s x V >= C

=> s >= C / [(p_H - p_L) x V]
```

In this example: s >= $20,000 / (0.20 x $200,000) = 0.50 = 50%

Both formulations are equivalent. Fine/bonus = equity sharing at the right fraction.

## Constrained Solution (No Fines)

If fine is unenforceable (legal constraint) or agent has no capital:

- base = 0, B = $100,000 (only paid on success)
- Expected pay = 0.80 x $100,000 = $80,000 > W = $70,000

Agent earns a $10,000 feasibility premium. Principal's average profit = $160,000 - $80,000 = $80,000 vs. $90,000 with fine solution. The $10,000 gap is the cost of moral hazard under the capital constraint.

## Risk Premium Complication

If the agent is risk-averse, they value the $100,000 bonus at less than its expected value ($80,000). They require additional compensation for bearing risk. The optimal solution is a compromise:
- Lower B (below $100,000) to reduce risk imposed on agent
- Higher base to compensate, accepting that this weakens effort incentive
- Result: effort somewhere between low and high; principal accepts some incentive loss to reduce risk premium cost

The more noise in the outcome (smaller p_H - p_L), the larger B must be to incentivize effort, and the more risk is imposed on a risk-averse agent. This is why high-noise environments tend toward fixed salaries.

FILE:references/carrot-stick-analysis.md
# Carrot vs. Stick Analysis

## Mathematical Equivalence

Any incentive spread can be framed as either a bonus (carrot) or a fine (stick). The two are economically equivalent when:
- The spread (difference between good and bad outcome payments) is the same
- The average payment (weighted by probabilities) is the same

**Example with spread = 50, average = 100, and 2% chance of exceptional event:**

Carrot scheme:
- Normal outcome: 99
- Exceptional good performance: 149 (probability 2%)
- Expected pay: 99 + (0.02 x 50) = 100

Stick scheme:
- Normal outcome: 101
- Exceptionally poor performance: 50 (probability 2%)
- Expected pay: 101 - (0.02 x 51) = 100

Both schemes have identical average payment (100) and identical incentive spread (50). They are mathematically identical from the standpoint of expected value.

## Why They Differ in Practice

**1. Reference point effects**

Behavioral economics shows agents respond differently to losses than gains of equal size. A fine of $10,000 is experienced as more painful than missing a $10,000 bonus, even though the financial outcome is identical. Sticks may therefore provide stronger behavioral incentive per dollar of expected value — but also create stronger resistance, anxiety, and resentment.

**2. Monitoring reliability**

Sticks require reliable detection of bad outcomes. If bad outcomes are sometimes falsely triggered (good workers occasionally punished), the stick:
- Punishes workers who do not deserve it (destroys trust)
- Weakens the link between effort and outcome (reduces incentive value)
- Causes workers to leave rather than accept arbitrary punishment risk

**Stalin's failure:** The Soviet punishment scheme was structurally a stick — work hard or go to Siberia. It would have been powerful if punishment were reliably tied to shirking. It failed because people were punished whether they worked hard or not (arbitrary purges, quota-meeting regardless of effort). When punishment is decoupled from effort, the incentive disappears entirely. Workers asked: "Why work hard if I might get punished anyway?"

**3. Legal and contractual constraints**

Employment law in most jurisdictions prohibits negative base pay and involuntary wage deductions for performance failures. Sticks are therefore frequently infeasible for employees. They are more available for:
- Contractors and vendors (performance bonds, liquidated damages)
- Equity holders (dilution)
- Executives with clawback provisions

**4. Agent wealth and capital**

A fine works only if the agent can pay it. If the agent lacks assets, fines are not credible — a bankrupt agent cannot be fined. This is why the constrained equity-sharing solution (no fine, bonus only) arises when the agent has no capital.

**5. Outside alternatives**

When the agent has strong outside options, the stick may cause them to simply leave rather than accept punishment risk. The effective threat of the stick depends on the gap between current wage and outside option. This is why above-market pay (efficiency wages) is a prerequisite for stick strategies to function.

## CEO Compensation as Carrot Structure

Top executive compensation is structured almost entirely as a carrot:
- Strong performance: very large bonus, stock options, accelerated vesting
- Average performance: moderately large salary plus some bonus
- Poor performance: "golden parachute" — still significant severance

The spread is enormous (billions in difference between great and poor outcomes), but the average is far above what would be needed to meet any reasonable participation constraint. The excess reflects competition for CEO candidates: the participation constraint is not "driving a cab" but "the package at Company X." European companies pay less and still attract capable CEOs because the European alternative is other European packages, not US ones.

## Efficiency Wages: Stick Built on Ongoing Relationship

The efficiency wage is a stick structure where the punishment is job loss rather than a fine. It works as follows:

**Setup:**
- Market wage: $40,000 (dead-end jobs requiring no special effort)
- Job value to employer: $60,000/year
- Cost of effort to worker: $8,000/year
- Probability of detecting shirking in any year: 25%
- Interest rate: 10%

**Efficiency premium calculation:**

The worker gains $8,000 from shirking (effort cost saved). They risk losing the premium X every year thereafter. If detected, lose X annually forever; present value of that loss at 10% interest rate = 10X.

No-shirk condition: $8,000 < 0.25 x 10X → X > $3,200

Minimum efficiency wage: $40,000 (market) + $8,000 (effort premium) + $3,200 (efficiency premium) = $51,200

**Why above $48,000 and not just $48,000?**

$48,000 = market wage + effort cost. At exactly $48,000, the worker is indifferent between the job and the outside option even while working hard. The extra $3,200 is the efficiency premium — it creates a cushion that makes shirking not worth the risk of losing the job.

The threat only works if:
- Detection probability is meaningful (not zero)
- The worker will be fired (not just warned)
- The word will spread (blacklisting is credible)
- The relationship is ongoing (the premium recurs; losing it matters)

FILE:references/efficiency-wages-and-intrinsic-motivation.md
# Efficiency Wages and Intrinsic Motivation

## Efficiency Wages

### The Problem They Solve

Sometimes performance is not measurable continuously, but gross failure (shirking being discovered) occurs with a detectable probability. In this context, a performance-based bonus formula is not available — you cannot set B because there is no continuous outcome metric. But you can still deter shirking through the threat of job loss.

An efficiency wage is above-market pay that creates a valuable job worth keeping, combined with a credible threat to fire (and blacklist) any worker caught shirking. The worker shirks only if the one-time gain from shirking outweighs the expected present value of losing the wage premium.

### The Efficiency Wage Formula

**Variables:**
- W_eff = efficiency wage
- W_m = market wage (outside option)
- C = cost of effort (one-time gain from shirking = C saved)
- q = probability of detecting shirking in any period
- r = interest rate (for discounting future income streams)

**No-shirk condition:**

The worker shirks if:
```
C (gain from shirking) > q x PV(W_eff - W_m) (risk of losing premium)
```

Present value of losing the premium forever at interest rate r:
```
PV(W_eff - W_m) = (W_eff - W_m) / r
```

Wait — but the worker faces detection risk each year. Annual premium at risk: W_eff - W_m.
Present value of permanent stream: (W_eff - W_m) / r

No-shirk condition:
```
C < q x (W_eff - W_m) / r
=> W_eff - W_m > C x r / q
=> W_eff > W_m + C x r / q
```

**Worked example (text):**
- W_m = $40,000, C = $8,000, q = 0.25, r = 0.10
- Required efficiency premium: $8,000 x 0.10 / 0.25 = $3,200
- Efficiency wage: $40,000 (market) + $8,000 (effort cost premium) + $3,200 = $51,200

Verification: One-time gain from shirking = $8,000. Risk = 0.25 chance of losing $3,200/year forever; PV = $3,200/0.10 = $32,000; expected loss = 0.25 x $32,000 = $8,000. Exactly indifferent at $51,200. Any higher efficiency wage makes shirking strictly worse.

### When Efficiency Wages Work

**Required conditions:**

1. **Detectable failure:** Shirking must come to light with meaningful probability. If q ≈ 0, the premium required becomes infinite (no finite wage deters shirking when detection is impossible).

2. **Credible termination:** You must actually fire workers caught shirking. If workers know termination is unlikely (political protection, union rules, social norms), the threat loses credibility.

3. **Credible blacklisting:** The premium is only valuable if the fired worker cannot immediately get an equally good job elsewhere. "Spreading the word" among employers makes the blacklist credible.

4. **Ongoing relationship:** The premium is valuable because it recurs. A one-shot contract cannot use this mechanism effectively.

### Efficiency Wages in Daily Life

The principle extends beyond formal employment:
- **Regular mechanic:** Paying slightly more than the lowest rate makes the relationship worth preserving for the mechanic. Cheating (overcharging, doing unnecessary work) risks losing a steady client. The premium is for honesty, not efficiency.
- **Long-term contractors:** Vendors who depend significantly on your business have strong incentive to maintain quality — losing the account is costly.
- **Author-publisher relationships:** Publishers who pay higher advances have more to lose from a book's failure and are more invested in its success — and authors are more motivated to deliver.

## Intrinsic Motivation and Crowding Out

### The Gneezy/Rustichini Experiment

Subjects were given 50 questions from an IQ test. Four groups:

| Group | Incentive | Average correct |
|---|---|---|
| 1 | None — "do your best" | 28 |
| 2 | 3 cents per correct answer | 23 |
| 3 | 30 cents per correct answer | 34 |
| 4 | 90 cents per correct answer | 34 |

**Key findings:**
- Groups 3 and 4 outperformed the no-incentive group (financial incentive works when large enough)
- Group 2 performed worst of all — worse than no incentive
- Small payments destroyed intrinsic motivation without replacing it with sufficient extrinsic motivation

**Mechanism:** Introducing money changes the frame from "I am doing this because it matters / is interesting" to "I am doing this for pay." At 3 cents, the "pay" frame is active but the pay itself is insulting — it signals the task is not worth real money. At 30 cents, the pay frame is active and the pay is large enough to be motivating. The transition from intrinsic to extrinsic motivation is discontinuous: once money is introduced, intrinsic motivation is partially displaced.

**Secondary effect:** Small payments may also signal that the task itself has low value or low importance. "They are only paying me 3 cents per answer — this must not matter much."

### Practical Conclusion

Gneezy and Rustichini conclude: **pay significantly or do not pay at all.** There is no safe small-payment strategy when intrinsic motivation is present.

### Identifying Intrinsic Motivation

| Signal | Implication |
|---|---|
| Agent works below market rate voluntarily | Mission, reputation, or identity motivation likely present |
| Quality is highest when financial monitoring is lowest | Intrinsic standards drive performance, not fear of penalty |
| Agent describes work in mission/impact terms, not compensation terms | Strong non-monetary motivation |
| Agent is in a profession with strong ethical norms (medicine, teaching, law) | Professional identity substitutes for financial incentive |
| Agent seeks the work out proactively rather than being recruited | Genuine interest, not just market labor supply |

### Career Concerns as Intrinsic Substitute

Career concerns are not intrinsic in the same sense as mission motivation, but they function similarly — they provide strong incentive for effort without requiring direct financial incentive tied to performance.

**Career concerns are strongest when:**
- Agent is early in career with many years of future earnings at stake
- Agent is in a field where reputation and track record are persistent (academia, law, finance)
- Current work is visible — results will be known to future employers or clients
- Agent expects to stay in the field long-term

**Career concerns are weakest when:**
- Agent is near retirement
- Agent is planning to leave the field
- Work is confidential and cannot signal reputation
- Agent has already established peak career position

**Design implication for managers:**

For early-career employees in visible roles with strong career concerns:
- Reduce direct monetary incentive intensity (career concerns substitute)
- Invest in visibility (prominent projects, external speaking, bylines)
- Provide honest feedback and development — career concern is only useful if performance actually signals ability

For late-career employees or contractors without ongoing relationships:
- Increase direct monetary incentive intensity
- Career concerns are weaker; financial incentives must do more of the work

### The Double-Burden of Risk and Incentive

When an agent is both risk-averse and must be given financial incentives, the employer faces a double cost:
1. **Effort cost compensation:** Expected bonus must cover cost of effort (participation constraint)
2. **Risk premium:** Agent must be compensated for bearing outcome risk (beyond expected value)

The risk premium grows with:
- The size of the bonus (larger gamble → larger risk premium)
- The level of noise in the outcome (higher p_H - p_L means less noise; risk premium is smaller)
- The agent's degree of risk aversion

This is why high-noise environments (where p_H - p_L is small) naturally tend toward fixed salaries or efficiency wages rather than performance pay: the risk premium required to impose large incentives in noisy environments is prohibitively expensive.

FILE:references/multi-task-incentive-design.md
# Multi-Task Incentive Design

## The Multi-Task Problem

Most agents perform multiple tasks simultaneously. When you add a strong financial incentive to one task, it affects effort allocation across all tasks. Whether this is helpful or harmful depends on the relationship between the tasks.

**Core insight (Holmstrom and Milgrom):** An agent with limited effort will direct that effort toward tasks with higher expected marginal payoff. If incentives are unequal across tasks, effort flows to the high-incentive task at the expense of others.

## Substitutes vs. Complements

**Substitute tasks:** More effort on Task A reduces the net productivity of effort on Task B. They compete for a shared resource — time, attention, physical energy, cognitive capacity.

*Diagnostics:*
- Working harder on A makes the agent more tired for B
- Time spent on A is unavailable for B
- Skills required for A crowd out skills for B
- Being very good at A makes being good at B harder (specialization)

*Examples:*
- Corn farming and dairy farming (shared labor hours and physical energy)
- Customer service calls (measured) vs. quality follow-up (not measured)
- Short-term revenue (this quarter) vs. customer satisfaction (long-term)
- Teaching (time-intensive) vs. research (also time-intensive, if not complementary)

**Complement tasks:** More effort on Task A raises the net productivity of effort on Task B. Effort on one makes the other more productive.

*Diagnostics:*
- Success on A makes B easier or more valuable
- Skills built for A transfer to B
- Insights from A generate better outputs in B
- A and B share upstream inputs that benefit both

*Examples:*
- Beekeeping and apple orcharding (bees pollinate apple trees)
- Research and teaching at research universities (research informs lectures; teaching sharpens research questions)
- Prospecting and closing sales (better prospects → easier closes)
- Security and checkout at airports (both part of the same passenger flow process)

## Design Rules

**When tasks are substitutes:**

Strong incentives on Task A divert effort from Task B. If B is important but hard to measure (quality, safety, long-term value), incentivizing only A destroys B performance.

*Rules:*
1. Equalize incentive strength across all tasks you care about
2. Use weaker incentives on each task than you would for a single-task role
3. Consider whether some tasks can be unmeasured entirely — but only if they are genuinely secondary
4. If you cannot avoid unequal measurement, ensure the well-measured task is the most important one

*Example:* Teachers whose pay is tied to standardized test scores (measurable) have incentives to teach to the test at the expense of critical thinking, creativity, and subjects not on the test. The fix: either include the hard-to-measure dimensions in the evaluation or reduce the weight on the easily-measured dimension.

**When tasks are complements:**

Strong incentives on all tasks amplify each other. No tradeoff between tasks; pushing hard on A helps B too.

*Rules:*
1. Use strong incentives on all tasks in the bundle
2. Bundle complementary tasks together under one person (or one team with shared incentives)
3. The complementarity creates a multiplier: incentive investment yields above-proportional returns

*Example:* A research university professor incentivized strongly on both publications and teaching quality performs better on both than one incentivized only on publications. The synergy is real.

## Organizational Design Implications

**Group complementary tasks together; separate substitute tasks:**

- Assign sets of complementary tasks to the same person, team, or division
- Give them strong incentives on all tasks in their bundle
- Assign substitute tasks to different people or divisions with independent incentive schemes
- This way, each person faces a bundle of complements with no internal diversion problem

**The Heathrow Airport failure:**

Airport operations are complementary: check-in, security, shopping, boarding, and ground transport are sequential steps in a single passenger flow. All should be managed by one entity with one aligned incentive: maximize passenger throughput and satisfaction.

UK government structure:
- British Airports Authority (BAA) owns and manages the shopping areas, landing fee setting, and physical terminals
- Police manage security (separate budget, separate incentive)
- Regulators set landing fees (separate mandate: minimize airline costs)

Results:
- BAA profits from leases on shops → allocates too little space to security checks (security reduces shopping time)
- Regulator sets landing fees too low → too many airlines choose Heathrow → congestion
- Police budget not tied to passenger throughput → security staffing often inadequate at peak times

Each entity's incentive partially cancels others'. The complementary tasks are in practice governed like substitutes under an adversarial multi-owner structure.

**Multiple owners / bosses — the incentive dilution formula:**

When one agent answers to N bosses with opposed interests, the total incentive strength is approximately:

```
total_incentive_strength ∝ 1 / N
```

This explains why it is hard to get anything done in international bodies (196 sovereign nations as bosses), large committees, and complex coalition governments. Mathematical models show that in the limit of fully opposed principals with equal power, incentive strength approaches zero — "no man can serve two masters."

**Research vs. teaching as a test case:**

*Substitutes hypothesis:* Research takes time and energy from teaching. Optimize by separating: specialized research institutes (CNRS model in France) and teaching-only universities.

*Complements hypothesis:* Research deepens teaching (researchers bring frontier knowledge to the classroom); teaching sharpens research (student questions reveal gaps; teaching forces clarity). Optimize by combining: research universities (US model).

*Evidence:* The comparative success of US research universities in producing both top research and high-quality graduates relative to the French model suggests complements is the better characterization — at least for top-tier universities. The combination does work better when both tasks genuinely feed each other.

## Competition Between Workers as Relative Performance

When many workers do the same task simultaneously under similar conditions, their luck is correlated (all face the same market, weather, or economic conditions). Comparing performance across workers removes the common luck component and yields a cleaner measure of relative effort.

**Relative performance incentives (tournaments):**

- Investment fund managers ranked by returns relative to peer group benchmark
- Dual-source suppliers evaluated against each other on the same contract
- Two proofreaders assigned overlapping pages — both measure the same "true" number of errors, so relative performance reveals who shirked

*Advantages:*
- Filters out common luck; evaluates effort more accurately
- Does not require knowing the absolute level of effort required
- "Your peers performed much better" is a credible response to bad luck excuses

*Disadvantages:*
- Creates incentive for sabotage of peers (reduce their performance rather than improve your own)
- Requires comparable tasks and correlated conditions
- Agents should not know who their comparison group is (prevents strategic coordination)
- Creates a "rat race" dynamic where all agents work extremely hard but relative rankings are unaffected

*The $2/typo first-finder tournament (Yale proofreading experiment):*
All copies of the book distributed to students; reward $2/typo but only to the first finder. The optimal strategy: start from the back of the book (fewer competitors have read the last chapters). Catherine Pichotta won by thinking ahead about competitor behavior, not just working harder. Strategic behavior in tournaments can be as important as raw effort.

FILE:references/nonlinear-schemes-and-gaming.md
# Nonlinear Incentive Schemes and Gaming

## Linear vs. Nonlinear: The Core Tradeoff

**Linear (piece-rate) scheme:**
- Payment = base + rate x performance
- Incremental reward is constant at every performance level
- No threshold effects; agent faces the same marginal incentive throughout

**Nonlinear (quota-bonus) scheme:**
- Payment = low_base if performance < quota; high_base if performance >= quota
- Incremental reward is essentially zero everywhere except near the quota threshold
- Near the threshold, the expected gain from a small effort increase is very large (crossing the quota means getting the full bonus)

## Quota-Bonus Incentive Dynamics

**When the quota works:**
The quota is effective when the agent currently sits just below the threshold — a little more effort meaningfully increases the probability of crossing it. The agent faces a binary choice: work hard to make the quota or accept failure. The entire bonus rides on this marginal effort.

**Example:** Salesperson with $1M annual quota in October, currently at $900K. One strong quarter can clear the quota. The agent has maximum incentive to push hard.

**When the quota fails:**

*Case 1 — Too difficult (unreachable quota):*
Salesperson has a bad first half and is at $300K in June with a $1M annual quota. The quota is now mathematically unreachable. Expected gain from any further effort = 0 (cannot hit the quota regardless). Agent stops trying for the rest of the year. The flat-incentive zone below the threshold expands to cover the entire remaining period.

*Case 2 — Already met (no incremental reward):*
Salesperson meets the $1M quota in October. There is no reward for exceeding $1M. Agent stops and coasts through November and December — or deliberately holds orders to get a head start on next year's quota.

*Case 3 — Strategic order timing (sandbagging):*
Salesperson has $950K in sales with a week to go. She has two large orders she could close now, but if she closes them this year, she will start next year at $0. She can close one order now to make quota, and hold the other order for a strong January start. This sandbagging serves the agent's interest (head start on next year's quota) but damages the employer's interest (lost revenue this year, distorted reporting).

## The Enron Pattern

Enron's incentive scheme created massive rewards for reported revenue and profit — without distinguishing genuine revenue from accounting constructs. The quota-like structure (meet earnings targets = large bonus) created incentives to:
- Book speculative future revenues as current income
- Reclassify off-balance-sheet liabilities
- Enter related-party transactions to manufacture apparent revenue

The scheme was powerful near the earnings threshold (as intended) but in a direction opposite to what the employer wanted. The lesson: nonlinear schemes amplify whatever is being measured; if the measurement can be gamed, the gaming will be amplified too.

## Real Estate Commission Analysis

**Setup:**
- Agent earns 6% commission on sale price
- A $20,000 price increase (from negotiating harder, waiting for a better offer, spending more time marketing) generates 6% x $20,000 = $1,200 for the agent
- But the agent's opportunity cost — time not spent on the next listing — may be worth more than $1,200

**Why the agent misaligns:**
The agent's marginal return on negotiating harder is only $1,200 regardless of how much harder they work. The seller's marginal return is $18,800 (their 94% share). The agent's incentive to maximize price is 15.6x weaker than the seller's.

**Why not just increase the commission rate?**
At 6% the misalignment is 15.6:1. At 50% commission it is 1:1 — perfect alignment — but the seller keeps only half the sale price. The agent would need 100% commission to be fully aligned, which makes no sense (the agent would become the effective owner).

**Better nonlinear design:**
Progressive commission that concentrates incentive where the seller wants effort:
- 3% on sale price up to $500,000 (baseline, agent has less incentive to push beyond this)
- 30% on every dollar above $500,000 (agent has strong incentive to negotiate beyond the base)

This gives the agent 30 cents per dollar above the reserve rather than 6 cents — 5x stronger incentive exactly where it matters.

**Failure mode of progressive structure:**
If the $500,000 reserve is set too high and the market won't support it, the agent faces the unreachable-quota problem (Case 1 above). Must calibrate reserve to a realistic market price.

## Hybrid Design (Best Practice)

The most robust approach combines:

1. **Base commission rate (linear):** Provides constant incentive across all performance levels; prevents complete disengagement when circumstances change

2. **Quota bonuses at thresholds:** 100%, 150%, 200% of base quota add concentrated bursts of incentive at realistic achievement targets

3. **No cliff above the top threshold:** Continue the linear commission above 200% to prevent coasting after the top bonus is hit

This preserves the power of threshold bonuses at realistic performance levels while ensuring the agent never faces a period of zero marginal incentive.

---
name: backward-reasoning-game-solver
description: "Solve sequential-move strategic games using backward induction. Use this skill when a user faces a multi-stage decision or negotiation where players alternate moves and each person's best action depends on what others will do later. Triggers include: user needs to determine the optimal opening move in a turn-based game or negotiation; user wants to know whether they can guarantee a win or favorable outcome before the game starts; user must sequence two risky actions and does not know which to attempt first; user is analyzing a multi-stage political, business, or competitive scenario where one party moves, the other responds, and so on; user has a finite-horizon sequential game with known player preferences and needs the game-theoretic solution; user suspects their opponent can anticipate their moves and wants to reason from end-states backward to their first action; user is building a game tree and needs to prune it to find dominant paths; user has a combinatorial takeaway game (like Nim variants) and wants the winning-position formula; user needs to understand why more flexibility or options can paradoxically hurt a player in a sequential game. This skill handles perfect-information, finite, sequential-move games. It does NOT cover simultaneous-move games (use a separate Nash equilibrium skill), incomplete-information games, or infinite-horizon repeated games."
version: 1.0.0
homepage: https://github.com/bookforge-ai/bookforge-skills/tree/main/books/the-art-of-strategy/skills/backward-reasoning-game-solver
metadata: {"openclaw":{"emoji":"📚","homepage":"https://github.com/bookforge-ai/bookforge-skills"}}
status: draft
source-books:
  - id: the-art-of-strategy
    title: "The Art of Strategy"
    authors: ["Avinash K. Dixit", "Barry J. Nalebuff"]
    chapters: [2]
tags: [game-theory, strategy, decision-making, sequential-games, backward-induction]
depends-on: []
execution:
  tier: 1
  mode: full
  inputs:
    - type: document
      description: "Description of the sequential game: players, sequence of moves, available actions at each decision point, and payoffs or preferences at terminal outcomes"
  tools-required: [Read, Write]
  tools-optional: []
  mcps-required: []
  environment: "Any agent environment; user describes the game situation in text or structured form"
discovery:
  goal: "Identify the game-theoretically optimal strategy for any sequential-move game by constructing a game tree and applying backward induction; deliver the opening move, the full contingent strategy, and the predicted outcome"
  tasks:
    - "Classify the interaction as sequential vs. simultaneous and confirm backward induction applies"
    - "Elicit the complete game structure: players, move order, actions at each node, and terminal payoffs or preference rankings"
    - "Construct or describe the game tree with nodes, branches, and terminal payoffs"
    - "Apply the backward induction procedure: start at terminal nodes, fold optimal choices back to the root"
    - "Identify winning positions vs. losing positions for combinatorial games using the k+1 formula where applicable"
    - "Determine risk sequencing: if multiple risky actions are needed, recommend attempting the harder one first while fallback options remain"
    - "Flag applicability limits: check for uncertainty about player motives, hidden information, or simultaneous moves that would require different tools"
    - "Deliver: optimal first move, full contingent strategy, predicted equilibrium outcome, and key reasoning"
  audience: "Strategists, negotiators, managers, game designers, analysts, and decision-makers facing any turn-based strategic interaction"
  when_to_use:
    - "User needs to solve a turn-based game or sequential negotiation"
    - "User must choose between attempting a risky action now vs. later and wants to know which sequence is optimal"
    - "User is designing rules for a competition and wants to predict outcomes before play begins"
    - "User faces a multi-stage business or political scenario where each party observes the other's move before responding"
    - "User wants to know if their current position in a game is a guaranteed win or guaranteed loss"
  quality:
    correctness: null
    depth: null
    actionability: null
    specificity: null
---

# Backward Reasoning Game Solver

## When to Use

Use this skill for any strategic interaction where players move one at a time, each player can observe what the previous player did, and the game ends in a finite number of moves.

The core principle: **Rule 1 — Look forward and reason backward.** Anticipate where your initial decisions will ultimately lead and use that information to calculate your best current choice. This rule applies whether the game lasts two moves or two hundred.

The key insight backward reasoning delivers: a future action that "lies in the future does not mean it is uncertain." If you can deduce what a rational opponent will choose at a later node — because you know their preferences — you can treat that future action as a known fact today when computing your own best move now.

**This skill applies when all of the following hold:**

- Players move in sequence (not simultaneously)
- Earlier moves are observable before the next player responds
- The game ends in a finite number of moves
- Preferences or payoffs at each terminal outcome are known or can be reasonably estimated

**This skill does NOT apply to:**

- Simultaneous-move games (rock-paper-scissors, sealed-bid auctions, price-setting)
- Games with hidden information or private cards
- Infinite-horizon repeated interactions where reputation effects dominate
- Situations with deep uncertainty about opponents' objectives (estimate preferences first, then apply)

---

## Context and Input Gathering

### Required (ask if missing)

- **Players and move order:** Who moves first, second, etc.? Does the sequence alternate, or does one player move multiple times in a row?
  -> Ask: "Walk me through who acts and in what order."

- **Actions at each decision point:** What choices does the player-to-move have at each node?
  -> Ask: "At each turn, what are the available actions?"

- **Terminal payoffs or preference rankings:** What does each player receive or prefer at each possible end-state?
  -> Ask: "At each possible ending, how do you rank the outcomes? Best to worst is enough if exact numbers are unavailable."

- **Applicability check:** Is this sequential (players observe each other's moves before responding)?
  -> Ask: "Does each player see what the other did before choosing their own move?"

### Useful (gather if present)

- The number of total moves or stages (needed to apply the k+1 formula for combinatorial games)
- Whether a player needs their opponent to believe a threat is credible (signals a commitment/credibility issue that modifies pure backward reasoning)
- Whether the game is repeated (a one-time game vs. an ongoing relationship changes payoffs)

---

## Execution

### Step 1 — Classify the Game Type

**Why:** Backward induction is the correct tool only for sequential games with observable moves. Applying it to simultaneous games produces wrong answers. Spending thirty seconds on classification prevents wasted analysis.

**1a. Is it sequential or simultaneous?**

- Sequential: players observe previous moves before choosing. Backward induction applies.
- Simultaneous: players choose without seeing each other's current action. Use Nash equilibrium (different skill).
- Mixed: some stages are simultaneous, others sequential. Apply backward induction only to the sequential stages; treat simultaneous sub-games separately.

**1b. Is information perfect?**

- Perfect information: all previous moves are known to all players (chess, 21-flags, most negotiation sequences). Full backward induction works.
- Imperfect information: some moves are hidden (card games). Backward induction must incorporate probabilistic beliefs. Flag this and proceed with stated assumptions.

**1c. Is it finite?**

- Finite: the game ends within a known or bounded number of moves. Backward induction is exact.
- Potentially infinite: note limitations; backward induction gives a partial solution (start from whatever end condition you can identify).

---

### Step 2 — Construct the Game Tree

**Why:** The game tree is the visual logic of the game. It makes the sequence of decisions explicit and prevents the common error of optimizing the first branch without considering what happens downstream. A game tree and a decision tree differ in one critical way: at decision-tree nodes only one person chooses; at game-tree nodes, different players may choose at different nodes, each optimizing for their own preferences.

**Structure of a game tree:**

- **Root node:** The first decision point. Label it with the player who moves.
- **Branch:** Each available action at a node. Label branches with the action name.
- **Internal nodes:** Every subsequent decision point. Label with the player who moves there.
- **Terminal nodes (leaves):** End-states. Attach the payoffs or preference rankings for each player.

**Construction procedure:**

1. Start at the root. Draw one branch per available action.
2. For each branch, identify whether the game ends or another player chooses next. If it ends, attach payoffs. If not, draw a new node labeled with the next player.
3. Repeat until all paths reach terminal nodes.
4. If the tree is too large to draw fully, describe it verbally node by node or use the k+1 formula shortcut for combinatorial games (Step 4).

**Critical rule:** Even if you know that certain branches will never be reached (because a player will not rationally choose them), you must still resolve what each player *would* do at every conceivable node. The reason: a player at an earlier node makes choices based on what the other player would do *if* that branch were taken. Pruning a branch without resolving it first is a mistake.

---

### Step 3 — Apply Backward Induction

**Why:** Forward reasoning — picking the branch that looks best at the first move — fails because it ignores what rational opponents will do in response. The correct procedure is the reverse: start at the end, where outcomes are fully known, and fold optimal decisions back toward the start. This converts every future branch into a known consequence, making the first move's true value calculable.

**Backward induction procedure (step by step):**

1. **Identify all terminal nodes.** These have definite payoffs. No decision is needed here.

2. **Move one step back.** Find every internal node whose branches all lead directly to terminal nodes. These are the "penultimate" nodes — the last decision points in the game.

3. **At each penultimate node, find the optimal action for the player who moves there.** Optimal = the action leading to the terminal node with the best payoff for *that player* (not you, not the other side — the player who moves at that node). Mark this branch as the selected path (thicken it or mark with an arrow).

4. **Replace the penultimate node with its outcome.** Now treat that node as if it were a terminal node whose payoff is whatever the selected branch leads to. The subtree below it is resolved.

5. **Repeat steps 2-4, moving one level back each time,** until you reach the root node.

6. **At the root, the optimal action is determined.** The full path of selected branches from root to terminal is the predicted equilibrium path.

**Worked illustration (Charlie Brown / Fredo Investment):**

The same tree structure covers both cases:
```
Player A
├── Action 1 → Player B
│              ├── B's preferred action → Outcome: [A: bad, B: good]
│              └── B's other action   → Outcome: [A: good, B: okay]
└── Action 2 → Outcome: [A: neutral, B: neutral]
```
Backward induction at B's node: B prefers "B's preferred action." Fold back: if A takes Action 1, the realized outcome is [A: bad, B: good]. Compare to Action 2: [A: neutral]. A prefers neutral to bad → A takes Action 2. Equilibrium: Action 2 without ever reaching B's node.

**Common error to avoid:** Do not prune B's node without resolving it. A must compute what B would do if reached — even if A ends up not taking that branch.

---

### Step 4 — Combinatorial Game Shortcut (Winning-Position Formula)

**Why:** For games where players alternately remove objects from a pile (or similar combinatorial structures), drawing the full tree is impractical for large games. The pattern in the terminal logic propagates backward into a closed-form formula that immediately identifies winning and losing positions.

**The k+1 formula:**

In a game where each player may take 1 to k objects per turn, and the player who takes the last object wins:

- **Losing positions:** multiples of (k + 1). A player facing a multiple of (k+1) cannot win against a perfect opponent.
- **Winning positions:** any other count. The correct move is to take enough objects to leave the opponent at the nearest multiple of (k+1).

**21-flags example (k=3, so k+1=4):**
- Losing positions: 4, 8, 12, 16, 20
- Winning first move: take 1, leaving opponent with 20 (a multiple of 4)
- Opponent takes n (1, 2, or 3); you take 4-n; the count drops to 16, 12, 8, 4, then 0

**The "hot potato" variant (player who takes last loses):**
The same formula applies but the parity flips: losing positions are still multiples of (k+1), but now the positions are interpreted in reverse — check whether facing a multiple means you are forced to take the last item.

**Generalizing:** Any time a combinatorial game has a cycle length derivable from the rules, apply backward induction to the smallest cases (1, 2, 3, ... objects) to identify the pattern, then project it forward.

---

### Step 5 — Risk Sequencing: Attempt Riskier Actions First

**Why:** When a player needs multiple unlikely successes to reach a goal, and can attempt them in any order, the correct sequence is to attempt the *harder or riskier action first while fallback options remain open*. Attempting the easier action first and failing the harder one second results in a forced loss that the reversed sequence would have avoided.

**The Orange Bowl principle (Dixit and Nalebuff):**

Nebraska needed two touchdowns plus net extra points to win. The coach kicked the safe one-point conversion after the first touchdown. When the second touchdown was scored, he was forced to attempt the two-point conversion with no margin. The correct strategy: attempt the two-point conversion first. If it succeeds, the subsequent one-point kick covers the margin. If it fails, there is still a chance to tie via the one-point kick after the second touchdown. Attempting the risky action first keeps more options alive.

**Generalizable rule:** If you need both A (risky, ~50%) and B (safe, ~90%), and either order is feasible:
- Order B→A: if B succeeds but A fails → lose. No recovery.
- Order A→B: if A fails → still alive for the tie/partial win via B. If A succeeds → B is insurance.

Attempt the riskier action first. This applies to: new product launches before marketing commitments are locked in, career pivots before exhausting current options, negotiation concessions where the harder ask should come before the easier one is spent.

---

### Step 6 — Check Applicability Limits

**Why:** Backward induction gives the correct answer only given its assumptions. Violating those assumptions without acknowledging it produces overconfident, wrong predictions. Identifying limits is part of delivering a sound analysis.

**Three conditions that limit backward induction:**

1. **Natural uncertainty (chance nodes):** Some nodes are resolved by probability (a die roll, a market shock), not a player's decision. Incorporate expected values at chance nodes using the same backward induction structure. Note that the result is now a probabilistic expectation, not a guaranteed outcome.

2. **Unknown opponent objectives:** Backward induction requires predicting what the opponent will choose, which requires knowing their preferences. If you are uncertain, you must estimate. The result is a conditional recommendation: "If your opponent values X more than Y, then your best move is Z." If opponent motives are deeply uncertain, behavioral game theory (incorporating altruism, fairness, reputation) supplements the pure analysis.

3. **Strategic uncertainty in simultaneous sub-games:** If any stage involves players choosing simultaneously, that stage requires Nash equilibrium analysis. Solve the simultaneous sub-game first, substitute its equilibrium payoffs into the larger game tree, then continue backward induction.

**Paradox of expanded options:** More choices or greater freedom can hurt a player in a sequential game. Adding options to one player (like a presidential line-item veto) changes what the other player anticipates and may cause the other player to respond in a way that leaves the first player worse off. Backward induction reveals these paradoxes; intuition does not.

---

### Step 7 — Deliver the Solution

Structure your output as:

**Optimal opening move:** [Specific action for the first player, stated plainly]

**Full contingent strategy:** [Complete if-then plan: "If opponent does X, do Y; if opponent does Z, do W"]

**Predicted equilibrium outcome:** [The terminal state that backward induction leads to, and each player's payoff there]

**Key reasoning:** [The one or two backward induction steps that most determine the outcome — the nodes where the game is effectively decided]

**Applicability notes:** [Any assumptions made about opponent preferences or probability estimates, and how violations would change the answer]

---

## Key Principles

**Rule 1 is the foundation.** Look forward and reason backward. All other steps implement this rule.

**Future actions are predictable, not uncertain.** In a sequential game, a rational opponent's future choice follows from their preferences — it is deducible, not random. Treating predictable future actions as "uncertain" is hope over analysis.

**Resolve every node, even unreachable ones.** A player at node A chooses based on what would happen at node B *if* A chose to go there. If node B is unresolved, node A cannot be correctly decided.

**Game trees and decision trees differ in kind.** A decision tree has one optimizer throughout. A game tree has multiple optimizers, each at their own nodes, each optimizing for themselves. Do not collapse a game tree into a single-optimizer problem.

**More options can hurt.** In sequential games, gaining additional choices can signal different intentions to opponents and cause their responses to shift unfavorably. Backward reasoning reveals this paradox; forward intuition misses it.

**Attempt risky actions while fallback options remain.** Risk sequencing is a corollary of backward induction: work backward from what you need to achieve, and structure the order of attempts so that failure of the hardest step leaves the most options open.

**The first-mover advantage is derivable, not assumed.** Whether moving first helps or hurts depends entirely on the structure of the game tree — backward induction tells you which.

---

## Examples

### Example 1: The 21-Flags Game (Survivor: Thailand)

**Setup:** 21 flags; players alternate; each turn take 1, 2, or 3; player who takes the last flag(s) wins.

**Apply k+1 formula (k=3, so k+1=4):**
- Losing positions: 4, 8, 12, 16, 20
- First player wins by taking 1, leaving opponent with 20

**Full contingent strategy for first player:** Take 1. Whatever opponent takes (n), take 4-n. After your turn the count is always a multiple of 4. Opponent is trapped.

**Why Sook Jai lost:** They took 2 on the opening move, leaving 19 — not a multiple of 4. This handed the initiative to Chuay Gahn. The tribe needed to reason all the way back to the opening move; recognizing the 4-flag trap mid-game is too late.

---

### Example 2: The Fredo Investment Game

**Setup:** Charlie considers investing $100K with Fredo. If Charlie invests, Fredo can honor the contract (Charlie nets $150K, Fredo nets $250K) or abscond (Charlie loses $100K, Fredo keeps $500K). If Charlie does not invest, both get $0.

**Game tree:**
```
Charlie
├── Invest → Fredo
│             ├── Abscond → [Charlie: -$100K, Fredo: +$500K]
│             └── Honor   → [Charlie: +$150K, Fredo: +$250K]
└── Don't  → [Charlie: $0, Fredo: $0]
```

**Backward induction at Fredo's node:** Fredo prefers $500K over $250K → will Abscond.

**Fold back:** If Charlie invests, realized outcome is [Charlie: -$100K]. Compare to Don't: [Charlie: $0]. Charlie prefers $0 → Don't invest.

**Equilibrium:** No investment; both get $0. The mutually beneficial outcome ($150K/$250K) is blocked by the inability to commit credibly.

**Caveat:** This changes if the game is repeated or if Fredo has other US-dependent business interests — these create an ongoing game with reputation effects that can sustain cooperation.

---

### Example 3: The Orange Bowl Risk Sequencing

**Setup:** Nebraska needs 2 extra points net across two touchdowns. Option A: two-point conversion attempt (~50% success). Option B: one-point kick (~95% success). Osborne's order: B then A. Alternative: A then B.

**Backward induction on Osborne's order (B first):**
- If B succeeds and A succeeds → win (needed outcome)
- If B succeeds and A fails → lose (no recovery)
- If B fails and A succeeds → tie (acceptable)
- If B fails and A fails → lose

**Backward induction on alternative order (A first):**
- If A succeeds and B succeeds → win
- If A succeeds and B fails → still ahead; other outcomes now favorable
- If A fails → still score second touchdown; one-point kick gives tie (acceptable)
- If A fails and score another: now need exactly 2 points — same situation as Osborne faced but with a remaining chance

**Key insight:** The only scenario where order matters is when exactly one attempt fails. If A fails first, the fallback (B on the second touchdown) still yields a tie. If B fails first (Osborne's plan), the second touchdown forces a must-make two-pointer with no margin. Attempt the risky action first.

---

## References

- `references/game-tree-construction.md` — Detailed node-labeling conventions, tree notation for complex games, handling chance nodes
- `references/combinatorial-game-formulas.md` — k+1 formula derivations, hot-potato variants, multi-pile Nim, worked tables for games up to 100 objects
- `references/applicability-checklist.md` — Diagnostic questions for classifying a game (sequential vs. simultaneous, perfect vs. imperfect information, finite vs. infinite)
- `references/risk-sequencing-patterns.md` — Generalization of the Orange Bowl principle to business, negotiation, and career scenarios; worked examples with probability trees

## License

This skill is licensed under [CC-BY-SA-4.0](https://creativecommons.org/licenses/by-sa/4.0/).
Source: [BookForge](https://github.com/bookforge-ai/bookforge-skills) — The Art of Strategy by Avinash K. Dixit, Barry J. Nalebuff.

## Related BookForge Skills

This skill is standalone. Browse more BookForge skills: [bookforge-skills](https://github.com/bookforge-ai/bookforge-skills)

FILE:references/applicability-checklist.md
# Applicability Checklist for Backward Induction

Use this checklist to determine whether backward induction is the correct tool, and if not, what adjustment or alternative to use.

---

## Step 1: Sequential vs. Simultaneous

**Question:** Do players observe each other's moves before choosing their own?

| Answer | Implication |
|---|---|
| Yes, all moves are observed before the next player responds | Fully sequential. Backward induction applies directly. |
| No, players choose simultaneously without observing the current action | Simultaneous game. Use Nash equilibrium analysis instead. |
| Mixed: some stages are sequential, some simultaneous | Apply backward induction to sequential stages. Solve simultaneous sub-games using Nash equilibrium and substitute equilibrium payoffs into the tree. |

---

## Step 2: Perfect vs. Imperfect Information

**Question:** Do players know all previous moves and the full state of the game at their decision point?

| Answer | Implication |
|---|---|
| Yes: all history is observable (chess, negotiation sequences, flag games) | Perfect information. Backward induction is exact. |
| No: some prior moves or private information is hidden (card games, some auctions) | Imperfect information. Backward induction requires probability estimates at hidden-information nodes. Results are conditional on beliefs. |
| Partially: some information is public, some private | Model public history as the tree structure; treat hidden information as opponent-type uncertainty. Use stated assumptions and sensitivity analysis. |

---

## Step 3: Finite vs. Potentially Infinite

**Question:** Does the game end within a bounded number of moves?

| Answer | Implication |
|---|---|
| Yes: game ends in a finite, known number of moves | Finite. Backward induction is exact and terminates. |
| Yes, but very large (chess-scale): finite in principle, infeasible in practice | Apply backward induction to the endgame (few pieces / final stages). Use heuristic evaluation for midgame positions. Combine game theory with domain expertise. |
| No: the game can repeat indefinitely | Backward induction does not directly apply. Consider: discounted payoffs, trigger strategies, reputation models for infinite-horizon games. |

---

## Step 4: Known vs. Unknown Preferences

**Question:** Do you know (or can you reasonably estimate) what each player prefers at each terminal outcome?

| Answer | Implication |
|---|---|
| Yes: preferences are clearly known (win > tie > loss; higher payoff > lower) | Full backward induction. |
| Approximately known: preferences can be ranked but not precisely measured | Backward induction with ordinal payoffs. Results are robust to small changes in rankings. |
| Uncertain: you do not know whether your opponent values outcome A over B | You cannot run backward induction without an assumption. Options: (1) try both preference orderings and deliver conditional recommendations; (2) ask for more context on opponent motives; (3) acknowledge uncertainty in the deliverable. |
| Opponent has non-standard preferences (altruism, fairness, spite) | Behavioral game theory adjustment. Incorporate fairness concerns as an additional payoff component (e.g., opponent prefers equal split over maximizing own take). |

---

## Step 5: One-Shot vs. Repeated

**Question:** Is this game played once or repeatedly between the same players?

| Answer | Implication |
|---|---|
| One-shot: single play with no future interaction | Pure backward induction. No reputation effects. Fredo-type defection is rational; Charlie should not invest. |
| Repeated with known end: finite repetitions | Backward induction applies to the repeated game. In many cases (e.g., finite prisoner's dilemma), cooperation unravels back to the one-shot outcome. |
| Repeated indefinitely or with unknown end | Reputation and reciprocity become relevant. The "shadow of the future" can sustain cooperation. Pure backward induction understates the range of sustainable outcomes. |

---

## Quick Classification Card

Run through this five-question sequence. The first "No" determines the tool.

1. Sequential moves? → If No: use Nash equilibrium
2. Observable history? → If No: use beliefs + backward induction (Bayesian)
3. Finite game? → If No: use repeated-game / reputation models
4. Known preferences? → If No: run conditional analysis or gather more information
5. One-shot? → If No: consider reputation effects as payoff additions

If all five answers are Yes: proceed with standard backward induction (Steps 2-7 of the main skill).

---

## The Paradox of More Options

A recurring finding from backward induction analysis: **giving a player more choices can make them worse off**.

This occurs when expanded options change the other player's anticipated responses. The example from Chapter 2: a presidential line-item veto appears to increase presidential power. But backward induction shows that Congress, anticipating the veto, changes the bills it passes in ways that leave both players worse off than under the no-veto regime.

Check for this paradox whenever:
- A player gains a new action or capability
- You are modeling how that change affects both players' strategies
- The outcome for the "empowered" player seems counterintuitively worse

The mechanism: more options at your node can signal or enable different things to opponents at earlier nodes, shifting their behavior before you even act.

FILE:references/combinatorial-game-formulas.md
# Combinatorial Game Formulas Reference

## The k+1 Winning-Position Formula

### Setup
- A pile of N objects
- Players alternate turns
- On each turn, a player must take between 1 and k objects (inclusive)
- The player who takes the **last** object(s) **wins**

### Formula

**Losing positions:** Any count that is a **multiple of (k+1)**

If you face a multiple of (k+1) and your opponent plays perfectly, you cannot win.

**Winning positions:** Any count that is **not** a multiple of (k+1)

Your winning move: take enough objects to leave your opponent with the nearest multiple of (k+1).
- Winning move = (current count) mod (k+1)
- This leaves opponent with count = current - [(current mod (k+1))], which is a multiple of (k+1)

### 21-Flags Example (k=3)

k+1 = 4. Losing positions: 4, 8, 12, 16, 20.

| Count facing you | Your status | Correct move |
|---|---|---|
| 21 | Winning | Take 1 (leave 20) |
| 20 | Losing | No winning move |
| 19 | Winning | Take 3 (leave 16) |
| 18 | Winning | Take 2 (leave 16) |
| 17 | Winning | Take 1 (leave 16) |
| 16 | Losing | No winning move |
| 13 | Winning | Take 1 (leave 12) |
| 12 | Losing | No winning move |
| 9 | Winning | Take 1 (leave 8) |
| 8 | Losing | No winning move |
| 5 | Winning | Take 1 (leave 4) |
| 4 | Losing | No winning move |
| 3 | Winning | Take 3 (take last, win) |
| 2 | Winning | Take 2 (take last, win) |
| 1 | Winning | Take 1 (take last, win) |

### Verification by Backward Induction

From the terminal nodes backward:
- 1, 2, 3 objects → Winning (take them all)
- 4 objects → Losing (whatever you take — 1, 2, or 3 — opponent takes the rest)
- 5, 6, 7 objects → Winning (take 1, 2, or 3 to leave opponent with 4)
- 8 objects → Losing
- Pattern: every multiple of 4 is a losing position

This is why the formula is derivable from first principles rather than memorized: run backward induction on small cases, observe the pattern, project it forward.

---

## Hot Potato Variant (Last Taker Loses)

Same setup, but the player who takes the last object **loses**.

The losing positions shift by one: a player facing exactly **1** object loses (must take it). Back-propagate:
- 1 → Lose
- 2, 3, 4 → Win (take enough to leave opponent with 1)
- 5 → Lose (any move leaves opponent with 1, 2, 3, 4 — all winning for opponent)

**Revised formula:** Losing positions are counts of the form (k+1)·m + 1 for integer m ≥ 0.

For k=3: losing positions are 1, 5, 9, 13, 17, 21.

If starting count is 21 (a losing position in hot potato), the second player wins, not the first.

---

## Multi-Pile (Nim) Generalization

When there are multiple piles and a player takes any number from one pile per turn, the solution involves XOR (exclusive or) of pile sizes in binary representation. This is standard Nim theory.

For a quick heuristic (exact for two piles):
- Two piles of equal size → the player who faces them loses (second-mover wins by mirroring)
- Two piles of unequal size → first player wins by equalizing the piles

For three or more piles, use the XOR rule: if XOR of all pile sizes = 0, it is a losing position; otherwise it is a winning position. The winning move is to take from any pile such that XOR of remaining piles = 0.

---

## Worked Tables for Common Starting Counts

### k=2 (take 1 or 2), last taker wins

k+1 = 3. Losing positions: 3, 6, 9, 12, ...

| Starting count | First player status | Winning first move |
|---|---|---|
| 10 | Win | Take 1 (leave 9) |
| 9 | Lose | No winning move |
| 7 | Win | Take 1 (leave 6) |
| 6 | Lose | No winning move |

### k=4 (take 1, 2, 3, or 4), last taker wins

k+1 = 5. Losing positions: 5, 10, 15, 20, 25, ...

| Starting count | First player status | Winning first move |
|---|---|---|
| 21 | Win | Take 1 (leave 20) |
| 20 | Lose | No winning move |
| 17 | Win | Take 2 (leave 15) |
| 15 | Lose | No winning move |

---

## Deriving the Formula for a New Game

If you encounter a combinatorial game not matching these patterns, derive from scratch:

1. Solve the terminal cases (1, 2, 3, ... objects) by inspection.
2. Label each as W (winning) or L (losing).
3. A position is L if and only if all moves from it lead to W positions.
4. A position is W if and only if at least one move leads to an L position.
5. List the L positions: 1, ?, ?, ...
6. Identify the gap between consecutive L positions — that gap is (k+1) for standard takeaway games.
7. Project the pattern forward.

FILE:references/game-tree-construction.md
# Game Tree Construction Reference

## Node Labeling Conventions

### Standard Notation

- **Decision node:** Circle or labeled box with the player's name or identifier (P1, P2, etc.)
- **Chance node:** Diamond or circle marked "Nature" or "N" — used when probability resolves the branch rather than a player
- **Terminal node:** Rectangle or square with payoffs listed for each player in the order they entered the game (P1 payoff, P2 payoff, ...)
- **Branch:** Arrow from node to next node or terminal, labeled with the action name
- **Selected branch (backward induction result):** Thickened line or arrowhead added after solving

### Payoff Representation

Two formats work equally well:

**Ordinal (rankings):** Use when exact utilities are unknown but preference ordering is clear.
- "4 = best, 1 = worst" for each player listed separately
- Example terminal node: `[Congress: 3, President: 3]`

**Cardinal (utilities or money):** Use when magnitudes matter (expected value calculations require cardinal payoffs).
- Example terminal node: `[Charlie: -$100,000, Fredo: +$500,000]`

When preferences are mixed (one player's ranking is clear, another's is ambiguous), estimate and state assumptions explicitly.

---

## Tree Structure for Common Game Types

### Two-Player Alternating Game (Most Common)

```
P1 (root)
├── Action A → P2
│              ├── Response 1 → Terminal [P1: x, P2: y]
│              └── Response 2 → Terminal [P1: x', P2: y']
└── Action B → P2
               ├── Response 1 → Terminal [P1: a, P2: b]
               └── Response 2 → Terminal [P1: a', P2: b']
```

### Three-Stage Game

```
P1 (root)
├── A1 → P2
│         ├── B1 → P1
│         │        ├── C1 → Terminal
│         │        └── C2 → Terminal
│         └── B2 → P1
│                  ├── C3 → Terminal
│                  └── C4 → Terminal
└── A2 → Terminal
```

Note: P1 appears at both the root and at Stage 3. At Stage 3, P1 optimizes for their own payoff given what has already happened — they do not undo the prior move. The game is still solved by backward induction starting from the Stage 3 nodes.

### Game with Chance Node

```
P1 (root)
└── Risky Action → Nature (N)
                    ├── Success (prob p) → P2
                    │                       ├── ...
                    └── Failure (prob 1-p) → Terminal [P1: 0, P2: 0]
```

At a chance node, do not pick the "best" branch — compute the expected payoff (probability × payoff) and fold that expected value back to the previous decision node.

---

## Handling Very Large Trees

When the game has many players, many stages, or many actions per node, drawing the full tree is impractical. Two approaches:

### Approach 1: Logical Description by Layer

Instead of drawing, describe the tree layer by layer:

- Layer 0 (root): P1 chooses from {A, B, C}
- Layer 1: If A, P2 chooses from {D, E}; if B, P2 chooses from {F}; if C, game ends
- Layer 2: If D, P1 chooses from {G, H}; if E, game ends; if F, P1 chooses from {I, J}
- Terminal payoffs: G → [P1: 3, P2: 1]; H → [P1: 1, P2: 4]; E → [P1: 2, P2: 2]; I → [P1: 4, P2: 0]; J → [P1: 0, P2: 3]

Then apply backward induction layer by layer from the terminals.

### Approach 2: Pattern Recognition + Formula

For combinatorial games where the structure is regular (same choices available regardless of history, only the state size changes), derive the winning-position pattern from small cases and generalize. See `combinatorial-game-formulas.md`.

---

## The Critical Rule: Resolve All Nodes Including Counterfactual Ones

A node is "counterfactual" if the backward induction solution predicts it will never be reached. It must still be resolved.

**Why:** Player at node X makes decisions based on what would happen at node Y if X chose the branch leading there. If Y is unresolved, X's calculation is incomplete.

**Example:** In the Congress/President line-item veto game (p. 53-54 of source), Congress's optimal move depends on what the President would do if Congress passed each possible bill. Even if Congress ends up passing "neither" (and thus the President's nodes for "U only" and "M only" are never reached), Congress must reason through what the President would do there to know that passing neither is optimal.

Mark unresolved counterfactual nodes with dashed lines to distinguish them from the predicted equilibrium path.

---

## Common Construction Errors

| Error | Consequence | Fix |
|---|---|---|
| Pruning a branch before resolving it | Incorrect backward induction at parent node | Resolve all branches before pruning |
| Using your own payoffs at opponent's nodes | Wrong prediction of opponent's choice | At each node, use the payoffs of the player who moves there |
| Merging sequential and simultaneous stages | Invalid tree structure | Model simultaneous stages as a sub-game payoff matrix; substitute equilibrium payoffs into the tree |
| Omitting chance nodes | Overstated certainty | Insert Nature nodes wherever probability (not choice) determines outcomes |
| Stopping the tree before all paths terminate | Cannot apply backward induction | Extend every path until payoffs are reachable (add "game continues" node if truly unbounded) |

FILE:references/risk-sequencing-patterns.md
# Risk Sequencing Patterns Reference

## The Core Principle

When you need multiple risky actions to all succeed, and you can choose the order of attempts, **attempt the riskier action first** — while fallback options remain available.

This is a direct corollary of backward induction: reason from the end-state you need to reach, and structure the sequence so that failure of the hardest step leaves the maximum number of options open.

---

## The Orange Bowl Case (Source: Ch. 2, pp. 71-73)

**Situation:** Nebraska needs net +3 extra points across two touchdowns to win the championship.
- Option A: Two-point conversion (~50% success, needed twice for a certain win, or once for a tie)
- Option B: One-point kick (~95% success, "safe")
- Osborne's order: B first (safe kick after TD1), then A (two-pointer after TD2 — now forced)

**Why Osborne's order was wrong:**

The only scenario where order matters is when exactly one attempt fails. Consider each case:
- Both succeed: order is irrelevant. Win regardless.
- A fails, B succeeds: if B→A order, A is the last attempt. No fallback. Lose. If A→B order, A fails but B after TD2 still gives a tie (championship by record).
- B fails, A succeeds: under B→A, this means the kick after TD1 failed (very rare), but two-pointer after TD2 succeeds. Net result: 31-30, Nebraska wins. Under A→B, same outcome.
- Both fail: order is irrelevant. Lose regardless.

**The asymmetry:** The only scenario where order changes the outcome is "A fails, B succeeds." Under B→A (Osborne's plan), this means: one-point kick succeeds after TD1 → two-pointer attempted after TD2 → fails → lose. Under A→B (correct): two-pointer attempted after TD1 → fails → score TD2 → one-point kick → tied game → championship by record.

**Backward induction logic:** Work backward from "what do I need after the second touchdown?" If the two-pointer failed on the first TD, you need a two-point attempt after the second TD too — but then failure means game over with no remaining fallback. If the two-pointer succeeded on the first TD, the one-point kick after the second TD is pure insurance.

---

## Generalized Risk-Sequencing Rule

**Setup:** You need both action A (probability p_A of success) and action B (probability p_B of success) to achieve your goal, where p_A < p_B (A is riskier). You can attempt them in either order. Failure of one does not make the other impossible.

**Compare outcomes where exactly one attempt fails (the only scenario where order matters):**

| Scenario | Order A→B outcome | Order B→A outcome |
|---|---|---|
| A fails, B succeeds | Partial goal achieved (B succeeded; fallback) | Forced to attempt A after using up B |
| B fails, A succeeds | Partial goal achieved (A succeeded; fallback) | Same as A→B |

In Order A→B: if A fails first, B is still available and may recover a fallback position.
In Order B→A: if B succeeds but A fails, no fallback remains.

**Recommendation:** Always attempt the lower-probability (harder, riskier) action first when a fallback exists.

---

## Domain Applications

### Sports / Competition

**Tennis serves:** First serve = aggressive (high risk, high reward). Second serve = safe (lower risk). If you hit the easy serve first and miss the aggressive one, you double-fault. The correct sequencing is hardest first (aggressive serve), safe second. Tennis players do this instinctively; the game-theoretic logic is exactly the risk-sequencing principle.

**Golf approach shots:** If playing for a par save requires a risky chip followed by an easy putt, attempt the risky chip while you still have the putt as backup.

---

### Business / Product Development

**New product launch:** A team must both (A) secure a manufacturing partner (uncertain, hard negotiation) and (B) sign a marketing agency (easier, multiple options available). Correct order: negotiate manufacturing first. If it fails, the marketing spend has not been committed. If manufacturing is locked in first, marketing negotiations happen with all options open.

**Investment due diligence:** A startup needs both (A) regulatory approval (uncertain, slow) and (B) investor commitment (easier to get conditionally). Correct order: pursue regulatory approval while investor commitment is conditional. Lock in regulatory approval, then firm up the investment. Investor commitments typically allow conditional withdrawal; regulatory rejections do not refund marketing spend.

---

### Negotiation

**Multi-issue negotiation:** You need to resolve both a harder concession (salary, equity, price) and an easier one (start date, title, payment terms). Attempt the harder concession first. If it fails, the easier concession has not been spent as a goodwill gesture. If the harder concession succeeds, the easier one strengthens the deal.

**Warning:** The counterparty may prefer the reverse order — they want you to spend your easy concessions early so that the hard ask comes after you are invested in the deal. Recognize this tactic and sequence your concessions to preserve your fallback.

---

### Career Decisions

**Switching fields:** Moving from finance to technology requires both (A) building a technical portfolio (uncertain, 6-month effort) and (B) updating a network and personal brand (lower risk, ongoing). Correct order: invest in the portfolio first while still employed in finance (fallback intact). Do not announce the pivot publicly (B) before the portfolio demonstrates readiness (A). If A stalls, the fallback position in finance is still accessible.

---

## When the Principle Does Not Apply

Risk sequencing only matters when:

1. Both actions are needed (not just preferred)
2. The order is flexible (some sequences are forced by logistics)
3. Failure of one action does not eliminate the option to attempt the other
4. A fallback outcome exists if the harder action fails first

If there is no fallback (all-or-nothing), order is irrelevant. If the actions are independent in payoff, backward induction may reveal a different priority. Always draw the tree first.

---

## Quick Decision Check

Ask these questions before sequencing:

1. Which action has the lower probability of success? → Attempt this one first.
2. If the harder action fails, what fallback remains? → The answer defines your "tie" or partial win.
3. If the easier action is used first and the harder fails, is the fallback still accessible? → If no: you are locked in; Osborne's mistake.
4. What is the difference in outcome between "harder fails, easier succeeds" under each ordering? → Quantify the benefit of correct sequencing.

---
name: auction-bidding-strategist
description: "Apply the complete game-theoretic auction framework to determine the optimal bid in any auction format. Use this skill when a user is preparing to bid in an English (ascending) auction, a Japanese auction, a Vickrey (second-price sealed-bid) auction, a Dutch (descending-clock) auction, or a standard sealed-bid first-price auction, and wants the game-theoretically correct strategy rather than guesswork. Triggers include: user is deciding how much to bid in a competitive tender, procurement auction, real estate auction, eBay auction, spectrum license auction, or corporate acquisition; user is worried about overbidding and wants to know how to set a ceiling; user suspects they may be falling into the winner's curse — winning but regretting the price paid; user must classify whether the auction involves private values (each bidder's value is independent) or common values (the item has a single underlying value that all bidders are estimating), because the correct strategy differs sharply between the two; user is evaluating whether to participate in a dollar auction, bidding war, or war-of-attrition-style competitive spending contest and wants to know when to stop or avoid; user needs to shade a bid below their true value in a sealed-bid first-price format and wants the formula; user is designing an auction and wants to know which format will yield more seller revenue; user is bidding in multiple simultaneous auctions and needs to think across the games. This skill does NOT cover multi-round negotiation without a defined auction structure (use a negotiation skill instead), combinatorial auctions with complex package bids, or procurement auctions requiring cost estimation."
version: 1.0.0
homepage: https://github.com/bookforge-ai/bookforge-skills/tree/main/books/the-art-of-strategy/skills/auction-bidding-strategist
metadata: {"openclaw":{"emoji":"📚","homepage":"https://github.com/bookforge-ai/bookforge-skills"}}
status: draft
source-books:
  - id: the-art-of-strategy
    title: "The Art of Strategy"
    authors: ["Avinash K. Dixit", "Barry J. Nalebuff"]
    chapters: [10]
tags: [game-theory, auctions, bidding, competitive-strategy, winner-curse]
depends-on: []
execution:
  tier: 1
  mode: full
  inputs:
    - type: document
      description: "Description of the auction situation: format, number of bidders, your estimated value, uncertainty about the item's true value, and any known facts about competitors"
  tools-required: [Read, Write]
  tools-optional: []
  mcps-required: []
  environment: "Any agent environment; user describes auction context in text; agent produces bid recommendation and strategy plan"
discovery:
  goal: "Classify the auction format and value type, then apply the correct bidding rule — producing an explicit bid recommendation, the reasoning behind it, and the traps to avoid"
  tasks:
    - "Identify the auction format: English, Japanese, Vickrey, Dutch, or sealed-bid first-price"
    - "Classify value type: private values (independent), common values (shared underlying value with noisy estimates), or mixed"
    - "Determine the bidder's true value (walkaway number) by asking for all value components"
    - "Apply the format-specific optimal bidding rule"
    - "Apply the winner's curse diagnostic in common-value settings: compute the bid conditional on winning (presume-you've-won analysis)"
    - "Identify whether the situation is a dollar auction / war of attrition, and compute the exit threshold"
    - "Flag revenue equivalence implications if the user is on the seller side or comparing formats"
    - "Deliver: recommended bid, bid rationale, traps to avoid, and format-equivalence notes"
  audience: "Business strategists, procurement managers, real estate buyers, corporate development teams, product managers bidding for ad inventory, investors in M&A situations, and anyone entering a structured competitive bidding process"
  when_to_use:
    - "User is preparing a bid and wants the game-theoretically optimal number"
    - "User has won an auction before and regretted it — diagnosing and correcting winner's curse"
    - "User is unsure whether to shade their bid below their value or bid their true value"
    - "User is deciding whether to enter or continue in a bidding war or competitive spending contest"
    - "User is designing an auction mechanism and wants to understand revenue equivalence"
  quality:
    correctness: null
    depth: null
    actionability: null
    specificity: null
---

# Auction Bidding Strategist

## When to Use

Use this skill whenever a strategic situation has the structure of an auction: one or more items up for bid, a defined rule for who wins, and a defined rule for how much the winner pays. The framework applies far beyond the art auction room — procurement tenders, corporate acquisitions, spectrum licenses, eBay, online ad auctions, and competitive employment offers are all auctions.

**The core discipline of auction strategy: the right bid is not your value. It depends on the format.**

In a second-price (Vickrey) auction, bidding your true value is the dominant strategy. In a first-price sealed-bid auction, bidding your true value guarantees zero profit at best. Getting the format wrong is the single most common and costly bidding error.

**This skill applies when:**

- A defined bidding format governs who wins and how much they pay
- You need an explicit numerical recommendation, not just direction
- The situation involves information asymmetry, common-value uncertainty, or escalation risk

**This skill does NOT apply to:**

- Open-ended negotiations without a structured bid mechanism (use a negotiation skill)
- Complex package-bidding combinatorial auctions (these require specialized integer programming)
- Procurement situations where cost uncertainty dominates — you must first estimate your own cost before bidding strategy applies

---

## Context and Input Gathering

### Required (ask if missing)

- **Auction format:** Is this English (ascending bids, open), Japanese (simultaneous hand-raising, clock rises), Vickrey (sealed, winner pays second-highest bid), Dutch (descending clock, first to stop wins), or standard sealed-bid first-price (sealed, winner pays their own bid)?
  -> Ask: "How does the auction work mechanically — is it open or sealed, does price go up or down, and who pays what if they win?"

- **Your true value:** What is the maximum you would pay and still be satisfied? This is your walkaway number — the price at which you are indifferent between winning and losing.
  -> Ask: "At what price would you be completely indifferent between winning and walking away? What components make up that number — resale value, strategic value, avoided cost if rival wins?"

- **Value type:** Does each bidder's value depend only on their own use (private values), or does the item have an underlying worth that everyone will eventually realize (common values)?
  -> Ask: "Is this an item whose value to you is independent of what others think — like a gift for yourself — or is it an item whose true worth is unknown and all bidders are estimating the same underlying reality, like an oil lease or a company?"

- **Number of bidders:** How many competitors are expected? This affects the bid-shading formula in first-price auctions.

### Useful (gather if present)

- Competitor value estimates (even rough ranges)
- Whether bids are observable or secret during the auction
- Whether this is a one-shot bid or a multi-round process
- Whether multiple items are for sale simultaneously
- Whether the auction is structured as a war of attrition (outlasting, not outbidding)

---

## Execution

### Step 1 — Classify the Auction Format

**Why:** The optimal bidding rule differs sharply across formats. Applying the wrong rule — for example, bidding your true value in a first-price sealed-bid auction — produces zero profit even when you win. One minute of format classification prevents systematic overbidding or underbidding.

**The four canonical formats and their strategic equivalences:**

| Format | Mechanism | Strategic twin |
|--------|-----------|---------------|
| English (ascending) | Open bidding; price rises; last bidder wins | Strategically equivalent to Vickrey |
| Japanese (clock rising) | All hands raised; clock rises; drop out by lowering hand; irreversible | Strategically equivalent to English |
| Vickrey (second-price sealed) | Sealed bids; highest wins; pays second-highest bid | Strategically equivalent to English/Japanese |
| Dutch (descending clock) | Price starts high, falls; first bidder to stop clock wins at that price | Strategically equivalent to first-price sealed-bid |
| First-price sealed-bid | Sealed bids; highest wins; pays their own bid | Strategically equivalent to Dutch |

**Key equivalence pair 1 — English ≈ Vickrey ≈ Japanese:**
All three produce the same outcome: the highest-value bidder wins and pays the second-highest valuation. In English/Japanese formats, the auction ends when the second-to-last bidder drops out at their value. In Vickrey, the winner pays the second-highest bid, which equals the second-highest valuation (since everyone bids truthfully).

**Key equivalence pair 2 — Dutch ≈ First-price sealed-bid:**
Both require the bidder to commit to a price before knowing who else has bid and what they offered. The strategic decision is identical: choose a price that maximizes expected profit by balancing the gain from winning against the risk of losing.

---

### Step 2 — Determine Your True Value (Walkaway Number)

**Why:** Every format's optimal bid is anchored to your true value — either you bid it directly (Vickrey/English) or you shade it (first-price). Without an accurate walkaway number, the downstream bid calculation is wrong regardless of how carefully the formula is applied.

**Components of true value (add all that apply):**

- Direct use value: what the item is worth to you in use
- Strategic option value: premium for keeping the item out of a rival's hands
- Resale value: expected liquidation price if you later sell
- Excitement or prestige premium: if present, include it explicitly so you can manage it later

**The walkaway test:** "At this exact price, I am completely indifferent between winning and losing. One dollar more and I would rather not win. One dollar less and I am happy to win." Only one number passes this test. That number is your value.

**Common error:** Treating your value as a ceiling you hope not to reach, rather than the precise indifference point. This leads to soft dropping in English auctions — dropping out before your true value — and forfeiting wins that would have been profitable.

---

### Step 3 — Apply the Format-Specific Optimal Bidding Rule

**Why:** Each format has a provably optimal bid derived from game theory. Using the correct rule is not a refinement — it is the difference between a profitable bidding strategy and one that either leaves money on the table or produces regret.

#### English Auction (Ascending, Open)

**Rule:** Stay in the bidding until the current price exceeds your true value. Drop out the moment the price would require you to pay more than your walkaway number.

**Why this is optimal:** Dropping earlier forfeits wins you would have valued. Staying later means you would pay more than the item is worth to you. The optimal exit point is exactly at your value, which is a dominant strategy — no information about competitors changes it.

**Bidding increment issue:** If increments are coarse (bidding goes in units of $10 and your value is $95), be aware that the last bid you make (at $90) means someone else can win at $100. Account for whether strategic endgame matters near your value ceiling.

**In common-value settings, modify this rule with Step 5 (winner's curse correction) before applying.**

#### Japanese Auction (Clock Rising, Simultaneous Drop-Out)

**Rule:** Keep your hand raised until the clock reaches your true value. Lower your hand at that exact price.

**Why this is equivalent to English:** The Japanese format provides more information — you know exactly when each competitor drops out and at what price. In a private-value setting, this extra information is irrelevant to your strategy (your value is independent of what others think). In a common-value setting, seeing where others drop out is useful signal that should update your estimate before the auction ends.

**Common-value modification:** If others drop out unusually early, this signals the true value may be lower. Update your estimate downward before proceeding.

#### Vickrey Auction (Second-Price Sealed-Bid)

**Rule:** Bid your exact true value. Not a penny more, not a penny less.

**Why this is a dominant strategy (proof by dominance):**

Consider two cases where bidding below your true value ($50 instead of $60) matters:

1. Highest rival bid is above $60: You lose either way. No difference.
2. Highest rival bid is below $50: You win and pay that rival bid either way. No difference.
3. Highest rival bid is between $50 and $60 (say, $53): If you bid $60, you win and pay $53 — a profit of $7. If you bid $50, you lose. Since your value is $60, winning at $53 is desirable.

The only case where the two bids differ is case 3 — and in that case, bidding your true value is strictly better. The same argument shows bidding above your value is also dominated. Therefore bidding your true value is the unique dominant strategy. You do not need to know how many competitors there are or what they are bidding.

**eBay proxy bidding:** eBay's proxy system approximates a Vickrey auction. Bid your true value as your proxy. The system will only spend up to what is needed to stay ahead. **Exception:** If early bidding reveals information to other bidders about the item's quality or your willingness to pay, sniping (submitting your proxy at the last moment) prevents competitors from updating their estimates and bidding higher. Snipe when your bid would credibly signal higher value to others who might then raise their own valuations.

#### First-Price Sealed-Bid Auction (and Dutch Auction)

**Rule:** Shade your bid below your true value. **Never bid your true value** — this guarantees zero profit even if you win. Shade the bid to balance expected profit against the risk of losing to another bidder.

**The bid-shading formula (symmetric private values, N bidders, values uniform 0–V):**

```
Optimal Bid = V × (N-1) / N
```

Where V is your value and N is the number of bidders (including yourself).

**Examples:**
- 2 symmetric bidders, value = $100: Bid $50 (bid half your value)
- 3 symmetric bidders, value = $100: Bid $67
- 4 symmetric bidders, value = $100: Bid $75
- 10 symmetric bidders, value = $100: Bid $90

**Intuition:** With more competitors, shading less is optimal because the risk that someone else outbids you rises. With fewer competitors, shade more to capture profit margin.

**The "bid as if you've won" principle (always apply in first-price/Dutch):**

When writing down your bid, assume all other bidders are below you. Then ask: given that I am the highest bidder, what is my best bid? This is equivalent to asking: "If I had a confederate who could only lower my bid after I've won, what amount would I instruct them to lower it to?" The answer to that question is the bid you should write down from the start.

**Why this works:** If your original bid would have lost, the confederate's adjustment doesn't matter — you lose either way. If your original bid wins, the confederate lowers it to the same amount you would have written in the first place. The two approaches produce identical results, so you might as well bid the shaded amount from the start.

---

### Step 4 — Classify Private vs. Common Values and Apply Accordingly

**Why:** The English/Japanese/Vickrey dominant strategy of "bid your value" is only valid in private-value settings. In common-value settings, your naive estimate is systematically biased upward — the winner's curse — and the rules must be modified. Failing to identify the value type leads to predictable losses.

**Private values:** Your value is independent of what others think. Signed memorabilia, items for personal use, branded goods where you care only about your own use. Each bidder's value is their own, unaffected by others' valuations.

**Common values:** The item has a single underlying value that all bidders will eventually realize. Examples: offshore oil lease (the oil quantity is what it is, regardless of who extracts it), corporate acquisition (the company is worth what it will generate, regardless of buyer), real estate in a liquid market, Treasury bills. Each bidder has a private estimate of the same common value.

**Mixed:** Most real situations have both components. An oil company that is better at extraction has private value (the efficiency premium) plus common value (the oil itself).

**In common-value settings, jump to Step 5 before computing any bid.**

---

### Step 5 — Apply the Winner's Curse Diagnostic (Common Values)

**Why:** In a common-value auction, the bidder with the highest private estimate wins. But winning means you estimated higher than everyone else — which means your estimate is systematically above the true value. Ignoring this produces what the book calls the winner's curse: you win, but you pay more than the item is worth. The correction is mandatory, not optional.

**The diagnostic question (presume-you've-won analysis):**

Do not ask: "What do I think this is worth?" Ask instead: "Conditional on my bid winning — meaning everyone else estimated lower — what is the item actually worth?"

**Worked example (ACME acquisition):**

- Your due diligence places ACME's current value uniformly between $2M and $12M (average: $7M)
- Your operational expertise can increase value by 50%
- Naive bid: Average value $7M × 1.5 = $10.5M

**Why $10.5M is wrong:** If you offer $10.5M and they accept, the company is worth between $2M and $10.5M today (average: $6.25M). Your 50% improvement brings it to $9.375M — below what you bid. Accepting your $10.5M offer is bad news, not good news.

**Correct procedure:**
1. Presume your offer will be accepted
2. Conditional on acceptance, recompute the expected value
3. Find the bid B such that: (expected value conditional on acceptance) × (your improvement multiplier) = B

**For ACME:** Bid B. If accepted, current value is between $2M and B (average: (2+B)/2). Your 1.5× improvement: 1.5 × (2+B)/2 = B. Solve: 1.5(2+B)/2 = B → 1.5 + 0.75B = B → 1.5 = 0.25B → B = $6M.

At $6M, if accepted, expected current value = $4M, improved to $6M — exactly breakeven. Bid less than $6M to profit; never bid above $6M without accepting expected losses.

**General rule for common-value auctions:** Your bid must account for the adverse selection implicit in winning. Winning is informative: it tells you that everyone else estimated lower. That information should reduce your bid, not be ignored.

**In English/Japanese auctions with common values:** Observe where competitors drop out — this reveals their private estimates of the common value. Use each dropout as a downward signal. If many competitors drop out early, revise your own estimate down before the auction ends.

---

### Step 6 — Recognize and Escape the Dollar Auction Trap

**Why:** Some competitive bidding contests have a structure where sunk costs create escalation beyond rational stopping points. The dollar auction (Shubik's escalation game) illustrates a situation where both top bidders pay, creating a trap with no natural exit once entered. Recognizing this structure before entering — not after — is the key decision.

**The dollar auction structure:**

An auctioneer sells a dollar bill. Highest bidder wins and pays their bid. But the second-highest bidder also pays their bid and gets nothing. Bids start at pennies. Once two bidders are active, the second-place bidder always has an incentive to bid one increment higher (spend a bit more to win something rather than lose and pay). This logic escalates indefinitely — bidding has reached $5 for a $1 bill in classroom experiments.

**Why escalation is rational at each step but catastrophic in aggregate:** At each moment, the bidder in second place calculates: "I can either lose $X (my current bid) and get nothing, or bid $X+1 and have a chance at the $1 prize." This marginal logic is sound, but the cumulative cost is unbounded.

**War of attrition (BSB vs. Sky UK):** The same structure appears in competitive spending wars. BSB and Murdoch's Sky TV both lost £1.5 billion in their satellite TV competition before merging. Each side rationally stayed in because (a) the prize was enormous and (b) sunk losses were irrelevant to future decisions. But both overestimated their ability to outlast the other.

**Exit threshold for war of attrition:**

The rational time to continue is while the expected value of winning exceeds the cost of staying in one more period. But both sides cannot both rationally believe they will outlast the other — overconfidence is the structural cause of catastrophic losses.

**Decision rules:**
1. **Before entering:** Calculate the maximum total you can lose in a competitive spending contest. If that maximum exceeds your capacity, do not enter regardless of the prize's appeal.
2. **Once entered:** Treat sunk costs as irrelevant (they are). Decide each period based only on: does the additional cost of staying in one more period justify the probability of winning the prize? If no, exit immediately.
3. **The best strategy in a dollar auction:** Do not play. The second-best strategy is to commit credibly before the auction that you will bid only up to a fixed ceiling and communicate this to competitors. A credible ceiling makes the auction unprofitable for both and may deter entry.

---

### Step 7 — Simultaneous Multi-Item Auctions (FCC Spectrum Logic)

**Why:** When multiple items are auctioned simultaneously and bidders value portfolios of items, single-item bidding rules produce systematically wrong results. The value of winning item A depends on whether you also win item B. Bidding on each independently ignores this interdependence and produces overbidding or underbidding.

**The FCC simultaneous auction design:** The FCC solved the sequential auction problem (bidding on NY first, then LA, with budget constraints carrying over) by auctioning all licenses simultaneously with open rounds. Bidders could shift bids across items between rounds. This allows the market to aggregate information about cross-item values.

**Key insight for multi-item strategy (AT&T/MCI example):**

When two bidders compete for two items and both can win both, the dominant bidder should recognize that the true cost of winning both items is higher than the sum of winning prices — because aggressively pursuing both items forces prices up on both. The cost of winning the second item may include the incremental price increase it caused on the first.

**Optimal multi-item strategy:**
- Identify which items form your target portfolio and what portfolio value is
- If the combined cost of winning all items exceeds portfolio value, consider deliberately winning only a subset (even the less-preferred item at a lower price) and let the rival win the remainder at a higher price
- Do not bid your full value on all items simultaneously — calculate the marginal value of each additional item conditional on already winning the others

---

### Step 8 — Deliver the Bid Recommendation

Structure your output:

**Auction format identified:** [Format name and strategic equivalent]

**Value type:** [Private / Common / Mixed, with rationale]

**True value (walkaway number):** [Amount and components]

**Recommended bid:** [Explicit number or formula applied to their stated value]

**Bid rationale:** [Which rule was applied and why; key calculation shown]

**Winner's curse adjustment (if common values):** [Revised estimate conditional on winning, and resulting corrected bid]

**Traps to avoid:** [Top 1-2 risks in this specific situation — overbidding, escalation, value type misclassification]

**Format equivalence note (if relevant):** [If the user is comparing formats or could choose, note revenue equivalence implications]

---

## Key Principles

**Format determines the rule.** The bid that is optimal in a Vickrey auction is catastrophic in a first-price sealed-bid auction. Identify the format before computing anything.

**Revenue equivalence theorem.** Under private values and symmetric bidders, English, Japanese, Vickrey, Dutch, and first-price sealed-bid auctions all yield the same expected seller revenue and the same expected winner — just through different mechanisms. This is why changing the surface rules of a game does not change outcomes: bidders adjust their strategies to offset exactly.

**In Vickrey, bid your value exactly.** This is the only setting where true-value bidding is a dominant strategy. The dominant strategy proof means you do not need to know anything about other bidders to confirm this.

**In first-price, shade your bid.** The formula is V × (N-1)/N under symmetric uniform beliefs. More competitors mean less shading; fewer competitors mean more shading.

**In common-value auctions, winning is bad news.** Winning means everyone else estimated lower. Update your estimate downward conditional on winning. The presume-you've-won diagnostic is the core correction tool.

**In dollar auctions and wars of attrition, the best time to exit was before entering.** Once in, sunk costs are irrelevant. Exit when the marginal continuation cost exceeds the marginal expected benefit — and be honest about the probability that the other side exits first.

**Buyer's premiums are borne by sellers, not buyers.** If an auction house adds a 20% buyer's premium, rational bidders adjust by bidding 1/1.2 of their true value. The hammer price falls enough to leave the winning bidder's total payment unchanged. The auction house's take comes out of the seller's proceeds.

---

## Examples

### Example 1: Vickrey Auction (Corporate Software License)

**Setup:** A government agency is auctioning a multi-year software contract to the lowest bidder. This is a second-price procurement auction (winner pays the second-lowest bid). Your cost to deliver is $800K. You estimate one strong competitor at roughly $700K-$900K.

**Apply the rule:** In a Vickrey (second-price) procurement auction, the dominant strategy is to bid your true cost exactly. Bid $800K.

**Why:** If you win, you pay the second-lowest bid (competitor's bid). If competitor bids $750K, you lose — correct, because they can do it cheaper. If competitor bids $850K, you win and get paid $850K for $800K work: $50K profit. Bidding below $800K (say $750K) might win but you'd be paid $750K or less for $800K work — a guaranteed loss. Bidding above $800K (say $850K) only changes the outcome if the competitor bids between $800K-$850K, in which case you lose wins you would have profited from.

**Recommended bid:** $800K (your true cost).

### Example 2: First-Price Sealed-Bid (Real Estate Offer)

**Setup:** You are in a competitive offer situation on a house. True value to you: $620,000. You believe there are 3 competing bidders, all with values roughly similar to yours.

**Apply the formula:** N = 4 (you plus 3 others). Optimal bid = $620,000 × (4-1)/4 = $620,000 × 0.75 = $465,000.

**Sanity check:** This seems very aggressive shading. In practice, real estate values are not uniformly distributed across [0, V] — they cluster near the asking price. The formula is exact only under symmetric uniform beliefs. Adjust: if you believe competing values cluster around $580K-$620K, the effective range is narrow and shading should be modest (perhaps $595K-$605K). The formula gives a floor on shading; judgment about competitor value concentration adjusts from there.

**Recommended bid:** $600,000 (approximately V × 0.97, given tight competitor value clustering) with a clear ceiling at $620,000.

### Example 3: Winner's Curse Correction (Company Acquisition)

**Setup:** Your team estimates a target company is worth $50M-$90M today. You can improve operations by 40%. You are in a sealed-bid acquisition process.

**Naive calculation:** Average value $70M × 1.4 = $98M. "I can bid up to $98M."

**Winner's curse analysis:** If accepted at $98M, current value is between $50M and $98M — average $74M. Your 40% improvement: $74M × 1.4 = $103.6M. Profit: $103.6M - $98M = $5.6M. Still slightly positive.

**Find the breakeven bid B:** Accepted at B → current value between $50M and B → average (50+B)/2. Your improvement: 1.4 × (50+B)/2 = B. Solve: 70 + 1.4B/2 = B → 70 + 0.7B = B → 70 = 0.3B → B = $233M.

**Wait — this is above the stated range.** This means within the range $50M-$90M, your 40% improvement always generates enough value to justify winning. The winner's curse is not binding here because your operational uplift is large. In this case, bid your full expected-value calculation up to $98M, but confirm your improvement assumptions — they are doing all the work.

**When the winner's curse is binding:** It binds when your improvement multiplier is small (say 1.05×) and the value range is wide. In that case, the accepted-bid calculation reveals expected losses.

---

## References

- `references/auction-format-taxonomy.md` — Full mechanics of all five formats with decision trees
- `references/bid-shading-formula.md` — Derivation of V×(N-1)/N, worked examples across N, and adjustments for non-uniform value distributions
- `references/winners-curse-worksheet.md` — Step-by-step presume-you've-won calculation template with ACME example fully worked
- `references/dollar-auction-escalation.md` — Dollar auction mechanics, war-of-attrition model, BSB/Sky case, exit criteria

## License

This skill is licensed under [CC-BY-SA-4.0](https://creativecommons.org/licenses/by-sa/4.0/).
Source: [BookForge](https://github.com/bookforge-ai/bookforge-skills) — The Art of Strategy by Avinash K. Dixit, Barry J. Nalebuff.

## Related BookForge Skills

This skill is standalone. Browse more BookForge skills: [bookforge-skills](https://github.com/bookforge-ai/bookforge-skills)

FILE:references/auction-format-taxonomy.md
# Auction Format Taxonomy

Full mechanics and decision flows for the five canonical auction formats.

---

## English Auction (Ascending, Open)

**Mechanics:**
- Auctioneer calls out successively higher prices
- Bidders signal participation (raise paddle, call out bid)
- Anyone can drop out at any time; once dropped, can re-enter in some implementations but typically cannot
- Auction ends when only one bidder remains
- Winner pays the price at which the last competitor dropped out (approximately the second-highest valuation)

**Strategic optimal:** Bid (stay in) until price reaches your true value; drop at that price.

**Information revealed:** Partial. In an English auction, you see others' bids but not their private valuations — a bidder might be silent and then make a surprise late bid. You know what others bid, but not how high they would have gone.

**Common value relevance:** High. In English auctions with common values, each dropout reveals a signal about the common value. Update your estimate as bidders leave.

---

## Japanese Auction (Clock Rising, Simultaneous Commitment)

**Mechanics:**
- All bidders begin with hands raised (or button pressed)
- A clock rises from a starting price; all active bidders remain "in" while their hand is raised
- A bidder drops out by lowering their hand; this is irreversible
- Auction ends when only one bidder remains
- Winner pays the price at which the last other bidder dropped out

**Strategic optimal:** Keep hand raised until clock reaches your true value; lower hand at exactly that price.

**Difference from English:** Full transparency. In a Japanese auction, you know exactly how many competitors remain at every price point and the exact prices at which each drops out. There are no hidden bids or surprise late entrants.

**Information revealed:** Complete. Everyone knows exactly when every bidder dropped out and at what price. This is the most information-rich ascending auction format.

**Strategic equivalence with English:** Under private values, the extra information is irrelevant — your value is independent of what others think. Both auctions produce the same outcome: winner with highest value pays the second-highest valuation.

---

## Vickrey Auction (Second-Price Sealed-Bid)

**Mechanics:**
- All bidders submit a sealed bid simultaneously
- Bids are revealed; highest bid wins
- Winner pays the second-highest bid (not their own)
- Other bidders pay nothing

**Strategic optimal:** Bid your exact true value. This is the unique dominant strategy — it is best regardless of what others bid and regardless of how many others are bidding.

**Dominant strategy proof:**

Let your value = V. Consider any alternative bid B ≠ V.

Case 1: Highest rival bid H > V. You lose with bid V; you also lose with bid B unless B > H, in which case you "win" but pay H > V — a loss. Bidding V weakly dominates.

Case 2: Highest rival bid H < V. You win with bid V and pay H < V (profit). If B > H you also win and pay H (same outcome). If B < H you lose (forfeited profit). Bidding V weakly dominates in this case too.

Case 3 (the decisive case): V < H < V... wait, this is impossible if H < V is case 2. The case that makes B ≠ V strictly worse: B < V and B < H < V → you lose with B, win with V (paying H < V), profit = V - H > 0. Bidding V strictly dominates B.

Symmetric argument shows bidding above V is also dominated. Therefore V is the unique dominant strategy.

**Real-world applications:** Treasury bills (uniform price variant), Google AdWords second-price variant, some procurement auctions.

---

## Dutch Auction (Descending Clock)

**Mechanics:**
- Clock starts at a high price and descends
- First bidder to stop the clock wins at that price
- All other bidders pay nothing and receive nothing
- (Aalsmeer Flower Auction: 14 million flowers/day, 160-acre facility, descending clock)

**Strategic optimal:** Determine your bid before the auction begins (as if writing a sealed bid). When the clock descends to that price, stop it immediately.

**Why strategic equivalence with sealed-bid first-price:**
- Your "bid" in a Dutch auction is the price at which you plan to stop the clock
- This plan is made in advance, without seeing others' bids
- The person who plans to stop at the highest price (has the highest sealed bid) wins
- The winner pays the price at which they stopped (their "bid")
- Identical to first-price sealed-bid in every strategically relevant sense

**The one difference:** In a Dutch auction, you know you've won when the clock stops. In a first-price sealed-bid auction, you find out later. But per the "bid as if you've won" principle, in a sealed-bid auction you should already be assuming you've won when computing your bid. This removes the only apparent difference.

---

## First-Price Sealed-Bid Auction

**Mechanics:**
- All bidders submit a sealed bid simultaneously
- Bids revealed; highest bid wins
- Winner pays their own bid
- Others pay nothing

**Strategic optimal:** Shade your bid below your true value. The amount of shading depends on how many competitors you expect.

**Bid-shading formula (symmetric private values, uniform distribution [0, V]):**

```
Optimal Bid = V × (N-1) / N
```

N = total number of bidders (including yourself).

| N (bidders) | Formula | Fraction of V to bid |
|-------------|---------|---------------------|
| 2 | V × 1/2 | 50% |
| 3 | V × 2/3 | 67% |
| 4 | V × 3/4 | 75% |
| 5 | V × 4/5 | 80% |
| 10 | V × 9/10 | 90% |
| 20 | V × 19/20 | 95% |

**Intuition:** With more competitors, the probability that you have the highest value is lower, so you must bid closer to your value to avoid being outbid. With fewer competitors, you can shade more aggressively because the next-highest bidder is likely to have a lower value than yours.

**Why never bid your true value in first-price:** If you win by bidding V, you pay V and profit exactly $0. You can always do better by bidding slightly less — you still win (since others bid even less), but you pay less. Bidding V is weakly dominated by V - ε for any ε > 0.

---

## Buyer's Premium (Rule Modification)

**What it is:** Auction houses (Sotheby's, Christie's) add a percentage premium to the winning bid. The buyer pays their bid plus, say, 20% on top. A $1,000 winning bid becomes a $1,200 payment.

**Who actually bears the cost:** The seller, not the buyer.

**Why:** Rational bidders know the premium exists. If your true value is $1,200 (the total amount you're willing to spend), you will bid $1,000 (knowing that $1,000 × 1.2 = $1,200). The hammer price falls in proportion to the premium. The winner's total payment stays the same. The auction house's cut comes from what would otherwise be the seller's proceeds.

**Generalizable principle (Revenue Equivalence):** Any change to the payment rule that bidders can anticipate and incorporate will be perfectly offset by corresponding changes in bidding behavior. The seller's expected revenue remains unchanged. Players adapt their strategies to undo rule changes.

FILE:references/bid-shading-formula.md
# Bid Shading Formula

Derivation of the optimal bid in first-price sealed-bid (and Dutch) auctions, with worked examples.

---

## Setup

**Assumptions for the symmetric baseline:**
- N bidders total (including yourself)
- Each bidder's value is drawn independently from a uniform distribution [0, 100] (or equivalently [0, V])
- All bidders know N and the distribution; each knows only their own value
- Symmetric equilibrium: all bidders use the same bidding function

---

## Deriving the Formula

**Step 1 — Expected payment in Vickrey equals bid in first-price (Revenue Equivalence)**

Revenue equivalence theorem guarantees that in equilibrium, both auctions yield the same expected payment from the winner. In a Vickrey auction with N symmetric bidders and uniform [0, V] values, the winner (highest-value bidder) pays the expected value of the second-highest valuation given that they have the highest.

**Step 2 — Compute the expected second-highest value conditional on having the highest**

If your value is v (and you have the highest value), the other N-1 bidders all have values below v. Their values are i.i.d. uniform [0, v]. The expected maximum of N-1 uniform [0, v] draws is:

```
E[max of N-1 uniform(0,v)] = v × (N-1)/N
```

**Step 3 — By revenue equivalence, your optimal first-price bid equals your expected payment in Vickrey**

```
Optimal bid = v × (N-1)/N
```

This is the optimal symmetric bid. Every bidder using this function produces a Nash equilibrium in the first-price auction.

---

## The Formula in Practice

**Formula:**
```
Bid = V × (N - 1) / N
```

Where:
- V = your true value (walkaway number)
- N = total number of bidders (including you)

**Worked examples:**

| Your Value | Bidders | Formula | Optimal Bid | Shading Amount |
|-----------|---------|---------|-------------|---------------|
| $100 | 2 | 100 × 1/2 | $50 | $50 |
| $100 | 3 | 100 × 2/3 | $66.67 | $33.33 |
| $100 | 4 | 100 × 3/4 | $75 | $25 |
| $100 | 5 | 100 × 4/5 | $80 | $20 |
| $100 | 10 | 100 × 9/10 | $90 | $10 |
| $60 | 2 | 60 × 1/2 | $30 | $30 |
| $60 | 4 | 60 × 3/4 | $45 | $15 |
| $200K | 3 | 200K × 2/3 | $133K | $67K |
| $1M | 5 | 1M × 4/5 | $800K | $200K |

---

## Adjustments for Non-Uniform Distributions

The formula V × (N-1)/N is exact only when competitor values are uniform [0, V]. In practice:

**When competitor values cluster near your value (competitive setting):**
- True optimal bid is closer to V than the formula suggests
- Shade conservatively (perhaps 5-10% off V rather than the formula's amount)
- Example: real estate competitive offers where all bidders are serious buyers cluster near market price

**When competitor values are sparse / few strong bidders:**
- Formula may understate how much to shade
- With only 1-2 serious competitors in a wide value range, shade more aggressively

**Rule of thumb for non-uniform settings:**
Use the formula as a floor — never bid above V × (N-1)/N in a first-price auction. Adjust upward toward V only when the distribution of competitor values is highly concentrated near your own.

---

## Why Bidding Your True Value Is Always Wrong in First-Price

**Algebraic proof:**

Suppose you bid exactly your value V. If you win, you pay V and profit = V - V = $0.

Now consider bidding B = V - ε for small ε > 0. You still win whenever your value is the highest (which is all that matters, since V is still your value). But now you pay B = V - ε and profit = V - (V - ε) = ε > $0.

Therefore bidding V is strictly dominated by bidding V - ε. QED.

**Bidding above V is even worse:** If you win, you pay more than V and profit < 0. This is dominated by not bidding at all.

---

## Revenue Equivalence Theorem

**Statement:** Under private values, symmetric bidders, and risk-neutrality, all four canonical auction formats (English, Japanese, Vickrey, Dutch/first-price) yield the same expected revenue to the seller and the same expected payment from the winner.

**Why it holds:** Bidders' strategies adjust to perfectly offset changes in format rules. If the seller imposes a buyer's premium, bidders shade down their bids proportionally. If the seller switches from second-price to first-price, bidders shade their bids down by exactly the right amount. The expected payment stays constant.

**Practical implication for sellers:** You cannot systematically increase revenue by choosing a different standard auction format. You can increase revenue by:
- Setting a binding reserve price (excludes low-value bidders, raises payment from the winner)
- Attracting more bidders (increases N, which by the formula tightens bid shading)
- Designing formats that reduce common-value uncertainty (reducing the winner's curse discount)

**Practical implication for bidders:** If you are in a format-equivalent auction, focus on getting your value estimate right, not on outguessing the format. The format doesn't change your expected payment.

FILE:references/dollar-auction-escalation.md
# Dollar Auction and Escalation Traps

Mechanics of escalation auctions, war-of-attrition dynamics, exit criteria, and avoidance strategies.

---

## The Dollar Auction (Shubik's Escalation Game)

**Setup:**
- An auctioneer offers a $1 bill for sale
- Standard ascending-bid rules — highest bidder wins the dollar
- One twist: **both the highest AND second-highest bidder pay their bids**
- The second-highest bidder pays but receives nothing

**Why it escalates:**

Once two bidders are active at bids of, say, $0.50 (second place) and $0.60 (first place):

The second-place bidder calculates: "I am going to lose $0.50 and get nothing. If I bid $0.70, I am the high bidder and will likely win $1.00 for a net gain of $0.30. My alternative is to lose $0.50 for certain."

This logic is locally rational at every step. But:
- If the second-place bidder bids $0.70, the previous leader is now at $0.60 (losing $0.60 for nothing) and faces the same calculation
- Bidding continues past $1.00 because at that point both sides have sunk more than the prize value — they bid to minimize total loss, not to profit

**Classroom result:** Bids regularly exceed $5 for a $1 bill. The auctioneer profits; all active bidders lose.

**Structural cause:** The combination of (1) sunk cost logic creating a trap and (2) no coordination between the two trapped bidders. Neither can unilaterally escape without absorbing a loss, but continuing escalation makes both worse off.

---

## War of Attrition (Business Context)

**Definition:** A competitive situation where both parties spend resources over time, the winner is the last one standing, and both pay their costs regardless of who wins.

**Structure as an auction:**
- Each party "bids" the total financial loss they are willing to absorb
- The party with the higher "bid" (more loss tolerance) wins
- Both parties pay their costs while the contest lasts

**BSB vs. Sky (1989-1990, UK Satellite TV):**

- BSB held the official license; Sky launched without one using a different satellite
- Both firms bid for exclusive content (Hollywood movies, sports rights) at inflated prices
- After 18 months: combined losses of £1.5 billion
- Murdoch understood BSB wouldn't fold easily. BSB's strategy was to outlast Murdoch (who was personally exposed financially)
- Resolution: merger forced by mutual exhaustion and the government's desire to preserve one functioning firm

**Why both continued despite mounting losses:**
1. Sunk costs are irrelevant to future decisions — the £600M already lost was gone either way
2. The prize (monopoly over UK satellite TV with potential £2B/year revenue) justified continued spending at each marginal decision point
3. Each side overestimated their ability to outlast the other — overconfidence plus no coordination mechanism

**Lesson:** The willingness to absorb losses determined the merger split. Murdoch's deeper pockets and higher personal financial exposure (credibly committed him to either win or go bankrupt) gave him leverage.

---

## Exit Criteria for Wars of Attrition

**The decision at each period:**

Stay if: Expected value of winning × probability of winning > Additional cost of staying one more period

Exit if: Additional cost of staying > Expected benefit

**The problem with this calculation:** Both parties may simultaneously believe that the other is "about to fold." There is no consistency check — both sides can hold this belief simultaneously and both be wrong.

**Mathematical formulation (optional depth):**

In the continuous-time war of attrition model, the equilibrium condition for staying in is:

```
p(t) + q(t) ≤ 1
```

Where p(t) is your probability of winning if you stay one more unit of time and q(t) is the rival's probability of winning if they stay one more unit. When p(t) + q(t) > 1, the expected costs exceed benefits and at least one party should exit.

**Practical interpretation:** If your probability of outlasting the rival is less than the fraction of the prize value consumed by one more period of costs, exit immediately.

---

## Decision Framework: Should You Enter?

**Before entering a competitive spending contest:**

1. **Calculate your maximum total loss.** In a dollar auction / war of attrition, the worst case is not "I lose the bid" — it is "I continue to a point where I have spent far more than the prize is worth." What is your hard cap?

2. **Assess credibility of commitment.** Can you credibly pre-commit to a ceiling bid? If your competitor knows you will walk at $X, they may choose not to enter or to stop before $X. Pre-commitment reduces escalation.

3. **Is there a coordinated exit available?** In BSB/Sky, the merger was the coordinated exit. Can you engineer one? Tacit coordination (both sides signaling through bid levels that a certain price is the ceiling) may be possible in multi-round auctions.

4. **Is the prize truly winner-take-all?** Wars of attrition are worst when the prize is indivisible. If the prize can be split (spectrum licenses across geographic areas, as in FCC auctions), there may be an implicit deal available where each side wins something.

---

## Avoidance Strategies

**Strategy 1: Do not enter.** The best outcome in a dollar auction is to not play. If you can observe that the auction structure creates escalation (both top bidders pay), avoid it entirely.

**Strategy 2: Pre-commit publicly to a ceiling.** "We will not spend more than $X to win this." If your ceiling is credible and below the prize value, you can sometimes deter competitors from entering, since they know the auction will be rational (you won't escalate past $X even to "recover" sunk costs).

**Strategy 3: Change the game.** If you are in a competitive spending war, look for ways to change the structure — partnership, merger, exit deal, or third-party arbitration. The BSB/Sky resolution was a merger at the eleventh hour; the ability to absorb losses determined the split.

**Strategy 4: Once in, treat sunk costs as zero.** At every decision point, the question is purely: does the expected future value of continuing exceed the marginal future cost? Prior losses are irrelevant. This is psychologically hard but analytically mandatory.

**The dollar auction trap in disguise:**

Many real business situations have dollar-auction structure without being called auctions:
- Competitive hiring wars (both companies spend on counter-offers; one wins, both pay)
- Patent litigation (both sides pay legal costs; winner gets the patent)
- Price wars to drive out competition (both lose margin; the survivor gets the market)
- Bidding wars for key hires or sports stars (both franchises pay escalating offers; one wins)

Recognize the structure. Once you name it as a dollar auction or war of attrition, the decision framework above applies.

FILE:references/winners-curse-worksheet.md
# Winner's Curse Worksheet

Step-by-step presume-you've-won calculation for common-value auctions.

---

## When to Use This Worksheet

Use whenever:
- The item has a single underlying value that all bidders will eventually realize (oil leases, company acquisitions, rights auctions, Treasury bills)
- Your estimate of that value is uncertain and potentially noisy
- Winning would reveal that you estimated higher than all competitors

Signs you are in a common-value setting:
- The item's value depends on facts that will eventually be known to everyone (oil quantity underground, future cash flows of a company)
- Other bidders are professional market participants who have done similar analysis
- You are estimating the same reality as your competitors, not valuing something according to personal taste

---

## The Presume-You've-Won Diagnostic

**The core reframing:**

Do NOT ask: "What do I think this item is worth?"

DO ask: "If my bid is accepted / if I win this auction, what does that tell me about the item's actual worth?"

Winning is informative. In a competitive common-value auction, winning means you estimated higher than everyone else. That is systematic, not random. The winner's estimate is biased upward relative to the true value.

---

## Worksheet: Step-by-Step Calculation

### Step 1 — Establish Your Prior Range

What is your estimated range for the item's current value?

- Low estimate: $______
- High estimate: $______
- Distribution assumption: __________ (uniform is simplest; use if no reason to weight one end)
- Prior average value: $________ (for uniform: (low + high) / 2)

### Step 2 — Identify Your Value Multiplier

What improvement or extraction advantage do you bring over the baseline?

- Your multiplier: ________ (e.g., 1.5 = you can improve value by 50%)
- Naive maximum bid: Prior average × multiplier = $________

*This is what most people bid. It is usually wrong.*

### Step 3 — Apply the Presume-You've-Won Correction

Assume your bid B is accepted. Conditional on acceptance, the item's current value is somewhere between [Low, B], not [Low, High].

For uniform distribution: expected current value conditional on acceptance = (Low + B) / 2

The value you will realize from the item: Multiplier × (Low + B) / 2

For the deal to be at least breakeven, the value you realize must equal your bid:

```
Multiplier × (Low + B) / 2 = B
```

Solve for B:

```
Multiplier × Low / 2 + Multiplier × B / 2 = B
Multiplier × Low / 2 = B - Multiplier × B / 2
Multiplier × Low / 2 = B × (1 - Multiplier/2)
B = (Multiplier × Low / 2) / (1 - Multiplier/2)
```

Or equivalently:

```
B = (Multiplier × Low) / (2 - Multiplier)
```

*This is your maximum breakeven bid. Bid less than this to expect a profit.*

---

## Worked Example: ACME Acquisition

**Given:**
- Value range: $2M to $12M (uniform)
- Your multiplier: 1.5 (50% improvement)
- Low = $2M

**Naive bid:** Average ($7M) × 1.5 = $10.5M

**Why $10.5M is wrong:**
If accepted at $10.5M, current value is $2M to $10.5M (average $6.25M). Your improvement: 1.5 × $6.25M = $9.375M < $10.5M. Expected loss: $1.125M.

**Correct calculation using the formula:**

```
B = (1.5 × 2) / (2 - 1.5) = 3.0 / 0.5 = $6M
```

**Verify:** Bid $6M. If accepted, current value is $2M to $6M (average $4M). Improvement: 1.5 × $4M = $6M = bid. Exactly breakeven. ✓

**Practical bid:** Bid less than $6M to expect profit (e.g., $5M to expect $500K profit on average).

---

## Worked Example: Oil Lease Auction

**Given:**
- Your seismic survey estimates 1M to 5M barrels at $20/barrel net value
- Range: $20M to $100M
- Your drilling efficiency advantage: 1.1× (10% better than average)
- Low = $20M

**Naive bid:** Average ($60M) × 1.1 = $66M

**Winner's curse check:**

```
B = (1.1 × 20) / (2 - 1.1) = 22 / 0.9 = $24.4M
```

**Why this is so low:** The 10% improvement multiplier is barely above 1.0. In this case the winner's curse is severe — your advantage is too small to overcome the adverse selection from winning. You would need to bid below $24.4M to expect any profit.

**At $66M (naive bid):** If accepted, current value is $20M to $66M (average $43M). Your value: 1.1 × $43M = $47.3M << $66M. Expected loss: ~$18.7M.

**Lesson:** When your improvement multiplier is close to 1.0 (little differentiation from rivals), the winner's curse can make common-value auctions unprofitable at any reasonable bid. Consider whether to participate at all.

---

## The "Conditional on Acceptance" Intuition

Think of the seller as having private information about the true value. They accept your bid only when you are paying more than the item is worth to them. The act of acceptance is bad news.

This is the same structure as:
- A used car seller accepting your offer (they know more about the car's condition than you)
- A counterparty accepting your M&A price (they know more about the company's liabilities)
- A homeowner accepting your offer in a falling market (they know more about structural issues)

In every case: if they said yes, ask yourself why. The presume-you've-won analysis forces you to answer that question before, not after, committing.

---

## Common-Value Adjustments in English/Japanese Auctions

In ascending auctions with common values:

1. **Each dropout is a signal.** When a competitor lowers their hand at price P, they are revealing that their private estimate is approximately P (or they updated their estimate downward to P). This is information you did not have before.

2. **Update continuously.** As more bidders drop out at prices lower than you expected, revise your estimate of the common value downward. Early mass dropouts are stronger signals than gradual ones.

3. **The last surviving competitor's dropout price is the strongest single signal.** The price at which the second-to-last bidder drops out tells you the second-highest private estimate. This is valuable: if you had estimated $100K but the second-to-last competitor drops at $70K, your estimate should move toward $70K-$80K range.

4. **In a Japanese auction, this information is more complete.** You know every dropout price, not just the final one. Use all of them as a Bayesian update on the common value.

---
name: target-market-selection-pvp-index
description: "Use this skill to select the ideal target market or customer
  segment for a small business using the PVP Index (Personal fulfillment,
  Value to marketplace, Profitability). Triggers when a user asks to choose a
  target market, pick a niche, identify their ideal customer, score market
  segments, find the best customers to focus on, decide which customer type to
  target, narrow marketing focus, build a customer avatar, define an ideal
  customer profile, stop trying to serve everyone, or fill square #1 of the
  1-Page Marketing Plan. Also activates for 'who should I market to', 'how do
  I choose a niche', 'which customers are most profitable', 'I serve too many
  segments', or similar market-selection questions."
version: 1.0.0
homepage: https://github.com/bookforge-ai/bookforge-skills/tree/main/books/the-1-page-marketing-plan/skills/target-market-selection-pvp-index
metadata: {"openclaw":{"emoji":"📚","homepage":"https://github.com/bookforge-ai/bookforge-skills"}}
status: published
source-books:
  - id: the-1-page-marketing-plan
    title: "The 1-Page Marketing Plan"
    authors: ["Allan Dib"]
    chapters: [1]
tags:
  - marketing
  - target-market
  - segmentation
  - small-business
  - niche
  - customer-avatar
depends-on: []
execution:
  tier: 1
  mode: full
  inputs:
    - type: document
      description: >
        Business description including current products/services and the
        market segments the business currently serves (or is considering).
  tools-required: [Read, Write]
  tools-optional: [Grep]
  mcps-required: []
  environment: >
    Document set — business description and market research notes in markdown.
    No code execution required.
discovery:
  goal: >
    Help the user select their ideal target market segment using the PVP Index
    scoring framework, producing a ranked segment table, a selected primary
    target, and a customer avatar.
  tasks:
    - "Score candidate market segments on Personal fulfillment (P), Value to
       marketplace (V), and Profitability (P)"
    - "Rank segments by total PVP score (max 30)"
    - "Select primary target market (highest score)"
    - "Resolve ties using secondary criteria"
    - "Build a customer avatar for the selected segment"
    - "Write target-market.md with full output"
  audience:
    roles:
      - small-business-owner
      - solopreneur
      - entrepreneur
      - freelancer
      - startup-founder
    experience: beginner-to-intermediate
  when_to_use:
    triggers:
      - "User wants to choose a target market or niche"
      - "User serves too many segments and wants to focus"
      - "User needs to define an ideal customer profile"
      - "User is filling square #1 of the 1-Page Marketing Plan"
      - "User asks which customers are most profitable or enjoyable"
    prerequisites: []
    not_for:
      - "Enterprise marketing with large teams and budgets (use rigorous
         market research methodologies instead)"
      - "Businesses that have already validated a clear, focused niche and
         are happy with it"
  environment:
    codebase_required: false
    codebase_helpful: false
    works_offline: true
  quality:
    scores:
      with_skill: 86
      baseline: 7
      delta: 79
    tested_at: "2026-04-09"
    eval_count: 1
    assertion_count: 14
    iterations_needed: 1
---

# Target Market Selection with the PVP Index

A structured process for small business owners to stop marketing to everyone
and start marketing to the right customers. Produces a scored segment
comparison table, a primary target market decision, and a customer avatar
— the required content for square #1 of the 1-Page Marketing Plan canvas.

The PVP Index (attributed to Frank Kern, adapted by Allan Dib) scores each
market segment on three dimensions (each 0–10):

- **P** — Personal Fulfillment: How much do you enjoy working with this type?
- **V** — Value to Marketplace: How much does this segment value your work?
- **P** — Profitability: How profitable is the engagement after all costs?

Total out of 30. Highest total = primary target for all marketing.

---

## When to Use

Use this skill at the START of any marketing effort — before writing ads,
choosing media, or crafting offers. It is the foundation because every
downstream marketing decision (message, channel, offer, lead magnet) depends
on knowing exactly who you are speaking to.

Also use it when:
- A business is spreading its marketing budget across too many segments
  and getting mediocre results everywhere
- A business owner feels they serve "everyone" and doesn't know where to
  focus
- Revenue exists but profitability is unclear (some segments may be
  loss-making despite high fees)
- The business is about to launch a new marketing campaign

Do NOT use this skill as a substitute for business strategy or product-market
fit validation. It assumes the business already has services to offer — it
selects *who to market them to*, not whether the business itself is viable.

---

## Context & Input Gathering

### Required (must have before proceeding)
- A description of the business: what it does, what services/products it
  offers
- A list of the current or candidate market segments (at least 2; ideally
  3–5)

### Observable / inferrable
- Industry and service type (often apparent from the description)
- Rough profitability (can be estimated from typical pricing and overhead)

### Defaults (apply if not provided)
- If the user can't name 3+ segments, prompt: "Think about the different
  types of customers you've served or want to serve. What are the main
  categories?"
- If profitability data is unavailable, ask the user to estimate relative
  profitability (high / medium / low) and convert to a 0–10 scale

### Sufficiency check
You have enough to proceed when you can name at least 2 segments and have
enough context to score all three PVP dimensions for each. If unsure about
personal fulfillment or profitability, ask directly — these are the dimensions
most often overlooked and most often decisive.

---

## Process

### Step 1: List candidate segments

Ask the user to name every distinct market segment they currently serve or
are considering. Record them as a simple list.

**WHY:** Without a named list, scoring cannot happen. Segments often blur
together in a business owner's mind; naming them forces clarity. Aim for
3–5 segments — fewer than 2 makes comparison trivial; more than 6 becomes
unwieldy.

### Step 2: Score each segment on Personal Fulfillment (P, 0–10)

For each segment, ask: "How much do you genuinely enjoy working with this
type of customer?" A score of 10 means you look forward to every engagement;
1 means you dread the work or the client relationship.

**WHY:** Most scoring frameworks (like ICE or RICE) omit personal fulfillment
entirely. Dib includes it because a segment you hate serving — even if
lucrative — is unsustainable. You will unconsciously underserve them, burn
out, or avoid marketing to them. Low P scores flag hidden long-term costs
that profit calculations miss.

### Step 3: Score each segment on Value to Marketplace (V, 0–10)

For each segment, ask: "How much does this type of customer value and pay for
your work?" A score of 10 means they actively seek specialists, pay premium
prices without complaint, and perceive high value. A score of 1 means they
price-shop aggressively or view your service as a commodity.

**WHY:** High value perception is what allows premium pricing. Segments that
don't perceive the value of your service will always push back on price,
require excessive justification, and create adversarial relationships. V
scores predict how price-sensitive future negotiations will be.

### Step 4: Score each segment on Profitability (P, 0–10)

For each segment, ask: "After all direct costs — time, materials, overhead
allocated to this segment — how profitable is a typical engagement?" Score
relative to your best-case scenario. High fees do not mean high profit;
account for delivery costs, revisions, support burden, and churn.

**WHY:** It is possible to be busy and broke. Dib's key distinction is
"turnover vs. left over." A segment generating $50K in revenue but $45K in
costs scores very low here. This dimension forces owners to examine actual
margin, not top-line revenue — which is often the insight that changes
everything.

### Step 5: Calculate totals and build the ranking table

Sum P + V + P for each segment. Rank from highest to lowest.

**WHY:** The total makes trade-offs visible. A segment that scores 8/3/9 (20)
versus one that scores 7/7/7 (21) reveals different problems — the first is
personally rewarding and profitable but undersells value; the second is
balanced. Seeing these numbers side by side prevents the most common error:
choosing based on gut feel alone.

### Step 6: Select the primary target market

**IF** one segment has a clearly higher total (3+ points above second place):
  → Select it as primary. Proceed to Step 7.

**IF** two segments are within 2 points of each other (a near-tie):
  → Apply the tie-breaker sequence:
  1. Which segment has the higher V score? (Value perception determines
     long-term pricing power — a critical lever for small businesses)
  2. Which segment has the larger addressable population in your geography?
  3. Which segment is easier to reach via your existing channels?
  → Select the segment that wins 2 of 3 tie-breakers. Document the
     reasoning.

**WHY:** Without an explicit decision rule, owners default to the segment they
are most comfortable with — which is often not the most strategic choice. The
tie-breaker sequence uses objective criteria rather than comfort.

### Step 7: Build the customer avatar

For the selected primary segment, construct a detailed profile. Cover:

**Demographics:**
- Age range, gender (if relevant), geography
- Job title / business type / role
- Income or business revenue range

**Psychographics (the critical layer):**
- What keeps them awake at night, worrying?
- What are they most afraid of?
- What frustrates them daily?
- What do they secretly want most?
- What is the dominant emotion this market lives with?

**Behavioral / contextual:**
- What publications, websites, or social feeds do they consume?
- What is a typical day like?
- Do they use industry-specific jargon?
- Are there built-in decision biases (e.g., highly analytical, risk-averse)?
- Who else is involved in the purchase decision? (If yes, create a second
  avatar for that influencer/gatekeeper)

**WHY:** The avatar turns an abstract segment into a specific person. All
downstream marketing — ads, offers, headlines, nurture emails — is written
to this person, not to a demographic bracket. Without a vivid avatar, copy
becomes generic and fails to trigger the "hey, that's for me" response that
makes direct-response marketing work.

### Step 8: Write target-market.md

Save the full output (scoring table + selection rationale + avatar) as
`target-market.md` in the user's working directory.

**WHY:** Square #1 of the 1-Page Marketing Plan must be documented to anchor
every subsequent marketing decision. Without a written record, the target
market selection reverts to vague memory and gets overridden under business
pressure ("maybe we should also try to reach X...").

---

## Inputs

| Input | Format | Required |
|-------|--------|----------|
| Business description | text / .md | Yes |
| Current or candidate market segments | list in text | Yes |
| Typical pricing per segment | rough estimate | Recommended |
| Gross cost/overhead per engagement | rough estimate | Recommended |
| Personal fulfillment (owner self-report) | 0–10 per segment | Gathered in Step 2 |

---

## Outputs

Primary output: `target-market.md`

```markdown
# Target Market: [Business Name]

## PVP Index Scoring

| Segment | Personal (P) | Value (V) | Profit (P) | Total /30 |
|---------|:------------:|:---------:|:----------:|----------:|
| [Seg A] | X            | X         | X          | XX        |
| [Seg B] | X            | X         | X          | XX        |
| [Seg C] | X            | X         | X          | XX        |

**Primary Target Market: [Segment Name] — Score: XX/30**

Selection rationale: [1–2 sentences explaining why this segment won,
including any tie-breaker logic applied]

---

## Customer Avatar: [Avatar Name]

**Demographics**
- Age: ...
- Role / business type: ...
- Geography: ...
- Revenue / income: ...

**Psychographics**
- Biggest fear: ...
- Keeps them awake: ...
- Daily frustration: ...
- Secretly desires: ...
- Dominant emotion: ...

**Behavior & Context**
- Reads / watches: ...
- Typical day: ...
- Decision biases: ...
- Jargon they use: ...

**Decision-Making Unit**
- Primary decision-maker: [Avatar Name]
- Influencer / gatekeeper: [Second avatar if applicable]

---

_Square #1 of the 1-Page Marketing Plan canvas: filled._
```

---

## Key Principles

**1. Niching makes price irrelevant.**
A specialist is sought out; a generalist is price-shopped. When you dominate
a specific niche, prospects come to you because you are the obvious expert for
their situation — the way a heart surgeon is not compared to a GP on fees. Do
not fear the narrow niche. Fear the broad, undifferentiated positioning.

**2. An inch wide, a mile deep — then expand.**
Start by dominating one highly specific segment. Once you own that niche,
expand to another. Trying to own multiple niches simultaneously dilutes
budget, message, and credibility. Growth comes from serial dominance, not
parallel mediocrity.

**3. Personal fulfillment is a business metric, not a luxury.**
Most frameworks optimize for revenue or margin and ignore whether you enjoy
the work. Low personal fulfillment is a hidden cost: it produces worse
delivery, higher churn, and owner burnout. A segment that scores 9 on profit
but 2 on fulfillment is not a good long-term bet.

**4. Profitability ≠ revenue.**
High fees from a high-effort, high-overhead segment can result in negative
margins. Always score profitability as "left over," not "turnover." The
segment that appears least lucrative by fee rate may be most profitable in
practice.

**5. Each service category needs its own campaign.**
A photographer who does weddings and corporate work can serve both — but each
requires a completely separate ad targeting a completely separate audience. A
single laundry-list ad speaks to neither. The PVP output tells you which
campaign to build first.

---

## Examples

### Example 1: Photographer (canonical book example)

**Scenario:** A freelance photographer serves four segments — Weddings,
Photojournalism, Corporate, and Family Portraits — and wants to know where
to focus marketing spend.

**Trigger:** "I do photography for four different markets and my marketing
isn't working. Where should I focus?"

**PVP Scoring:**

| Segment          | Personal (P) | Value (V) | Profit (P) | Total |
|------------------|:------------:|:---------:|:----------:|------:|
| Weddings         | 5            | 7         | 9          | 21    |
| Photojournalism  | 9            | 7         | 2          | 18    |
| Corporate        | 3            | 6         | 9          | 18    |
| Family Portraits | 9            | 8         | 9          | **26** |

**Output:** Family Portraits wins decisively — the only segment that scores
high on all three dimensions. Photojournalism scores high on fulfillment but
is barely profitable (equipment, time, low publication fees). Weddings are
profitable but the photographer finds them stressful. Corporate pays well but
the work is joyless.

**Avatar:** Sarah, 34, married with one toddler. Feels this is a once-in-a-
decade moment that will be on the walls of her home for decades. Her dominant
emotion is anticipation mixed with anxiety about capturing the moment
perfectly. She reads parenting blogs and Pinterest. She wants an emotional
story, not a technical portfolio.

---

### Example 2: Independent Management Consultant

**Scenario:** A consultant serves startups (project-based, $5K–$15K
engagements), mid-market companies ($20K–$60K retainers), and non-profits
(grants-funded, $8K–$20K). She finds startup work energizing but chaotic;
non-profit work fulfilling but underpaid; mid-market work steady but
bureaucratic.

**Trigger:** "I'm spread thin across three client types. Which should I focus
my marketing on?"

**PVP Scoring:**

| Segment         | Personal (P) | Value (V) | Profit (P) | Total |
|-----------------|:------------:|:---------:|:----------:|------:|
| Startups        | 8            | 5         | 4          | 17    |
| Mid-Market      | 4            | 8         | 9          | 21    |
| Non-Profits     | 7            | 3         | 3          | 13    |

**Output:** Mid-Market wins (21). Startups are energizing but score low on
both value perception and profitability — startups are budget-constrained,
slow to pay, and often require extensive scope-creep management. Non-profits
are personally rewarding but chronically undervalue consulting fees. Mid-
market companies have budget authority, respect expertise, and generate
reliable margins.

**Tie note (if Startups had scored 20):** The tie-breaker would favor
Mid-Market on V score (8 vs 5) — mid-market clients have higher value
perception and are less likely to push back on fees, which is the more
sustainable long-term lever.

**Avatar:** Marcus, 48, VP of Operations at a 200-person manufacturing
company. His dominant frustration: processes that worked at 50 people are
breaking at 200. He reads Harvard Business Review. His biggest fear is being
seen as the bottleneck that's holding the company back. He wants a consultant
who has "seen this exact problem before."

---

### Example 3: Quick-turn (user provides segment list)

**Scenario:** A web design agency owner says: "I serve real estate agents,
restaurants, law firms, and e-commerce brands. Real estate pays $2K–$5K but
I find it boring. Restaurants pay $1K and churn fast. Law firms pay $5K–$15K
but are very demanding and slow to approve work. E-commerce pays $3K–$8K
and I enjoy the work."

**Trigger:** "Which should I focus on?"

**Process:** Translate owner statements to scores:
- Real estate: P=3 (boring), V=6 (pay okay, not premium), P=6 (decent margin)
  → 15
- Restaurants: P=5 (neutral), V=3 (low price, high churn), P=2 (thin margin)
  → 10
- Law firms: P=4 (demanding = friction), V=8 (high fees, respect expertise),
  P=7 (good margin despite slow pace) → 19
- E-commerce: P=8 (enjoyable), V=6 (pay well, value speed), P=7 (good margin)
  → 21

**Output:** E-commerce wins (21). Law firms are a strong second and worth
targeting once e-commerce is established. Restaurants should be deprioritized
or dropped from marketing entirely.

---

## References

- `book-profile.json` — Full book metadata and terminology mappings
- `hunter-report.md` — sk-01 entry: PVP Index found by 5 hunters, density 5
- Research summary: `.meta/research/target-market-selection-pvp-index.md`
  (complete scoring tables, avatar questions, anti-patterns, Max Cash and
  Angela Assistant avatar examples)

## License

This skill is licensed under [CC-BY-SA-4.0](https://creativecommons.org/licenses/by-sa/4.0/).
Source: [BookForge](https://github.com/bookforge-ai/bookforge-skills) — The 1-Page Marketing Plan by Allan Dib.

## Related BookForge Skills

No direct dependencies. Install the full book set from GitHub.

Or install the full book set from GitHub: [bookforge-skills](https://github.com/bookforge-ai/bookforge-skills)

---
name: sales-conversion-trust-system
version: 1.0.0
homepage: https://github.com/bookforge-ai/bookforge-skills/tree/main/books/the-1-page-marketing-plan/skills/sales-conversion-trust-system
metadata: {"openclaw":{"emoji":"📚","homepage":"https://github.com/bookforge-ai/bookforge-skills"}}
status: published
quality:
  scores:
    with_skill: 92.9
    baseline: 0
    delta: 92.9
  tested_at: "2026-04-09"
  eval_count: 1
  assertion_count: 14
  iterations_needed: 1
description: >
  Use when ready to convert nurtured leads into paying customers. Triggers on:
  "sales conversion", "close sales", "convert leads", "outrageous guarantee",
  "risk reversal", "3-tier pricing", "tiered pricing", "try before you buy",
  "free trial design", "trust signals", "testimonials", "customer objections",
  "fill square 6", "weak satisfaction guarantee", "pricing structure",
  "reduce sales friction", "everyone is in sales". Produces a sales-conversion.md
  document with a trust checklist, guarantee statement, 3-tier pricing menu,
  try-before-you-buy mechanic, and friction audit.
source-books:
  - title: "The 1-Page Marketing Plan"
    author: Allan Dib
    chapters: [6]
    pages: "163-188"
tags: [marketing, sales, conversion, guarantees, pricing, small-business]
depends-on:
  - lead-nurture-sequence-design
  - irresistible-offer-builder
output-files:
  - sales-conversion.md
---

# Sales Conversion Trust System

Convert nurtured leads into customers using a trust-first system: manufacture
credibility signals, craft an outrageous guarantee that addresses real fears,
design 3-tier pricing with an ultra-high-ticket anchor, add a try-before-you-buy
mechanic, and remove friction so saying yes is effortless.

## When to Use

Use this skill after leads have been nurtured through an education-based
sequence and are showing buying intent — requesting quotes, asking questions,
engaging with your content. The goal at this stage is never to hard-sell; it is
to remove the remaining barriers of distrust, commitment anxiety, and inertia.

If leads are arriving cold (no nurture) and you are having to convince or push
them hard, the conversion process has been entered too early. In that case:
- IF leads are not yet being nurtured → invoke `lead-nurture-sequence-design`
- IF the offer itself is not yet defined → invoke `irresistible-offer-builder`
  OR ask the user to describe their current offer and nurture state before
  proceeding.

## Context and Input Gathering

Before running the process, collect:

1. **Current sales process** — how do prospects currently move from interest to
   purchase? What happens at the handoff point?
2. **Top 3 customer objections or fears** — what do prospects say, ask, or
   hesitate about most? (Source these from sales calls, support tickets, reviews)
3. **Current pricing structure** — single price, multiple tiers, or none yet?
4. **Nurture status** — are prospects arriving warm (educated, pre-motivated) or
   cold? If cold, flag and redirect as above.
5. **Existing guarantee** — does one exist? What does it say?
6. **Payment and checkout friction** — what methods are accepted? What forms or
   steps are required to buy?

## Process

### Step 1 — Verify Nurture Is Working

Check whether the nurture sequence has done its job. Well-nurtured leads should
arrive pre-framed, pre-motivated, and pre-interested — essentially asking to buy.
Signs the nurture has failed: prospects need heavy convincing, price objections
dominate, ghosting is common after a first positive interaction ("hopeium").

**Why:** If you are hard-selling at this stage, the problem is upstream, not at
conversion. Fixing the conversion process without fixing nurture is patching the
wrong pipe.

### Step 2 — Audit Trust Signals (Trust Manufacture Checklist)

Work through the checklist below and mark each item as present, absent, or needs
improvement. Small businesses start behind large ones in perceived trustworthiness
— technology and professional presentation can close this gap immediately.

**Website:**
- [ ] Phone number prominently listed at top of every page
- [ ] Physical business address displayed (not a PO Box; use a virtual office if
      working from home)
- [ ] Privacy policy and terms of use page present
- [ ] Professional design — no cheap or amateurish templates
- [ ] Business-domain email address (not Gmail or Hotmail)

**Social proof:**
- [ ] Testimonials from real customers (with names, not anonymous)
- [ ] Case studies showing before/after outcomes
- [ ] Media mentions or third-party endorsements
- [ ] Reviews on relevant platforms

**Credentials and team:**
- [ ] Certifications, accreditations, or industry memberships displayed
- [ ] Years in business stated
- [ ] Real team photos (not stock photography)
- [ ] Named team members publicly listed

**Operations:**
- [ ] Ticketing or support system so requests are trackable (reduces fear of
      falling into a black hole)
- [ ] CRM in use for consistent follow-up

**Why:** Prospects are guilty-until-proven-innocent in their relationship with
any unfamiliar business. Every absent trust signal is a reason to delay or defect.
These signals are visible in seconds; the prospect never asks for them explicitly
but will disqualify you if they are missing.

### Step 3 — Identify the Prospect's Top 3 Fears

Brainstorm or research the specific fears and uncertainties your prospects hold
about buying from you. Do NOT use generic fears. Source them from:
- Lost-deal conversations
- Support tickets and FAQs
- Competitor reviews on third-party platforms
- Your own sales call notes

Write them down explicitly:
1. Fear 1: ___
2. Fear 2: ___
3. Fear 3: ___

**Why:** A guarantee that addresses fears no one actually has is worthless. The
guarantee must speak to the specific anxieties in the prospect's mind at the
moment of commitment — not your assumptions about what they might worry about.

### Step 4 — Draft an Outrageous Guarantee

**Anti-pattern to avoid:** "Satisfaction guaranteed" and "money-back guarantee"
are weak, vague, and ignored. They are so overused they register as noise. They
do not address any named fear and therefore reduce no real anxiety.

**The outrageous guarantee formula:**
1. Name a specific fear or set of fears from Step 3
2. Promise a specific, measurable outcome that neutralises each fear
3. State the consequence you bear if you fail to deliver it
4. Make the consequence meaningful (double-credit, full refund, free service)

**IT company example (verbatim from source):**
> "We guarantee that our certified and experienced IT consultants will fix your
> IT problems so they don't recur. They'll also return your calls within fifteen
> minutes and will always speak to you in plain English. If we don't live up to
> any of these promises, we insist that you tell us and we'll credit back to your
> account double the billable amount of the consultation."

**Pest control example (verbatim from source):**
> "We guarantee to rid your home of ants forever, without the use of toxic
> chemicals, while leaving your home in the same clean and tidy condition we
> found it. If you aren't absolutely delighted with the service provided, we
> insist that you tell us and we'll refund double your money back."

Note the structure of both: named fears → specific promise per fear → named
consequence if broken. Neither says "satisfaction guaranteed."

**On abuse risk:** Even after accounting for the small number of people who
abuse a strong guarantee, you are far ahead — a strong guarantee attracts more
customers than a weak one loses to abuse. If your guarantee terrifies you to
write, that is a signal to improve your delivery quality, not to weaken the
guarantee.

**Why:** Risk reversal shifts perceived risk from the buyer to the seller. When
the seller bears the cost of failure, the prospect's decision becomes much
easier. It also forces internal delivery quality improvement.

### Step 5 — Design 3-Tier Pricing (with Ultra-High-Ticket Anchor)

Offer exactly three tiers. Never a single price. Never more than three or four
options (the Columbia jam study: 30% of buyers purchased from 6 options vs 3%
from 24 options — choice overload prevents any decision).

**Tier structure:**
- **Basic** — core deliverable only; entry point for price-sensitive buyers
- **Standard** — core deliverable plus meaningful additions; ~30% of buyers will
  choose this or above; price it as your main revenue engine
- **Premium** — standard plus the highest-value additions you can deliver; price
  at ~50% above standard but deliver twice or more the value

**Ultra-high-ticket anchor:**
Add a fourth item (or make Premium the anchor) priced at 10x or 100x the
Standard. Rule of thumb: 10% of your customer base would pay 10x more; 1% would
pay 100x more. Even selling one ultra-high-ticket item per month can make up a
large percentage of net profit.

The anchor also serves a psychological function: it makes Standard look
reasonably priced by comparison. Ultra-high-ticket buyers also attract
prestige-oriented customers who shop on service and convenience, not price.

**"Unlimited" variation (risk reversal for high-volume buyers):**
For services with variable usage anxiety (data, consulting hours, support calls),
offer an unlimited tier at a fixed monthly fee. People overestimate how much they
will use a service at point of purchase; the unlimited option removes the fear
of surprise charges while costing you very little on average.

**Never discount.** Unless you have an explicit loss-leader strategy, resist the
urge. Discounting damages positioning and margin. Instead, increase value:
bundle bonuses, add services, increase quantity.

**Why:** Price is a positioning signal, not just a number. A single price gives
buyers no choice architecture. Three tiers create anchoring, upsell paths, and
the illusion of control. Ultra-high-ticket items are pure profit and attract a
different, higher-quality customer segment.

### Step 6 — Add a Try-Before-You-Buy Mechanic

Design one of:
- Free first consultation or discovery session
- 30-day free trial with full feature access
- Free first month of service
- "Puppy dog" delivery — send the product first, invoice later, prospect must
  actively return it to cancel

**Why this works:** Try-before-you-buy breaks down commitment anxiety (it
feels reversible), but inertia then works in your favour. Once the prospect has
the product or has experienced the service, returning it requires active effort.
A genuine customer who is getting value will almost never make that effort. The
burden of reversing the sale passes to the buyer.

**Why:** Reducing perceived commitment risk at the point of entry dramatically
increases conversion without requiring any pressure tactics.

### Step 7 — Train Everyone in Sales

Make it explicit to every staff member that sales are the lifeblood of the
business and that everyone — not just the sales department — is in sales.

- Share the cues that signal a buying opportunity (requests, questions, usage
  patterns, service interactions)
- Create an incentive programme: sales get rewarded regardless of which staff
  member identified the opportunity
- The easiest sale is to an existing satisfied customer; equip all staff to
  recognise these moments

**Why:** The BMW service clerk who failed to offer a test drive when the
customer asked to borrow a newer model illustrates what happens when staff
think sales is "not my job." Every touchpoint is a potential conversion
moment; untrained staff close those doors silently.

### Step 8 — Remove Friction (Close the Sales Prevention Department)

Audit every step between "I want to buy" and "purchase complete." Any step that
makes yes harder or no easier is a friction point. Common offenders:

- [ ] Accepting only cash or refusing certain card types
- [ ] Surcharges for credit card or preferred payment methods
- [ ] Long or unnecessary sign-up forms
- [ ] Processes designed around your convenience, not the buyer's
- [ ] No payment plan or finance option for high-ticket items

**Why:** Prospects have already decided to buy. Every unnecessary step introduces
doubt, delay, and dropout. Factor payment processing fees into your prices and
absorb them — "Cash Only" signs are stepping over dollars to pick up pennies.

### Step 9 — Produce the sales-conversion.md Document

Compile all outputs from steps above into a single reference document.

## Inputs

| Input | Source |
|-------|--------|
| Top 3 customer fears/objections | Sales calls, support tickets, reviews |
| Existing pricing structure | Internal records |
| Current guarantee copy | Website, sales materials |
| Payment methods and checkout flow | Website / ops team |
| Nurture sequence status | `lead-nurture-sequence-design` output |
| Offer definition | `irresistible-offer-builder` output |

## Outputs

**sales-conversion.md** containing:

1. **Trust checklist** — all 14 signals with status (present / absent / needs
   work) and priority order for fixing absent items
2. **Guarantee statement** — final guarantee copy using the outrageous guarantee
   formula, with named fears, specific promises, and named consequence
3. **3-tier pricing menu** — Basic / Standard / Premium with price points,
   feature list per tier, ultra-high-ticket anchor item, and optional unlimited
   variant
4. **Try-before-you-buy mechanic** — chosen mechanic with implementation steps
   and inertia rationale
5. **Friction audit** — list of identified friction points with recommended
   removals and payment method coverage

## Key Principles

**Trust is manufactured, not claimed.** Generic claims ("best quality",
"excellent service") register as noise. Trust comes from specific, verifiable
signals: named staff, real photos, support systems, credentials, and testimonials
with real attribution.

**Specific beats vague in every guarantee.** "Satisfaction guaranteed" is
forgettable. A guarantee that names the prospect's actual fears and assigns a
specific consequence to the seller is memorable, differentiating, and
conversion-generating.

**3-tier pricing is the default.** A single price leaves money on the table and
removes choice architecture. Always offer Basic, Standard, and Premium. Always
include an ultra-high-ticket option — someone in every market will pay 10x or
100x more.

**Ultra-high-ticket anchors make Standard look reasonable.** This is not a
gimmick; it is how choice architecture reduces price resistance across the
entire product line.

**Inertia favours the seller in try-before-you-buy.** The mechanic is not
charity — it is psychology. Getting the product into the prospect's hands shifts
the burden of reversing the sale from you to them.

**Friction kills conversions.** At the moment of purchase, the prospect has
already decided. Every unnecessary step is a chance for second thoughts. Make
yes the path of least resistance.

**If you have to hard-sell, the nurture failed.** Conversion is the downstream
outcome of a working nurture process. Conversion tactics cannot compensate for
cold, unprepared leads.

## Examples

### Example 1 — IT Support Company (3-Tier Pricing)

| Tier | Price | Includes |
|------|-------|----------|
| Basic | $499/mo | Remote support, 8-hour response SLA |
| Standard | $899/mo | Remote + on-site, 2-hour response, quarterly review |
| Premium | $1,399/mo | Standard + dedicated consultant, 15-min response, monthly strategy session |
| Concierge (anchor) | $8,500/mo | Premium + CTO-on-call, custom security audit, unlimited on-site |

Guarantee (verbatim): "We guarantee that our certified and experienced IT
consultants will fix your IT problems so they don't recur. They'll also return
your calls within fifteen minutes and will always speak to you in plain English.
If we don't live up to any of these promises, we insist that you tell us and
we'll credit back to your account double the billable amount of the consultation."

Try-before-you-buy: First 30-day trial at Standard rate, full cancellation rights.

### Example 2 — Pest Control Company

Customer fears identified:
1. Pests return after treatment
2. Toxic chemicals harm family or pets
3. Technician leaves the house dirty

Guarantee (verbatim): "We guarantee to rid your home of ants forever, without
the use of toxic chemicals, while leaving your home in the same clean and tidy
condition we found it. If you aren't absolutely delighted with the service
provided, we insist that you tell us and we'll refund double your money back."

Note: each named fear maps directly to a named promise. This is the formula.

### Example 3 — Business Coaching (Applying All 5 Elements)

**Trust checklist gaps identified:** No physical address, no team photos, no
testimonials with named clients. Priority: add testimonials first (highest trust
impact, lowest cost).

**Guarantee:** "We guarantee you will have a complete 90-day action plan with
measurable targets within 3 sessions. If you don't, we'll coach you at no charge
until you do — or refund your investment in full."

**3-tier pricing:**
- Starter ($497): 3 sessions + email support
- Growth ($997): 6 sessions + weekly accountability check-in + template library
- Accelerator ($1,997): Growth + priority access + done-with-you plan build
- VIP Intensive (anchor, $12,000): Full-day immersive + 12 months on-call access

**Try-before-you-buy:** Free 45-minute strategy session (first value delivery,
creates inertia toward the paid programme).

**Friction removed:** Stripe payment plan (3 monthly instalments) available on
all tiers. Single-page booking form. Calendar booking self-serve.

## References

- Source: Allan Dib, *The 1-Page Marketing Plan*, Chapter 6 "Sales Conversion",
  pp 163–188
- Depends on: `lead-nurture-sequence-design` (warm leads pre-condition),
  `irresistible-offer-builder` (defined offer pre-condition)
- Related: `customer-lifetime-value-growth` (post-conversion value extraction),
  `customer-experience-systems-design` (delivery quality that backs the guarantee)
- Columbia jam study referenced in pricing section: Iyengar & Lepper (2000),
  "When Choice is Demotivating", *Journal of Personality and Social Psychology*

## License

This skill is licensed under [CC-BY-SA-4.0](https://creativecommons.org/licenses/by-sa/4.0/).
Source: [BookForge](https://github.com/bookforge-ai/bookforge-skills) — The 1-Page Marketing Plan by Allan Dib.

## Related BookForge Skills

Install related skills from ClawhHub:
- `clawhub install bookforge-lead-nurture-sequence-design`
- `clawhub install bookforge-irresistible-offer-builder`

Or install the full book set from GitHub: [bookforge-skills](https://github.com/bookforge-ai/bookforge-skills)

---
name: referral-system-design
description: Design an active referral system for a small business. Use this skill when you want to get more referrals, build a referral system, stop relying on word of mouth, ask for referrals without seeming desperate, write a referral script, create a gift card referral mechanism, apply the bystander effect override to referral asks, set upfront referral expectations with new customers, find joint venture referral partners, write a JV referral outreach email, profile complementary business partners for lead exchange, fill square 9 of the 1-Page Marketing Plan, apply the law of 250, design an orchestrated referral ask cadence, or turn customer satisfaction into a systematic lead source.
version: 1.0.0
homepage: https://github.com/bookforge-ai/bookforge-skills/tree/main/books/the-1-page-marketing-plan/skills/referral-system-design
metadata: {"openclaw":{"emoji":"📚","homepage":"https://github.com/bookforge-ai/bookforge-skills"}}
status: published
source-books:
  - id: the-1-page-marketing-plan
    title: "The 1-Page Marketing Plan: Get New Customers, Make More Money, and Stand Out from the Crowd"
    authors: ["Allan Dib"]
    chapters: [9]
tags: [marketing, referrals, joint-ventures, word-of-mouth, small-business]
depends-on: ["customer-experience-systems-design"]
execution:
  tier: 1
  mode: full
  inputs:
    - type: document
      description: "Business description — what the business sells, who the ideal customer is, and a rough sense of customer satisfaction level"
  tools-required: [Read, Write]
  tools-optional: []
  mcps-required: []
  environment: "Document set. Typical files: business-description.md, customer-list.md or CRM export (optional), referral-system.md (to be created)."
---

# Referral System Design

## When to Use

You are a small business owner who wants referrals to be a deliberate, reliable lead source — not an occasional windfall you hope for. Typical triggers:

- You get some referrals but they happen randomly; you have no system driving them
- You feel awkward asking for referrals, as if it makes you seem needy or desperate
- You have satisfied customers who never refer — because no one asked them to
- You attend networking events and ask for "anyone who needs X" without getting results
- You want to set up a joint venture with a complementary business to generate warm leads
- You want to fill square 9 of your 1-Page Marketing Plan canvas

**Dependency check:** This skill assumes customers are genuinely satisfied. If customer experience baseline is unknown, invoke `customer-experience-systems-design` first, or verify that delivered results are strong before running referral scripts. A referral system built on a poor experience accelerates negative word of mouth, not positive.

Before starting, clarify:
- **What is the business?** (product/service, industry, approximate customer base size)
- **What is the typical lifetime value of a customer?** (needed to size gift card and JV incentives)
- **Are there complementary businesses the owner already has relationships with?** (head start for JV partner identification)
- **Is there a current referral practice?** (Audit starting point)

---

## Context & Input Gathering

Ask the user for the following if not already provided:

1. **Business description** — What does the business sell, and to whom?
2. **Satisfaction signal** — Are customers currently happy? Any testimonials, repeat business, or positive feedback?
3. **Customer lifetime value** — Rough estimate. This determines how much to invest in a gift card or JV voucher.
4. **Existing referral practice** — Do you ask for referrals now? How? What happens?
5. **Complementary businesses** — Name any businesses that serve your ideal customer before or after you do.

---

## Process

### Phase 1: Referral Practice Audit

**Step 1 — Classify current referral behavior.**

Determine which pattern describes the current state:

| Pattern | Description | Risk Level |
|---------|-------------|-----------|
| **Passive hope** | Do great work, hope referrals come. No ask, no system. | High — single uncontrollable source |
| **Vague ask** | Sometimes mention "if you know anyone..." to happy customers. No gift, no specifics. | Medium — triggers bystander effect |
| **Active ask, no system** | Ask directly but inconsistently, no cadence, no gift card, no upfront expectation. | Medium — sporadic results |
| **Active system** | Upfront expectation-setting + post-delivery ask + gift card + JV partners + ask cadence. | Low — reliable lead flow |

WHY: Most small businesses are in "passive hope." The goal of this phase is to name the gap honestly before writing scripts. A business that never asks cannot improve its referral rate by changing its ask wording — it must first commit to asking.

---

### Phase 2: Write the Upfront Expectation-Setting Script

**Step 2 — Craft the expectation-setting script for the start of each new customer engagement.**

Use this template, adapted to the business:

> *"[Customer name], I'm going to do an awesome job for you, but I need your help also. Most of our new business comes through referrals. This means that rather than paying for advertising to get new clients, we pass the cost savings directly to you. We typically get about three referrals from each new customer. When we're finished working together and you're 100% satisfied with the work we've done, I'd really appreciate it if you could keep in mind three or more other people who we could also help."*

Four elements this script delivers:
1. **Promise of result** — "I'm going to do an awesome job for you" — power stays with the customer; the referral is conditional on your performance
2. **Their benefit** — referral model = cost savings passed directly to them
3. **Specific number** — "three or more" plants a number without pressure; the customer begins thinking ahead
4. **Future-oriented** — delivered at the START, when the relationship is fresh and expectations are high

WHY: Setting the referral expectation upfront is far more effective than a surprise ask at the end. The customer has the entire engagement to mentally prepare a short list. It also reframes referrals as the normal operating model of the business — not as begging.

---

### Phase 3: Write the Post-Delivery Direct Ask Script with Gift Card

**Step 3 — Craft the post-delivery direct ask.**

Use this template:

> *"[Customer name], it's been such a pleasure working with you. If you know anyone who's in a similar situation to yourself, we'd love you to give them one of these gift cards which entitles them to $[amount] off their first [consultation/purchase/session] with us. One of the reasons we're able to keep the cost of our service down is because we get a lot of our business through referrals from people like you."*

Three elements this script delivers:
1. **Acknowledgment** — opens by recognizing the customer; people love being acknowledged
2. **Gift, not favor** — you are not asking the customer for something; you are giving them something valuable (a gift card) to pass on to a friend
3. **Reason why** — explaining the referral-based business model makes the ask feel principled, not transactional

**Gift card sizing:** The face value should feel meaningful to the recipient (your prospect) while being affordable relative to your customer lifetime value. If a new customer is worth $5,000 in lifetime revenue, a $100 gift card is a 2% acquisition cost — far cheaper than advertising.

WHY: Framing the ask as offering a gift to the referrer's friend removes the psychological barrier of "begging for a favor." The referrer becomes a value-provider to their network, not a sales agent for your business. This taps the real psychology of referrals: people share because it makes them look and feel good — not as a favor to you.

---

### Phase 4: Apply the Bystander-Effect Override Formula

**Step 4 — Make asks specific, not general.**

Never use: *"If you know anyone who needs [service], please send them my way."*

"Anyone" is "somebody else." Everyone assumes someone else will refer. This is the bystander effect: in a crowd emergency, when everyone assumes someone else will act, no one acts. The same diffusion of responsibility kills generic referral asks.

The fix: be specific. For a referral to occur in a conversation between two people, three things must happen:
1. The referrer notices the conversation is relevant to what you do
2. They think of YOU specifically
3. They introduce you into the conversation

**Formula for a specific ask:**

Instead of "anyone who needs X" → name a specific type of person AND a specific trigger situation.

| Generic (bystander-effect) | Specific (bystander-effect override) |
|---------------------------|-------------------------------------|
| "Anyone who needs a financial planner" | "Someone nearing retirement age who's recently been made redundant" |
| "Anyone who needs a pet food recommendation" | "A new dog owner your vet just prescribed XYZ food for" |
| "Anyone who needs a lawyer" | "A friend who's just started a business and hasn't set up their contracts yet" |

WHY: A specific person + specific trigger creates personal responsibility in the referrer. They think: "Actually, my colleague John just got made redundant last month — this is exactly for him." The abstract "anyone" never triggers that mental shortcut.

---

### Phase 5: Build the JV Partner Profile

**Step 5 — Identify joint venture referral partner types.**

Ask: who serves your ideal customer BEFORE or AFTER they come to you? Those are your JV candidates.

Examples:
- Lawyer → Accountant (same client, different need, non-competing)
- Car detailer → Mechanic (sequential relationship with the same car owner)
- Pet food retailer → Veterinarian (the vet prescribes what the retailer sells)
- Financial planner → Real estate agent (both serve clients with major financial transitions)

**Step 6 — Build a referral profile for each JV partner type.**

For each partner type, answer the five profiling questions:

| Question | What to Determine |
|----------|------------------|
| **Who do they know?** | What types of prospects are in their client or contact base? |
| **What trigger event makes referral timely?** | What happens in the prospect's life that makes your service relevant RIGHT NOW? |
| **What specific problem does the prospect have?** | Not "needs a [professional]" — name the exact problem (redundancy, new puppy, just signed a lease) |
| **How does the referrer look good by referring?** | What value does the referrer provide their client by passing your asset on? |
| **What asset can you give the referrer to pass on?** | A gift card, a lead magnet, a special report, a voucher — something with tangible value |

WHY: A cold request to a JV partner ("send me your clients") fails for the same reason a generic referral ask fails. The partner has no personal responsibility and no tool. A profiled approach gives the partner a specific trigger to watch for and a specific asset to offer — making them a value-provider to their clients, not a salesperson for you.

---

### Phase 6: Write the JV Outreach Email

**Step 7 — Draft the JV outreach email for the primary JV partner type.**

Use this structure (adapted from the Financial Planner → Real Estate Agent example):

> *"Hey [Partner name],*
>
> *If you have anyone who [specific trigger situation — e.g., wants to buy or sell a property, or who's about retirement age and has recently been made redundant], I've got something that I think would really help them out. I've put together a [asset — e.g., special report, gift card, voucher] titled "[Asset name]." If you have anyone who would benefit from this, please call or text me, and I'll [send you a copy to pass on / give you gift cards to hand out]."*

Key design principles:
- **Trigger-specific** — name the exact situation that makes your service relevant now
- **Warm, not cold** — the referrer passes on an asset; they do not give out the prospect's contact details
- **Referrer looks good** — they provide genuine value to their client, not a sales pitch
- **Low friction** — the partner just has to forward something; no explanation, no selling

WHY: Not asking for a cold referral is critical. A cold referral requires the referrer to convince their client to call you — a significant social cost. A warm pass-through (share this report/voucher) requires almost no effort and creates goodwill between the referrer and their client.

---

### Phase 7: Design the Orchestrated Ask Cadence

**Step 8 — Schedule the referral ask touchpoints.**

Document the full referral ask cadence as a repeatable process:

```
REFERRAL ASK CADENCE

Touchpoint 1 — Onboarding (Day 1 of engagement):
  Deliver: upfront expectation-setting script
  Goal: plant the referral expectation early; let the customer begin thinking

Touchpoint 2 — Mid-engagement check-in (if engagement > 2 weeks):
  Deliver: satisfaction check ("How are things going so far?")
  Goal: confirm experience is positive before post-delivery ask

Touchpoint 3 — Completion / Delivery (Day of handoff):
  Deliver: post-delivery direct ask script + gift cards
  Goal: capture referral commitment when satisfaction is highest

Touchpoint 4 — Follow-up (14–21 days post-delivery):
  Deliver: brief follow-up call/email checking in on results + gentle ask
  ("We'd love to help someone else in your network — do you still have
  that gift card? Happy to send more if you'd like.")
  Goal: capture referrals from customers who forgot or were busy at delivery

JV Outreach Cadence (independent of individual customer cadence):
  Quarterly: review JV partner list; identify new trigger events (legislation
  changes, local business news, seasonal patterns); send fresh JV outreach email
  to each partner type with an updated asset or gift card batch
```

WHY: Most referral asks happen once, at delivery, if at all. A cadence spreads the ask across four touchpoints and catches customers at different stages of post-purchase satisfaction. The JV cadence treats partner relationships as a recurring asset to be actively maintained — not a one-time deal struck and forgotten.

---

## Inputs

| Input | Required | Notes |
|-------|----------|-------|
| Business description (industry, product/service) | Yes | Needed to write specific scripts |
| Approximate customer lifetime value | Yes | Needed to size gift card amount |
| Current referral practice | Yes | Needed for audit in Phase 1 |
| Complementary businesses (JV candidates) | Recommended | Can be identified from scratch if absent |
| Customer list or CRM export | Optional | Helps segment referrer profiles |

---

## Outputs

This skill produces a single document: **referral-system.md**

Contents:

1. **Referral practice audit** — current state classification (passive hope / vague ask / active ask / active system)
2. **Upfront expectation-setting script** — adapted to the business, ready to use at onboarding
3. **Post-delivery direct ask script** — adapted to the business, with gift card amount determined
4. **Gift card mechanism** — face value, redemption condition, delivery method (physical card, digital code, voucher)
5. **Bystander-effect override** — specific person type + trigger situation table for use in direct and JV asks
6. **JV partner list** — two to four partner types profiled using the five-question framework
7. **JV outreach email template** — one per primary partner type, ready to send
8. **Orchestrated ask cadence** — four-touchpoint customer cadence + quarterly JV cadence, formatted as a checklist

---

## Key Principles

**Active over passive.** Passive word-of-mouth is a free lunch — nice, but not a strategy. A referral system requires deliberate orchestration: scripts, gifts, timelines, and JV partnerships. Hope is not a plan.

**They refer because it benefits them, not you.** People share recommendations to look good and to help their friends. Every referral script and JV mechanism must be designed around what's in it for the referrer and their contact — not around your need for new business.

**Specific over general.** "Anyone who needs X" is the perfect ingredient for the bystander effect. Name a specific person type and a specific trigger event. Personal responsibility and mental specificity collapse the gap between "I should refer someone" and actually making the introduction.

**Reward referrers — give them something to offer.** A gift card or voucher transforms the referrer from a salesperson into a value-provider. They are not asking their friend to call you; they are handing their friend $50.

**Set the expectation upfront.** Asking at delivery lands as a surprise. Setting the expectation at onboarding gives the customer the entire engagement to identify three names. It also frames referrals as the normal operating model of the business.

**The tribe is a prerequisite.** Referral systems accelerate what is already happening — a positive customer experience. Building a referral system on top of a mediocre experience produces negative word of mouth at scale. Verify satisfaction before running scripts.

**Law of 250.** Every customer has about 250 people in their life who are important enough to invite to a wedding or funeral. Treat every customer relationship as a node in a network of 250 potential referrals. One bad experience loses 250 prospects.

---

## Examples

### Example 1: Mike's Pet World — JV Gift Card Mechanism (verbatim case)

Mike's Pet World is a pet food retailer. Their JV partner: a local veterinarian who recommends XYZ dog food to clients.

The arrangement: the vet receives a supply of $50 gift cards redeemable at Mike's Pet World. After a consultation where the vet recommends XYZ food, the vet hands the client a card:

*"I recommend XYZ dog food. You can buy it at most pet food retailers but you're a good customer so here's a $50 voucher that you can redeem at Mike's Pet World, which is just down the road. They always carry plenty of stock of XYZ dog food."*

The outcome:
- Vet creates massive goodwill — handing a client $50 for free
- Client receives an unexpected discount; no sales pressure
- Mike's Pet World acquires a new customer whose lifetime value is approximately $5,000
- Mike's cost: a $50 voucher (wholesale cost even less)
- Mike's also acquires the trust the client has with their vet — the referral carries that goodwill

Most customers redeem the card because throwing out a voucher with a monetary value attached feels wasteful. The conversion rate on gift card referrals far exceeds cold advertising.

### Example 2: Financial Planner → Real Estate Agent — JV Outreach Email (verbatim case)

A financial planner wants referrals from real estate agent clients. JV profiling:
- Trigger event: clients of the real estate agent who are nearing retirement age and have recently been made redundant
- Problem: how to leverage a redundancy package to fund a comfortable retirement
- Referrer looks good by: giving their client a free special report with practical guidance
- Asset to pass on: a free report titled "The 7 Keys to Leveraging Your Redundancy Package and Ensuring a Fully Funded Retirement"

The JV outreach email sent to six real estate agent clients:

> *"Hey Bob,*
>
> *If you have anyone who wants to buy or sell a property or who's about retirement age and has recently been made redundant, I've got something that I think would really help them out. I've put together a special report titled, "The 7 Keys to Leveraging Your Redundancy Package and Ensuring a Fully Funded Retirement." If you have anyone who would benefit from this, please call or text me, and I'll send you a copy of the report to pass on."*

This email is:
- Specific about the trigger (redundancy + retirement age)
- Not asking for cold contact details — the agent just forwards the report
- Designed so the agent looks good to their client (providing genuine value)
- Low friction — the call to action is "call or text me"

### Example 3: Home Renovation Contractor — Full System Applied

A home renovation contractor with 40 completed projects and good reviews but zero referral system.

**Upfront expectation-setting (delivered at contract signing):**
> "Mrs. Johnson, I'm going to do a great job on your kitchen. Most of our new projects come through referrals from happy clients — it keeps our overhead low and we pass those savings on. When we're done and you love the result, I'd really appreciate it if you could think of three neighbors or friends who might be planning something similar."

**Post-delivery ask (delivered on final walkthrough):**
> "Mrs. Johnson, it's been a real pleasure working with you. Here are three gift cards — each one gives a friend $500 off a kitchen or bathroom project with us. If you know anyone planning a renovation, we'd love you to pass these on."

**Gift card sizing:** Customer lifetime value = $20,000 (repeat renovations + referrals). Gift card face value = $500. Acquisition cost = 2.5%.

**JV partner profile:** Interior designers (serve homeowners before renovation begins). Trigger event: client has just signed a lease or purchased a new home. Asset: $500 gift card toward renovation. The designer hands the client the card as a gesture of added value.

**Bystander-effect override in the post-delivery ask:** "Do you have a neighbor who mentioned they've been thinking about redoing their kitchen? Anyone who's just bought a new place nearby?"

---

## References

- Dib, A. (2016). *The 1-Page Marketing Plan*, Chapter 9: Orchestrating and Stimulating Referrals (pp. 250–266).
- Law of 250: Joe Girard — developed while observing attendance at Catholic funerals and wedding guest counts. Average person has ~250 people important enough to invite to either. Each customer represents 250 potential referrals (or 250 potential enemies).
- Bystander effect: first documented by Darley and Latané (1968) following the Kitty Genovese case. Diffusion of responsibility in crowds — the same mechanism that kills generic referral asks at networking events.
- `customer-experience-systems-design` — prerequisite skill. Referral systems accelerate the existing customer experience trajectory. Run this skill first if customer satisfaction is unverified.

## License

This skill is licensed under [CC-BY-SA-4.0](https://creativecommons.org/licenses/by-sa/4.0/).
Source: [BookForge](https://github.com/bookforge-ai/bookforge-skills) — The 1-Page Marketing Plan by Allan Dib.

## Related BookForge Skills

Install related skills from ClawhHub:
- `clawhub install bookforge-customer-experience-systems-design`

Or install the full book set from GitHub: [bookforge-skills](https://github.com/bookforge-ai/bookforge-skills)

---
name: marketing-plan-canvas
description: "Use this skill to build, assemble, or audit a complete 9-square
  1-Page Marketing Plan canvas for any small or medium business. Triggers when a
  user asks to create a marketing plan, build a 1-page marketing plan, use the
  1PMP framework, fill in the 9-square canvas, create a marketing strategy, get
  a complete marketing plan, start marketing a business, fix broken marketing,
  audit their marketing, or doesn't know where to start with marketing. Also
  triggers for: 'marketing plan', '1-page marketing plan', 'create marketing
  plan', 'marketing strategy', 'complete marketing plan', '1PMP canvas', '9
  square canvas', 'marketing plan template', 'small business marketing plan',
  'need a marketing plan', 'don't know where to start with marketing', 'my
  marketing is a mess', 'random acts of marketing', 'Allan Dib framework',
  'before during after marketing', 'prospect lead customer framework', 'I have
  no marketing system', 'pull my marketing together', 'marketing audit'. This
  is the hub skill — invoke it whenever a user wants end-to-end marketing
  clarity, not just one tactical element."
version: 1.0.0
homepage: https://github.com/bookforge-ai/bookforge-skills/tree/main/books/the-1-page-marketing-plan/skills/marketing-plan-canvas
metadata: {"openclaw":{"emoji":"📚","homepage":"https://github.com/bookforge-ai/bookforge-skills"}}
status: published
source-books:
  - id: the-1-page-marketing-plan
    title: "The 1-Page Marketing Plan"
    authors: ["Allan Dib"]
    chapters: [0, 1, 2, 3, 4, 5, 6, 7, 8, 9]
tags:
  - marketing
  - marketing-plan
  - strategy
  - small-business
  - meta-framework
depends-on:
  - target-market-selection-pvp-index
  - marketing-message-and-usp-crafting
  - advertising-media-roi-framework
  - lead-capture-ethical-bribe-design
  - lead-nurture-sequence-design
  - sales-conversion-trust-system
  - customer-experience-systems-design
  - customer-lifetime-value-growth
  - referral-system-design
  - irresistible-offer-builder
execution:
  tier: 1
  mode: full
  inputs:
    - type: document
      description: >
        Business description at minimum (one paragraph: what the business does,
        who it currently serves, what it sells). Richer inputs are any existing
        outputs from the 10 dependent skills — the skill will integrate whatever
        exists and ask for minimums on whatever is missing.
  tools-required: [Read, Write]
  tools-optional: [Grep]
  mcps-required: []
  environment: >
    Document set — business description and any existing skill outputs in
    markdown. No code execution required. Can produce a minimum-viable canvas
    in a single conversation from nothing but a business description.
discovery:
  goal: >
    Guide the user through all 9 squares of the 1-Page Marketing Plan, produce
    a coherent single-document canvas (marketing-plan-canvas.md), and bias
    toward shipping an 80% plan over endlessly refining a perfect one.
  tasks:
    - "Explain the 9-square framework and 3-act structure to the user"
    - "Assess current state per square: have output / partial / missing"
    - "For each square, decide: reference existing skill output, invoke
       dedicated skill, or gather minimum viable input directly from user"
    - "Fill all 9 squares in Act order: Before (1-2-3), During (4-5-6),
       After (7-8-9)"
    - "Perform coherence review across all 9 squares"
    - "Write marketing-plan-canvas.md — the single output document"
    - "Apply 80-percent-ships principle — ship the plan, iterate later"
  audience:
    roles:
      - small-business-owner
      - solopreneur
      - entrepreneur
      - freelancer
      - startup-founder
    experience: beginner-to-intermediate
  when_to_use:
    triggers:
      - "User wants a complete marketing plan from scratch"
      - "User has partial marketing work and wants to integrate it"
      - "User's marketing is broken or random and they want a system"
      - "User is auditing their marketing against a proven framework"
      - "User mentions the 1-Page Marketing Plan or Allan Dib"
    prerequisites: []
    not_for:
      - "Enterprise/large company brand marketing with multi-million-dollar
         budgets (this framework is explicitly designed for small to medium
         businesses using direct response marketing)"
      - "Detailed execution of a single square — invoke the dedicated
         component skill directly instead"
  environment:
    codebase_required: false
    codebase_helpful: false
    works_offline: true
  quality:
    scores:
      with_skill: 100
      baseline: 21.4
      delta: 78.6
    tested_at: "2026-04-09"
    eval_count: 1
    assertion_count: 14
    iterations_needed: 1
---

# The 1-Page Marketing Plan Canvas

A meta-orchestrator that assembles all 9 squares of the direct response
marketing lifecycle into a single canvas document. Covers the full prospect
→ lead → customer journey across three acts: Before (get them to KNOW you),
During (get them to LIKE you and buy), After (get them to TRUST you, buy
again, and refer others).

This skill works in three starting states — from nothing, from partial work,
or as an audit of an existing plan — and always produces a single output
document: `marketing-plan-canvas.md`.

---

## When to Use

Use this skill when the goal is the **complete marketing system**, not one
piece. It is the entry point for any of these situations:

- **Starting from scratch:** no marketing plan, no clarity, just a business
  description
- **Integrating partial work:** some squares filled via other skills, but no
  unified view
- **Auditing a broken plan:** marketing exists but is random, fragmented, or
  not converting

Do NOT use this skill when the user wants deep work on a single square.
Invoke the dedicated component skill directly for that (e.g.,
`target-market-selection-pvp-index` for square #1 alone).

---

## Context and Input Gathering

Before starting, determine the user's starting state:

**Minimum viable input (always ask if absent):**
- What does your business do, in one sentence?
- Who do you currently serve (rough description)?
- What do you sell, and at roughly what price?

**Check for existing outputs from component skills.** If the user has worked
through any of the 9 dependent skills, those outputs become the content for
the corresponding square. Ask:

> "Have you worked through any of the individual marketing squares before?
> If so, share those documents and I'll integrate them. Otherwise I'll ask
> you the minimum question for each square."

**For each square, the decision rule is:**
1. IF a dedicated skill output exists → reference it, extract a 3-5 bullet
   summary for the canvas
2. ELSE IF the user wants depth on this square → invoke the dedicated skill
   before proceeding
3. ELSE → ask the minimum viable question and record the answer directly

---

## Process

### Step 1 — Explain the Framework (WHY: shared vocabulary prevents wasted effort)

Orient the user before gathering any input. Explain:

The 1-Page Marketing Plan organizes all marketing into 9 squares across
3 acts, reflecting the 3 stages every customer passes through:

```
┌──────────────────────────────────────────────────────────────┐
│                  MY 1-PAGE MARKETING PLAN                     │
├─────────────────┬──────────────────┬────────────────────────┤
│  ACT I: BEFORE  │  (Prospect)      │  Goal: they KNOW you   │
├─────────────────┼──────────────────┼────────────────────────┤
│ #1 My Target    │ #2 My Message    │ #3 Media to            │
│    Market       │    to Them       │    Reach Them          │
├─────────────────┴──────────────────┴────────────────────────┤
│  ACT II: DURING │  (Lead)          │  Goal: they LIKE you   │
├─────────────────┬──────────────────┬────────────────────────┤
│ #4 Lead         │ #5 Lead          │ #6 Sales               │
│    Capture      │    Nurture       │    Conversion          │
├─────────────────┴──────────────────┴────────────────────────┤
│  ACT III: AFTER │  (Customer)      │  Goal: they TRUST you  │
├─────────────────┬──────────────────┬────────────────────────┤
│ #7 World-Class  │ #8 Increase      │ #9 Orchestrate         │
│    Experience   │    Customer LTV  │    Referrals           │
└─────────────────┴──────────────────┴────────────────────────┘
```

Key insight: this is a direct response framework for small and medium
businesses. It is NOT mass marketing or branding. Every square is designed
to produce a measurable, trackable result.

### Step 2 — Assess Current State (WHY: avoid redoing completed work)

For each of the 9 squares, determine the status:
- **Have:** a skill output or documented answer exists
- **Partial:** some thinking done but not fully worked through
- **Missing:** no prior work

Present a quick status table to the user and confirm before proceeding.

### Step 3 — Act I: Before (squares 1, 2, 3 in order)

**WHY for the order:** target market must be decided before writing a message,
and media channel only makes sense once you know who you're reaching and
what you're saying. Reversing this order is the definition of "random acts
of marketing."

**Square #1 — Target Market**
- Skill: `target-market-selection-pvp-index`
- IF output exists → extract: primary target segment + customer avatar
  (3-5 bullets)
- ELSE IF user wants depth → invoke `target-market-selection-pvp-index`
- ELSE → ask: *"Who is your primary target customer? Describe them in one
  sentence — their role, their problem, their situation."*
- Note: `irresistible-offer-builder` outputs feed into the offer framing
  for this segment (cross-reference).

**Square #2 — Message to Target Market**
- Skill: `marketing-message-and-usp-crafting`
- IF output exists → extract: USP statement + headline (3-5 bullets)
- ELSE IF user wants depth → invoke `marketing-message-and-usp-crafting`
- ELSE → ask: *"What is your unique selling proposition in 1-2 sentences?
  Why should your target customer choose you over every alternative,
  including doing nothing?"*
- Note: `irresistible-offer-builder` is the cross-cutting input here —
  a strong offer makes the message concrete.

**Square #3 — Advertising Media**
- Skill: `advertising-media-roi-framework`
- IF output exists → extract: primary channels + budget allocation approach
  (3-5 bullets)
- ELSE IF user wants depth → invoke `advertising-media-roi-framework`
- ELSE → ask: *"Where will you reach your target customers? Name 1-2
  specific channels (e.g., Facebook ads, LinkedIn, local newspaper,
  email list, Google search)."*

### Step 4 — Act II: During (squares 4, 5, 6 in order)

**WHY for the order:** leads must be captured before they can be nurtured,
and nurturing must happen before conversion. The sequence is the system.

**Square #4 — Lead Capture**
- Skill: `lead-capture-ethical-bribe-design`
- IF output exists → extract: lead magnet name + capture mechanism (3-5
  bullets)
- ELSE IF user wants depth → invoke `lead-capture-ethical-bribe-design`
- ELSE → ask: *"How do you capture interested prospects' contact details?
  What do you offer them in exchange (e.g., free report, checklist,
  webinar, free consultation)?"*

**Square #5 — Lead Nurture**
- Skill: `lead-nurture-sequence-design`
- IF output exists → extract: nurture medium + sequence length + cadence
  (3-5 bullets)
- ELSE IF user wants depth → invoke `lead-nurture-sequence-design`
- ELSE → ask: *"How do you follow up with leads who haven't bought yet?
  What's your current sequence (email, phone, direct mail) and how long
  does it run?"*

**Square #6 — Sales Conversion**
- Skill: `sales-conversion-trust-system`
- IF output exists → extract: conversion mechanism + trust elements
  (3-5 bullets)
- ELSE IF user wants depth → invoke `sales-conversion-trust-system`
- ELSE → ask: *"How do you convert a nurtured lead into a paying customer?
  What's the final step that gets someone to buy (sales call, proposal,
  checkout page, in-person meeting)?"*
- Note: `irresistible-offer-builder` is a direct feed into square #6 —
  the offer structure determines conversion rate.

### Step 5 — Act III: After (squares 7, 8, 9 in order)

**WHY for the order:** experience must exist before you can measure
lifetime value, and referrals only happen when the experience is strong
enough to earn recommendation. Act III is where most businesses leave
the most money on the table.

**Square #7 — World-Class Experience**
- Skill: `customer-experience-systems-design`
- IF output exists → extract: key experience systems + wow moments (3-5
  bullets)
- ELSE IF user wants depth → invoke `customer-experience-systems-design`
- ELSE → ask: *"What happens after someone becomes a customer? How do you
  make sure they feel looked after and impressed — not just served?"*

**Square #8 — Increase Customer Lifetime Value**
- Skill: `customer-lifetime-value-growth`
- IF output exists → extract: upsell/cross-sell paths + LTV levers (3-5
  bullets)
- ELSE IF user wants depth → invoke `customer-lifetime-value-growth`
- ELSE → ask: *"What do customers buy after their first purchase? Do you
  have a higher-tier offer, a subscription, or a natural next product
  they should move to?"*

**Square #9 — Orchestrate and Stimulate Referrals**
- Skill: `referral-system-design`
- IF output exists → extract: referral mechanism + incentive structure
  (3-5 bullets)
- ELSE IF user wants depth → invoke `referral-system-design`
- ELSE → ask: *"How do you generate referrals? Is there a systematic
  process, or does it happen by accident when a happy customer mentions
  you?"*

### Step 6 — Coherence Review (WHY: incoherent plans waste money faster than no plan)

Before writing the canvas document, check:

1. **Market ↔ Message match:** Does the USP in square #2 speak directly
   to the pain of the target in square #1? If they are misaligned, the
   advertising in square #3 will reach the right people but say the wrong
   thing.

2. **Message ↔ Media match:** Does the channel in square #3 actually reach
   the target from square #1? (Example: LinkedIn reaches B2B professionals;
   Facebook reaches consumers and local businesses. Wrong channel =
   broadcasting to the wrong audience.)

3. **Lead magnet ↔ Target match:** Does the ethical bribe in square #4
   appeal to the specific target from square #1? A generic lead magnet
   attracts generic, low-quality leads.

4. **Offer ↔ Message consistency:** Does the irresistible offer (cross-
   cutting) back up the promise made in square #2? The message sets
   expectations; the offer must fulfil them.

5. **After phase completeness:** At least one element should exist in each
   of squares 7, 8, and 9. A business that excels at Before and During but
   ignores After is leaving its most profitable customers under-served.

Flag any misalignments to the user before writing the canvas. Suggest the
appropriate dedicated skill to resolve each gap, or ask the minimum
corrective question.

### Step 7 — Write `marketing-plan-canvas.md` (WHY: a plan that exists only in conversation cannot be executed)

Produce the single output document. Format:

```
# My 1-Page Marketing Plan
Business: [name]
Date: [date]
Status: [Draft / In Progress / Live]

---

## ACT I: BEFORE — Getting Prospects to KNOW You

| Square | Content |
|--------|---------|
| #1 Target Market | [3-5 bullets from skill output or user input] |
| #2 Message | [3-5 bullets: USP, headline, core promise] |
| #3 Media | [Top 1-2 channels, allocation rationale] |

---

## ACT II: DURING — Getting Leads to LIKE You and Buy

| Square | Content |
|--------|---------|
| #4 Lead Capture | [Lead magnet name, capture mechanism] |
| #5 Lead Nurture | [Sequence summary: medium, cadence, length] |
| #6 Sales Conversion | [Conversion mechanism, trust elements, offer summary] |

---

## ACT III: AFTER — Getting Customers to TRUST You, Buy Again, Refer

| Square | Content |
|--------|---------|
| #7 World-Class Experience | [Key experience systems, wow moments] |
| #8 Increase Customer LTV | [Upsell path, retention mechanism, subscription offer] |
| #9 Orchestrate Referrals | [Referral mechanism, incentive, timing] |

---

## Cross-Cutting: Irresistible Offer
[Summary of offer structure — feeds squares #2 and #6]
Source: irresistible-offer-builder skill output (if available)

---

## Coherence Notes
[Any misalignments flagged in Step 6, with recommended next actions]

---

## Next Actions
[Prioritized list: which squares need deeper work via dedicated skills]
```

For each square, if a dedicated skill output exists, add a reference link:
`→ Full detail: skills/[skill-name]/[output-file].md`

### Step 8 — Apply the 80% Principle (WHY: a shipped plan beats a perfect draft)

After writing the canvas, make this explicit to the user:

> "This plan is ready to execute. It is 80% complete — and 80% out the
> door beats 100% in the drawer every time. Marketing is not an event; it
> is a process. Start executing Act I immediately. Refine as you get
> real-world feedback. The squares that are thin will become clearer once
> you are in market."

Identify the single highest-leverage next action (usually: get the target
market clear first if missing, or start running Act I media if it exists).

---

## Inputs

**Minimum required:**
- Business description (one paragraph): what the business does, who it
  serves, what it sells

**Higher-value inputs (optional, but each one reduces the questions asked):**
- Outputs from any of the 10 dependent skills (in full or as summaries)
- Current marketing assets: ads, lead magnets, email sequences, sales
  scripts, onboarding processes

**Not required:**
- Prior marketing experience
- Large budget
- A completed business plan

---

## Outputs

**Primary output:**
- `marketing-plan-canvas.md` — single document with all 9 squares filled
  at minimum viable detail, plus coherence notes and next actions

**The output has two levels of completeness:**
- **Minimum viable canvas:** all 9 squares answered with 3-5 bullets each,
  gathered directly from user in this skill. Fast to produce; good enough
  to start executing.
- **Integrated canvas:** 9 squares populated from dedicated skill outputs,
  with references to full detail documents. Deeper; better for businesses
  that have worked through the component skills.

Both are valid outputs of this skill. The goal is a canvas that can be
stuck on a wall and used to guide daily marketing decisions.

---

## Key Principles

**Systematic, not random**
Random acts of marketing — trying Facebook ads this week, a podcast next
week, a direct mail campaign the month after — with no connecting strategy
is the single biggest reason small business marketing fails. The 9-square
canvas is the strategy that connects all tactics.

**Direct response, not mass marketing**
This framework is designed for small and medium businesses. It is NOT
brand marketing. Every element must be trackable and measurable. If you
cannot measure whether a square is working, it is not a direct response
element — revisit it.

**80% ships, 100% stays in the drawer**
Paralysis by analysis kills more marketing plans than bad strategy. A
good-enough plan executed today beats a perfect plan executed never.
Set a deadline: this canvas must be complete within one session. Gaps
become the next iteration, not a reason to delay.

**Iterate, don't perfect**
The first version of the canvas is a hypothesis. Real customers will
tell you what to change. Treat the canvas as a living document — review
it quarterly, update squares that are underperforming, and add depth via
dedicated skills when a specific square becomes a bottleneck.

**Marketing is a process, not an event**
High-growth businesses make marketing a daily routine. Failed businesses
treat marketing as something they do when revenue drops. The canvas gives
you the system; you provide the consistent execution.

---

## Examples

### Example A — Starting from Scratch

**Situation:** User runs a bookkeeping business for small trades businesses
(plumbers, electricians). No marketing plan. Gets clients by word of mouth.

**How this skill runs:**
1. Explain the 3-act framework.
2. Assess: all 9 squares missing.
3. Ask the minimum question for each square — takes ~15 minutes.
4. Coherence check: confirm the message in #2 speaks to tradesperson pain
   (cash flow visibility, ATO compliance stress) not generic bookkeeping.
5. Write canvas: minimum viable, 9 squares filled with user's answers.
6. Flag: square #5 (nurture) is thin — recommend invoking
   `lead-nurture-sequence-design` as next step.
7. Ship the plan. Identify Act I (LinkedIn for local trades) as first
   execution priority.

### Example B — Integrating Existing Work

**Situation:** User has outputs from `target-market-selection-pvp-index`,
`marketing-message-and-usp-crafting`, and `irresistible-offer-builder`.
Squares #1, #2, and the cross-cutting offer are done. Squares #3-9 need
work.

**How this skill runs:**
1. Read in the three existing skill outputs.
2. Assess: #1 and #2 have full outputs; cross-cutting offer done.
3. Summarize #1 and #2 into canvas format (3-5 bullets each).
4. Ask minimum questions for squares #3-9.
5. Coherence check: verify the offer (from `irresistible-offer-builder`)
   is consistent with the message in #2.
6. Write integrated canvas with references to the two full skill outputs.
7. Prioritize: squares #4 and #5 are thin — recommend
   `lead-capture-ethical-bribe-design` and `lead-nurture-sequence-design`.

### Example C — Auditing a Broken Plan

**Situation:** User has been running Facebook ads for 6 months with no
results. They have a landing page, some emails, and a sales process. Their
marketing "isn't working."

**How this skill runs:**
1. Map what they have to the 9 squares.
2. Assessment reveals: square #1 is vague ("small business owners"), #2
   has no USP (their ads look like mass marketing), #3 is Facebook only
   with no tracking.
3. Coherence failure: message (#2) is generic; media (#3) is running but
   without a targeted message, it attracts low-quality leads.
4. The root cause is that squares #1 and #2 are broken — not the Facebook
   channel.
5. Recommend: invoke `target-market-selection-pvp-index` to sharpen the
   niche, then `marketing-message-and-usp-crafting` to rebuild the message
   before spending another dollar on ads.
6. Write an audit canvas documenting current state and gap analysis.

---

## References

- `target-market-selection-pvp-index` — square #1: niche selection with
  the PVP Index (Personal fulfillment, Value, Profitability)
- `marketing-message-and-usp-crafting` — square #2: unique selling
  proposition, headline, and message construction
- `advertising-media-roi-framework` — square #3: channel selection and
  media ROI measurement
- `lead-capture-ethical-bribe-design` — square #4: lead magnets and
  ethical bribe design
- `lead-nurture-sequence-design` — square #5: multi-touch nurture
  sequences for unconverted leads
- `sales-conversion-trust-system` — square #6: trust-building conversion
  systems for the final buying decision
- `customer-experience-systems-design` — square #7: world-class experience
  delivery systems
- `customer-lifetime-value-growth` — square #8: upsell, cross-sell, and
  retention mechanisms
- `referral-system-design` — square #9: systematic referral orchestration
- `irresistible-offer-builder` — cross-cutting: offer construction that
  feeds squares #2 (message) and #6 (conversion)
- Source: *The 1-Page Marketing Plan* by Allan Dib — Introduction
  (framework, canvas, 3 acts), Conclusion (anti-patterns, 80% principle,
  marketing-as-process). Canvas downloadable at 1pmp.com.

## License

This skill is licensed under [CC-BY-SA-4.0](https://creativecommons.org/licenses/by-sa/4.0/).
Source: [BookForge](https://github.com/bookforge-ai/bookforge-skills) — The 1-Page Marketing Plan by Allan Dib.

## Related BookForge Skills

Install related skills from ClawhHub:
- `clawhub install bookforge-target-market-selection-pvp-index`
- `clawhub install bookforge-marketing-message-and-usp-crafting`
- `clawhub install bookforge-irresistible-offer-builder`
- `clawhub install bookforge-advertising-media-roi-framework`
- `clawhub install bookforge-lead-capture-ethical-bribe-design`
- `clawhub install bookforge-lead-nurture-sequence-design`
- `clawhub install bookforge-sales-conversion-trust-system`
- `clawhub install bookforge-customer-experience-systems-design`
- `clawhub install bookforge-customer-lifetime-value-growth`
- `clawhub install bookforge-referral-system-design`

Or install the full book set from GitHub: [bookforge-skills](https://github.com/bookforge-ai/bookforge-skills)

---
name: marketing-metrics-dashboard
version: 1.0.0
homepage: https://github.com/bookforge-ai/bookforge-skills/tree/main/books/the-1-page-marketing-plan/skills/marketing-metrics-dashboard
metadata: {"openclaw":{"emoji":"📚","homepage":"https://github.com/bookforge-ai/bookforge-skills"}}
status: published
quality:
  scores:
    with_skill: 92
    baseline: 31
    delta: 61
  tested_at: "2026-04-09"
  eval_count: 1
  assertion_count: 13
  iterations_needed: 1
description: >
  Build a marketing metrics dashboard tracking the 7 key numbers every small business must measure:
  Leads, Conversion Rate, Average Transaction Value (ATV), Break-Even Point, Gross Margin, Churn Rate,
  and Customer Lifetime Value (CLV) — plus Customer Acquisition Cost (CAC). Use this skill whenever the
  user wants to measure marketing performance, set up a KPI dashboard, track marketing ROI, fill in their
  key numbers, understand which metrics to watch, analyze leads and conversion, calculate CAC, compute CLV,
  identify profit improvement levers, run a marketing analytics review, understand small business metrics,
  build a marketing scorecard, or model what happens when they improve just 3 metrics by 10%. Demonstrates
  the compounding leverage insight: small improvements across multiple metrics multiply — not add — producing
  disproportionate profit gains. Also covers the CAC vs profit-per-sale decision rule for evaluating whether
  a marketing campaign is a winner or loser.
source:
  book: "The 1-Page Marketing Plan: Get New Customers, Make More Money, and Stand Out from the Crowd"
  author: "Allan Dib"
  chapters: ["Chapter 8: Increasing Customer Lifetime Value", "Chapter 3: Reaching Your Prospects"]
  pages: "237-241, 107-110"
tags: [marketing, metrics, analytics, small-business, kpi]
execution-tier: 1
execution-mode: full
density-score: 5
depends-on: []
---

# Marketing Metrics Dashboard

## When to Use

Use this skill when a business owner, solopreneur, or entrepreneur needs to:

- Identify which marketing numbers to track and why
- Build or populate a dashboard with current metric values
- Understand whether their marketing spend is profitable (CAC vs profit decision)
- Model the impact of small improvements across multiple levers
- Set targets and decision rules per metric before the next review cycle

**Preconditions:** The user should have at least rough estimates for their current business numbers. If they have none, the skill walks them through estimation. An existing dashboard can be updated or rebuilt from scratch.

**Before starting, verify:** Do you have current numbers, estimates, or are we starting from zero? Knowing the starting point determines whether this is a measurement session (fill in real numbers) or a planning session (set initial targets with estimates).

---

## Context & Input Gathering

### Required Context (ask if missing)

- **Business type and revenue model:** whether the business is transactional, subscription/recurring, or project-based — this determines which metrics matter most and which update cadence to use.
  → Check prompt for: descriptions of how they charge customers (per-job, monthly, subscription, retail)
  → If missing, ask: "Do customers buy once, come back repeatedly at their own pace, or pay you on a recurring subscription?"

- **Approximate monthly revenue range:** needed to sanity-check metric estimates and select the right dashboard format.
  → Check prompt for: dollar figures, revenue mentions
  → If missing, ask: "Roughly what is your monthly revenue — under $10K, $10K–$100K, or over $100K?"

### Observable Context (gather from environment)

- **Existing financial or marketing files:** look for any spreadsheets, CRM exports, ad spend reports, or revenue summaries in the working directory.
  → Look for: `*.csv`, `*.xlsx`, `revenue*.md`, `marketing*.md`, `dashboard*.md`
  → If unavailable: proceed with user-provided estimates; note all assumptions explicitly.

### Default Assumptions

- Measurement cadence: monthly (suitable for most small businesses; shift to weekly if revenue is high-velocity)
- Dashboard format: markdown document (universal; upgrade to Geckoboard or spreadsheet if user requests)
- CAC calculation: per-campaign basis, not blended average — unless user specifies otherwise

### Sufficiency Threshold

Proceed when: business type is known AND at least 3 of the 7 metrics have a current value or estimate.
Ask before proceeding when: no numbers are available at all — use estimation prompts to unblock.
Proceed with defaults when: some metrics are unknown; mark them as "TBD — estimate by [method]" in the dashboard.

---

## Process

### Step 1: Gather Current Metric Values

**ACTION:** Collect or estimate values for all 7 core metrics plus CAC. Work through each metric below. If the user doesn't know an exact number, ask them to estimate or calculate it from what they do know.

**WHY:** You cannot improve what you don't measure. Many business owners rely on gut feel or total revenue alone, which hides the levers that actually drive profit. Establishing baseline numbers — even rough estimates — is the prerequisite for all improvement.

For each metric, apply these collection questions:

| Metric | What to collect | Estimation fallback |
|--------|----------------|---------------------|
| **Leads** | New contacts entering the business per month | Count website visitors, inquiries, ad clicks, walk-ins |
| **Conversion Rate** | % of leads who become paying customers | (Customers acquired this month) ÷ (Leads this month) × 100 |
| **Average Transaction Value (ATV)** | Average dollar amount per purchase | Total revenue ÷ number of transactions |
| **Break-Even Point** | Monthly fixed cost floor | Sum: rent + staff + utilities + software + insurance + any recurring fixed costs |
| **Gross Margin** | Profit % after cost of goods/services | (Revenue − direct costs) ÷ Revenue × 100 |
| **Churn Rate** | % of recurring customers who cancel or stop buying per period | (Customers lost this period) ÷ (Customers at start of period) × 100 |
| **Customer Lifetime Value (CLV)** | Total revenue expected over a customer's full tenure | ATV × average purchases per year × average years retained |

**CAC (Customer Acquisition Cost):**
- Formula: Total campaign spend ÷ number of new customers acquired from that campaign
- Collect separately per marketing channel when possible

**IF** the user has no recurring customers → skip Churn Rate and CLV; note "not yet applicable — add subscription/repeat element to unlock."

**ELSE** → compute all 7 metrics and CAC.

---

### Step 2: Classify Each Metric by Health Status

**ACTION:** For each metric with a current value, assign a health status: **Healthy**, **Watch**, or **Act Now**. Use the interpretation guide below.

**WHY:** Raw numbers without context are meaningless. Classification converts data into decisions — it tells you where attention is required and where things can run on autopilot.

**Interpretation guide:**

| Metric | Healthy signal | Watch signal | Act Now signal |
|--------|---------------|--------------|----------------|
| Leads | Growing month-over-month | Flat 3+ months | Declining |
| Conversion Rate | >industry benchmark; improving | Flat | Declining or below 1% for transactional |
| ATV | Growing (upsell/pricing working) | Flat | Declining or customer pushback on price |
| Break-Even | Covered with >20% buffer | Covered with <10% buffer | Revenue at or below break-even |
| Gross Margin | >50% for digital/service; >30% for physical | 20–30% | <20% or declining |
| Churn Rate | <2%/month for subscriptions | 2–5%/month | >5%/month (leaky bucket problem) |
| CLV | CLV > 3× CAC | CLV 1–3× CAC | CLV < CAC (losing money per customer) |

---

### Step 3: Compute CAC vs Profit Decision

**ACTION:** For each active marketing campaign or channel, apply the CAC profitability test:

1. Calculate CAC = (campaign total spend) ÷ (customers acquired from campaign)
2. Calculate profit per new customer = ATV × Gross Margin % (front-end profit only)
3. Apply decision rule:
   - **If profit per sale > CAC → winning campaign** (keep, scale, or optimize)
   - **If profit per sale < CAC → losing campaign** (stop, unless CLV justifies front-end loss — see below)
   - **If CLV is high and well-understood → front-end loss may be acceptable** (subscription or high-repeat businesses can "go negative" on acquisition if lifetime value is proven)

**WHY:** Response rates and conversion rates in isolation are vanity metrics. The only number that matters is whether the campaign made money — which requires knowing both what you spent (CAC) and what you earned (profit per customer, optionally adjusted for CLV). A campaign with a 2% conversion rate can be a winner or a loser depending on these numbers.

---

### Step 4: Identify the 3 Highest-Leverage Metrics

**ACTION:** From the 7 core metrics, select the 3 that are:
(a) currently classified "Act Now" or "Watch", AND
(b) most directly under the business owner's control in the next 90 days

Priority ordering when multiple candidates exist:
1. Conversion Rate — often the highest-leverage single number because it multiplies everything downstream
2. Average Transaction Value — improves with upsells, bundles, or pricing changes; no new leads required
3. Leads — scalable but requires marketing spend or effort; slower to move

**WHY:** Attempting to improve all 7 metrics simultaneously spreads attention too thin and produces no meaningful result in any area. Focusing on 3 creates compounding: improvements multiply across each other rather than adding linearly, generating disproportionate profit impact from the same effort.

---

### Step 5: Model the 10% Compounding Scenario

**ACTION:** For each of the 3 chosen metrics, calculate what a 10% improvement produces — both individually and in combination.

Use this calculation sequence:

```
Current State:
  Total Conversions = Leads × Conversion Rate
  Total Revenue = Total Conversions × ATV
  Gross Profit = Total Revenue × Gross Margin %
  Net Profit = Gross Profit − Break-Even Point

After 10% improvement on all 3 chosen metrics:
  New Leads = Leads × 1.10  (if Leads is a chosen lever)
  New Conversion Rate = Conversion Rate × 1.10  (if Conv Rate is chosen)
  New ATV = ATV × 1.10  (if ATV is chosen)
  Recompute: New Total Conversions → New Revenue → New Gross Profit → New Net Profit

Net Profit Improvement % = (New Net Profit − Current Net Profit) / Current Net Profit × 100
```

**WHY:** The compounding effect is the most important insight in marketing metrics: because improvements multiply together, a 10% gain on 3 levers does not produce a 30% profit gain — it produces something far larger, especially when the break-even point is fixed. The worked example below shows a 431% net profit improvement from exactly this mechanism.

**Show both the "before" and "after" tables** so the magnitude of the leverage is visible.

---

### Step 6: Produce the Dashboard Document

**ACTION:** Write the completed dashboard as a markdown document. Include current values, health status, target values, decision rules, and the compounding scenario output.

**WHY:** A dashboard only works if it is reviewed regularly. A concrete document with clear decision rules and target values transforms the data into a system — something that gets checked weekly or monthly rather than forgotten. Tying metrics to decisions and owners makes it actionable.

Output format: see the **Outputs** section below for the complete dashboard template.

---

## Inputs

- Current business numbers: leads per month, conversion rate, ATV, fixed monthly costs, gross margin, churn rate (if recurring model), CLV estimate
- Marketing spend per campaign or channel (for CAC calculation)
- Business model type (transactional, recurring, project-based)
- Any existing dashboard files or financial summaries in the working directory

---

## Outputs

### Primary Output: `marketing-metrics-dashboard.md`

```markdown
# Marketing Metrics Dashboard
**Business:** [Business name]
**Period:** [Month/Quarter]
**Last Updated:** [Date]
**Update Cadence:** [Weekly / Monthly]

---

## Core Metrics Snapshot

| Metric | Current Value | Status | Target | Decision Rule |
|--------|--------------|--------|--------|---------------|
| Leads | [value] | [Healthy / Watch / Act Now] | [target] | If declining 2+ months → audit traffic source; review ad copy |
| Conversion Rate | [value]% | [status] | [target]% | If below [X]% → review sales process, guarantee, and follow-up sequence |
| Avg Transaction Value | [value] | [status] | [target] | If flat → add upsell, bundle, or raise prices on best-sellers |
| Break-Even Point | [value]/mo | [covered/at risk] | Fixed cost floor | If revenue ≤ break-even → freeze discretionary spend immediately |
| Gross Margin | [value]% | [status] | [target]% | If declining → review supplier costs or pricing; never discount without justification |
| Churn Rate | [value]%/mo | [status] | <[X]%/mo | If >5%/mo → prioritize retention over acquisition; diagnose top 3 exit reasons |
| Customer Lifetime Value | [value] | [status] | [target] | If CLV < 3× CAC → optimize retention before scaling acquisition |

---

## Customer Acquisition Cost (CAC) by Channel

| Channel | Spend | New Customers | CAC | Profit/Customer | Decision |
|---------|-------|--------------|-----|----------------|----------|
| [Channel 1] | [spend] | [count] | [CAC] | [profit] | Win / Loss / Front-end loss OK (high CLV) |
| [Channel 2] | | | | | |

**CAC Formula:** Total campaign spend ÷ customers acquired from that campaign
**Win condition:** Profit per sale > CAC
**Loss condition:** Profit per sale < CAC (stop unless CLV justifies it)

---

## 3 Highest-Leverage Metrics (90-Day Focus)

1. **[Metric 1]** — Current: [X], Target: [Y], Owner: [Name/Role]
   Action: [Specific intervention — e.g., "Add outrageous guarantee to checkout page"]

2. **[Metric 2]** — Current: [X], Target: [Y], Owner: [Name/Role]
   Action: [Specific intervention]

3. **[Metric 3]** — Current: [X], Target: [Y], Owner: [Name/Role]
   Action: [Specific intervention]

---

## Compounding Leverage Projection (10% on 3 Levers)

| | Before | After (+10% on 3 levers) |
|-|--------|--------------------------|
| Leads | [value] | [value × 1.10] |
| Conversion Rate | [X]% | [X × 1.10]% |
| Total Conversions | [L × CR] | [new L × new CR] |
| Average Transaction Value | [value] | [value × 1.10] |
| Total Revenue | [value] | [new convs × new ATV] |
| Gross Margin | [X]% | [X]% |
| Total Gross Profit | [rev × margin] | [new rev × margin] |
| Break-Even Point | [fixed] | [fixed] |
| **Net Profit** | **[GP − BE]** | **[new GP − BE]** |
| **Net Profit Improvement** | | **[improvement %]** |

**Key insight:** Fixed costs (break-even) don't increase when you improve these 3 levers.
Every additional dollar of gross profit above break-even flows directly to net profit.
This is why small percentage improvements compound into large net profit gains.

---

## Dashboard Format & Review Schedule

- **Format:** [Whiteboard with manual updates / Spreadsheet / Geckoboard / Internal web page]
- **Review frequency:** [Weekly / Monthly]
- **Review owner:** [Name]
- **Next review date:** [Date]

### Incentive Ties (optional)
- Team dinner if churn rate stays below [X]% this month
- Bonus pool unlocked when net profit exceeds [target]
```

---

## Key Principles

- **Measure 7 specific numbers, not everything** — most businesses track revenue and little else, which hides the levers driving it. The 7 metrics above cover the full revenue equation: how many enter (leads), how many convert, how much they spend, what it costs to run, and how long they stay.

- **Small improvements compound, not add** — a 10% gain on leads, conversion, and ATV produces far more than 30% profit growth because the break-even point is fixed. Every incremental dollar of gross profit above that fixed floor goes straight to net profit. This is the fundamental leverage point of direct-response marketing.

- **CAC is the north star of campaign evaluation** — response rates, click-through rates, and impressions are intermediate signals. The only campaign metric that matters is: did profit per customer exceed cost per customer acquired? If yes, scale it. If no, stop it or fix it.

- **Churn is the master retention metric for recurring businesses** — you cannot fill a leaking bucket. High churn means every new customer simply replaces a lost one, leaving revenue flat regardless of lead volume. Fix churn before scaling acquisition.

- **Front-end loss can be rational if CLV is known** — businesses with high customer lifetime value (subscriptions, repeat-purchase models) can profitably lose money on the first transaction because subsequent purchases more than compensate. This is only safe when CLV is well-understood from actual data, not assumption.

- **Track all 7, focus on 3** — keep the full dashboard current so you have situational awareness. But direct improvement energy at the 3 highest-leverage metrics for the current 90-day period. Attempting to optimize everything simultaneously produces nothing.

- **The dashboard must be visible and reviewed on a schedule** — a metrics document that no one checks is worthless. Tie a review meeting, team incentive, or personal habit to the cadence. The dashboard is an early-warning system, not a historical record.

---

## Examples

### Example 1: The Compounding Leverage Worked Case (Online Electronics Store)

**Scenario:** Small online consumer electronics store importing from China. Healthy 50% margin. Owner wants to improve profitability without increasing fixed costs.

**Starting metrics:**
- Leads: 8,000/month
- Conversion Rate: 5%
- Total Conversions: 400
- Average Transaction Value: $500
- Total Revenue: $200,000
- Gross Margin: 50%
- Total Gross Profit: $100,000
- Break-Even Point: $90,000/month (warehouse, staff, hosting)
- Net Profit: $10,000/month ($120,000/year)

**Three interventions applied:**
1. More compelling ad copy → Leads increase from 8,000 to 8,800 (+10%)
2. Outrageous risk-reversal guarantee → Conversion Rate increases from 5% to 5.5% (+10%)
3. Checkout upsell offer → ATV increases from $500 to $550 (+10%)

**After metrics:**
- Leads: 8,800
- Conversion Rate: 5.5%
- Total Conversions: 484
- ATV: $550
- Total Revenue: $266,200
- Gross Margin: 50%
- Total Gross Profit: $133,100
- Break-Even Point: $90,000 (unchanged)
- Net Profit: $43,100/month ($517,200/year)

**Result:** Net profit improved from $10,000 to $43,100 — a **431% improvement** — from only 10% gains across 3 levers. Owner income increased from $120,000 to $517,200 annually. Fixed costs did not change.

**Why the leverage is so large:** Each intervention multiplied the others. More leads × higher conversion = more customers. More customers × higher ATV = more revenue. Fixed break-even point meant all incremental gross profit became net profit. The math compounds; it does not add.

---

### Example 2: Local Fitness Studio (Subscription Model)

**Scenario:** A fitness studio with 200 active members at $80/month. Owner notices revenue is flat despite steady new member sign-ups.

**Dashboard reveals:**
- Leads: 40 free-trial bookings/month
- Conversion Rate: 50% (20 new members/month)
- ATV: $80/month
- Break-Even: $12,000/month
- Gross Margin: 70%
- Churn Rate: 10%/month ← Act Now flag
- CLV: $80 × 12 months average = $960

**CAC analysis:**
- Google Ads spend: $1,000/month → 10 new members → CAC = $100
- Profit per new member front-end: $80 × 70% = $56 (front-end loss of $44)
- CLV-adjusted: $960 × 70% = $672 gross profit over tenure → campaign is a strong winner on CLV basis

**Problem identified:** Churn at 10%/month means losing 20 members/month — exactly equal to new acquisition. Revenue stays flat. Filling the bucket but it leaks as fast as it fills.

**3 highest-leverage focus (90 days):**
1. Churn Rate: target <4%/month — introduce member check-ins at week 3, 6, 10 (high-cancellation risk windows)
2. ATV: target $90/month — introduce premium membership tier with one-on-one coaching add-on
3. Conversion Rate: target 60% — improve free-trial experience with structured onboarding session

**Compounding projection (if all 3 improve 10%):**
- Churn improvement alone: 10% → 9% = 2 fewer lost members/month = net positive growth begins
- Combined effect: revenue trajectory turns from flat to growing; no increase in marketing spend required

---

## References

- For calculating Customer Lifetime Value in detail and the 5 levers to increase it (raise prices, upsell, ascension, frequency, win-back): see Chapter 8 of the source book
- For campaign-level CAC analysis and the 3-scenario decision framework (stop / measure / scale): see Chapter 3 of the source book
- For tools to pull metrics automatically: Geckoboard integrates with common small business platforms (Stripe, Shopify, Google Analytics, QuickBooks)
- For dashboard cadence guidance: weekly works well for high-velocity businesses (e-commerce); monthly is appropriate for service, consulting, or project-based businesses

## License

This skill is licensed under [CC-BY-SA-4.0](https://creativecommons.org/licenses/by-sa/4.0/).
Source: [BookForge](https://github.com/bookforge-ai/bookforge-skills) — The 1-Page Marketing Plan by Allan Dib.

## Related BookForge Skills

No direct dependencies. Install the full book set from GitHub.

Or install the full book set from GitHub: [bookforge-skills](https://github.com/bookforge-ai/bookforge-skills)

---
name: marketing-message-and-usp-crafting
description: "Use this skill to craft a differentiated Unique Selling Proposition
  (USP), write a Problem/Solution/Proof elevator pitch, and engineer headlines that
  activate the 5 core emotional buying motivators. Triggers when a user asks to write
  a USP, unique selling proposition, craft a marketing message, build an elevator
  pitch, differentiate from competitors, answer 'why buy from me', write headlines,
  write copywriting or marketing copy, fix positioning, escape me-too marketing,
  stop competing on price, apply emotional marketing, target fear/love/greed/guilt/
  pride in copy, identify pain points for messaging, write sales copy, or fill
  square #2 of the 1-Page Marketing Plan canvas. Also activates for 'quality and
  great service as USP', 'we offer the best service', 'how do I stand out',
  'nobody knows what makes us different', 'my ads are not working', or similar
  messaging and positioning questions."
version: 1.0.0
homepage: https://github.com/bookforge-ai/bookforge-skills/tree/main/books/the-1-page-marketing-plan/skills/marketing-message-and-usp-crafting
metadata: {"openclaw":{"emoji":"📚","homepage":"https://github.com/bookforge-ai/bookforge-skills"}}
status: published
source-books:
  - id: the-1-page-marketing-plan
    title: "The 1-Page Marketing Plan"
    authors: ["Allan Dib"]
    chapters: [2]
tags:
  - marketing
  - copywriting
  - messaging
  - positioning
  - usp
  - small-business
depends-on:
  - target-market-selection-pvp-index
execution:
  tier: 1
  mode: full
  inputs:
    - type: document
      description: >
        Target market output (target-market.md from target-market-selection-pvp-index)
        OR a description of the business's primary customer segment and their top
        pain points, if target-market.md is not available.
  tools-required: [Read, Write]
  tools-optional: []
  mcps-required: []
  environment: >
    Document set — business description and target market notes in markdown.
    No code execution required.
discovery:
  goal: >
    Help small business owners craft a marketing message that compels their target
    market to act — producing a USP statement, a Problem/Solution/Proof elevator
    pitch, and 3–5 headline drafts — all in a single marketing-message.md deliverable.
  tasks:
    - "Gather target market and pain point context"
    - "Draft USP candidates and select the strongest one using the logo-swap test"
    - "Build an elevator pitch using the Problem/Solution/Proof formula"
    - "Select one or more emotional buying motivators from the 5-motivator taxonomy"
    - "Draft 3–5 headlines that activate those motivators"
    - "Audit all outputs against anti-patterns (quality/service USP, price USP,
       me-too positioning)"
    - "Write marketing-message.md with the complete message system"
  audience:
    roles:
      - small-business-owner
      - solopreneur
      - entrepreneur
      - freelancer
      - startup-founder
    experience: beginner-to-intermediate
  when_to_use:
    triggers:
      - "User wants to write or improve a USP"
      - "User is crafting a marketing message or sales copy"
      - "User wants to differentiate from competitors"
      - "User is writing headlines or ad copy"
      - "User is filling square #2 of the 1-Page Marketing Plan"
      - "User wants to stop competing on price"
    prerequisites:
      - "Primary target market known (from target-market-selection-pvp-index or
         provided directly)"
    not_for:
      - "Enterprise brand strategy requiring market research agencies and large
         positioning studies"
      - "Businesses without any product or service yet (message cannot precede offer)"
  environment:
    codebase_required: false
    codebase_helpful: false
    works_offline: true
  quality:
    scores:
      with_skill: 100
      baseline: 36
      delta: 64
    tested_at: "2026-04-09"
    eval_count: 1
    assertion_count: 14
    iterations_needed: 1
---

# Marketing Message and USP Crafting

A structured process for small business owners to stop blending in and start
standing out. Produces a Unique Selling Proposition (USP), a
Problem/Solution/Proof elevator pitch, and emotionally-engineered headline
drafts — the complete message system for square #2 of the 1-Page Marketing
Plan canvas.

Most small business ads are interchangeable: company name, logo, laundry list
of services, claim of "best quality and service," offer of a free quote. Swap
the name and logo — it could be any competitor. This skill eliminates that
failure by forcing explicit answers to the two questions every prospect asks
silently: "Why should I buy?" and "Why should I buy from you?"

---

## When to Use

Use this skill AFTER selecting your primary target market (via
`target-market-selection-pvp-index` or direct user input). Marketing message
depends entirely on knowing exactly who you are speaking to — the same business
must write completely different copy for different customer segments.

Also use it when:
- An existing business is getting price-shopped and doesn't know why
- Current ads produce no response or "me too" comparison shopping
- A business owner can't answer "What makes you different?" with a single clear
  sentence
- A new campaign is being built from scratch

Do NOT use this skill as a substitute for offer construction (pricing, bonuses,
guarantees). Message comes before offer in execution order, but the offer itself
is a separate skill.

---

## Context and Input Gathering

### Required (must have before proceeding)

**IF target-market.md exists** (output of `target-market-selection-pvp-index`):
  → Read it. Extract: primary target segment name, customer avatar, dominant
    emotion, biggest fears, daily frustrations, what they secretly want.

**IF target-market.md does not exist**:
  → Ask the user directly:
  1. "Who is your primary target customer? Be specific — describe a type of
     person, not a broad demographic."
  2. "What is the single biggest problem or frustration they face that your
     business can solve?"
  3. "What outcome do they actually want — not the feature you provide, but
     the result they are trying to achieve?"

Also gather:
- What the business sells (product/service description)
- 2–3 nearest competitors and what makes each of them different (or indistinct)
- Any existing USP attempts, taglines, or elevator pitch (to diagnose failures)

### Sufficiency check

You have enough to proceed when you can name: (1) who the customer is,
(2) what pain they are living with right now, and (3) what specific outcome the
business delivers. If any of these three is vague, ask before proceeding.

---

## Process

### Step 1: Identify the result the customer is actually buying

Ask: "What is the prospect buying — not the product or service, but the result
or outcome they want to achieve?"

A printer is not selling business cards. They are selling "more customers
walking through the door." A security system company is not selling cameras.
They are selling "the feeling that your family is safe when you are not home."
A management consultant is not selling advice. They are selling "operations
that scale without breaking."

Write a one-sentence outcome statement: "[Target customer] wants [specific
outcome], not [product/service feature]."

**WHY:** Selling features turns prospects into price shoppers who compare
specifications across competitors. Selling outcomes positions you as a problem
solver and pain reliever. Prospects are willing to pay far more for a cure than
for a feature — the same way someone with a splitting headache will pay double
or triple for pain relief without shopping around.

### Step 2: Draft 3–5 USP candidates

Answer these two questions for the business, in clear and quantifiable terms:

1. Why should they buy? (What problem exists if they do not act?)
2. Why should they buy from you? (What do you offer that no competitor offers
   in the same way — in the product, delivery, packaging, support, guarantee,
   or experience?)

For each candidate USP, write a single sentence that a prospect could read in
three seconds and immediately understand what is different and why it matters.
The uniqueness does not need to be in the product itself. It can be in:
- How it is delivered (e.g., installed and configured in your home vs. box
  in bag)
- How it is supported (e.g., dedicated account manager vs. support ticket queue)
- How it is packaged (e.g., flat monthly fee vs. hourly billing)
- What guarantee backs it (e.g., results in 30 days or full refund)
- What experience surrounds it (e.g., the CD Baby confirmation email)

**WHY:** Very few products are truly unique. The uniqueness must come from
somewhere — packaging, delivery, experience, or guarantee are all valid.
Generating 3–5 candidates prevents anchoring on the first idea, which is
usually the most generic.

### Step 3: Apply the logo-swap test to each candidate (gate)

For each USP candidate, ask: "If I removed the business name and logo from this
statement and placed it on a competitor's website, would it still make sense?"

**IF YES** (it passes to any competitor): the USP has failed. It is a "me too"
statement. Return to Step 2 and generate a more specific candidate.

**IF NO** (it could only belong to this specific business): the USP is viable.
Move to Step 4.

Common failures of the logo-swap test:
- "We provide quality service at competitive prices." → Fails. Any competitor
  can claim this.
- "We have 20 years of experience." → Fails. Generic credential.
- "We are locally owned and operated." → Fails. Every local competitor qualifies.
- "We offer free consultations." → Fails. Any competitor can match this tomorrow.

**WHY:** The logo-swap test is the single fastest diagnostic for a weak USP.
If a statement is interchangeable, it gives the prospect no reason to choose
you — they default to price comparison, which is the worst competitive position
for a small business. There will always be someone willing to go out of business
faster than you by discounting further. The only escape is a differentiated
position.

### Step 4: Select and sharpen the strongest USP

From the candidates that passed the logo-swap test, select the one that:
- Targets the most emotionally relevant pain or desired outcome for the avatar
- Is the hardest for competitors to copy quickly
- Can be stated in a single clear sentence without jargon

Sharpen it by:
- Making it specific: replace vague words with numbers, timeframes, or
  observable outcomes
- Making it outcome-focused: rewrite any feature language as a result statement
- Making it prospect-facing: use "you" language, not "we" language

**WHY:** Specificity is credibility. "We save small businesses an average of
12 hours per week on payroll administration" is more convincing than "We save
you time." Specific claims are harder to refute and easier to remember.

### Step 5: Build the elevator pitch using Problem/Solution/Proof

Use the formula: **"You know how [problem]? Well, what I do is [solution].
In fact, [specific proof or result example]."**

Fill each component:

**Problem:** Describe the pain the target market is currently experiencing.
Use their language. The problem should feel immediate and real — something they
are living with today, not a future risk they might face.

**Solution:** Describe the outcome you deliver, not the features or mechanism.
Focus on the transformation: before state → after state.

**Proof:** A single specific result that happened for a real customer. Include
a number, a timeframe, or a named outcome. "In fact, just last week a client of
mine..." is more compelling than "Our clients have seen great results."

The full pitch should be deliverable in 30–90 seconds. It is both a networking
tool and a clarity mechanism — if you cannot say it in 90 seconds, your message
is not clear enough to use in any marketing.

**WHY:** Bad elevator pitches are product-focused and self-focused — they talk
about the business, not the prospect. The Problem/Solution/Proof structure
forces customer-focus at every step. Prospects respond when they hear their own
situation described accurately; they tune out when they hear a business
describe itself.

### Step 6: Select emotional buying motivator(s) for headlines

Identify which of the 5 core emotional buying motivators are most active for
this target avatar:

1. **Fear** — especially fear of loss, fear of missing out, fear of a bad
   outcome. The most powerful motivator. The amygdala (the brain's survival
   center) processes threats first. "Fear of loss" consistently outperforms
   "desire for gain" in response rates.

2. **Love** — desire for connection, relationships, family protection,
   belonging. Activates when the product protects or enhances something the
   prospect loves deeply.

3. **Greed** — desire for more: more money, more time, more opportunity, more
   of what they value. "More for me" at lower cost or higher return.

4. **Guilt** — not doing right by family, employees, self, or others.
   Underutilized but powerful for audiences with clear obligations (parents,
   employers, professionals with a duty of care).

5. **Pride** — status, exclusivity, membership in an elite group, being seen
   as smart or successful. "People like you use this."

Select the 1–2 motivators that best match the avatar's dominant emotional state
(from the customer avatar built in `target-market-selection-pvp-index`). If the
avatar is not available, infer from the problem description.

**WHY:** People buy with emotion and justify with logic afterward. Copy that
fails to activate at least one of these five motivators is timid and
ineffective — it generates polite indifference, not action. Identifying the
correct motivator before writing headlines prevents generic copy that "sounds
professional" but triggers nothing.

### Step 7: Draft 3–5 headlines

Write 3–5 headline variants, each activating the selected motivator(s). A
headline's job is not to describe the product — it is to grab attention and
compel the reader to keep reading. Think of it as the ad for the ad.

Headline structures that consistently work (adapt to the business):
- Direct pain: "Attention [target market]: Are You Still [painful situation]?"
- Specific result: "How [business type] Clients [specific result] in [timeframe]"
- Fear of loss: "The [N] Most Expensive [mistakes/errors] [target market] Make
  — And How to Avoid Them"
- Social proof + fear: "Why [N] [target market] in [location] Have Switched to
  [solution]"
- Enemy in common: "[Problem] Is Costing You [specific loss] — Here's What
  to Do About It"
- Pride/exclusivity: "For [target market] Who Refuse to [accept painful status
  quo]"

Use emotionally charged words: Free, You, Save, Results, Proven, Money, New,
Easy, Safety, Guaranteed, Discovery. One word substitution can shift response.

**WHY:** The headline is read first and determines whether anything else gets
read. A weak headline means zero return on all other copy effort. Writing 3–5
variants allows testing — only real response data reveals which motivator
resonates most with this specific audience.

### Step 8: Audit all outputs against anti-patterns

Before finalizing, check each element against these failure modes:

| Anti-pattern | Test | Correct action |
|---|---|---|
| Quality/service USP | "Is 'quality' or 'great service' in the USP?" | Replace with specific, measurable differentiator |
| Price USP | "Does the USP rely on being cheapest?" | Shift to value, outcome, or experience-based differentiation |
| Me-too positioning | Does the logo-swap test fail? | Return to Step 3 |
| Feature language | Does copy describe specs rather than outcomes? | Rewrite as result statements |
| Prevention framing | Is copy selling future safety rather than current pain? | Reframe to address existing pain |
| Self-focused | Does copy say "we" more than "you"? | Flip to prospect perspective |

**WHY:** Anti-patterns are the default — they feel natural because they are how
most businesses talk about themselves. The audit step converts the instinctive
output into direct-response copy. Skipping this step typically means the final
document contains at least one critical failure that renders the whole message
generic.

### Step 9: Write marketing-message.md

Compile all outputs into a single deliverable. Save it as `marketing-message.md`
in the user's working directory.

**WHY:** Square #2 of the 1-Page Marketing Plan must be documented. Without a
written record, the message reverts under pressure to familiar but ineffective
patterns. The document also serves as the creative brief for all downstream
marketing: ads, landing pages, email sequences, and sales scripts.

---

## Inputs

| Input | Format | Required |
|-------|--------|----------|
| target-market.md (from target-market-selection-pvp-index) | .md | Preferred |
| Primary target segment + top pain points | direct user input | If no target-market.md |
| Business description (product/service) | text | Yes |
| Existing USP attempts or taglines | text | Recommended |
| 2–3 nearest competitors | text | Recommended |

---

## Outputs

Primary output: `marketing-message.md`

```markdown
# Marketing Message: [Business Name]

## Unique Selling Proposition

**USP Statement:**
[Single sentence. Outcome-focused. Passes logo-swap test.]

**Logo-swap test result:** PASS — [brief rationale for why this could only
belong to this business]

**Why it works:** [1–2 sentences connecting USP to target avatar's dominant
pain or desired outcome]

---

## Elevator Pitch (Problem/Solution/Proof)

**Problem:**
[1–2 sentences. Target market's current pain, in their language.]

**Solution:**
[1–2 sentences. Outcome delivered, not features provided.]

**Proof:**
[1 sentence. Specific result with a number, timeframe, or named outcome.]

**Full pitch (30–90 seconds):**
"You know how [problem]? Well, what I do is [solution]. In fact, [proof]."

---

## Headline Drafts

**Primary motivator(s) selected:** [Fear / Love / Greed / Guilt / Pride]
**Rationale:** [1 sentence connecting motivator to avatar's dominant emotion]

1. [Headline 1]
2. [Headline 2]
3. [Headline 3]
4. [Headline 4 — optional]
5. [Headline 5 — optional]

**Recommended first test:** Headline [N] — [brief reason]

---

## Anti-Pattern Audit

| Check | Result |
|-------|--------|
| Quality/service USP | CLEAR |
| Price USP | CLEAR |
| Logo-swap test | PASS |
| Feature vs. outcome language | CLEAR |
| Pain vs. prevention framing | CLEAR |
| Prospect-focused ("you" > "we") | CLEAR |

---

_Square #2 of the 1-Page Marketing Plan canvas: filled._
```

---

## Key Principles

**1. Logo-swap test is the gate.**
If your marketing message could belong to any competitor with a name swap, it
has failed — regardless of how professional it sounds. Every USP candidate must
pass this test before moving forward.

**2. Niching makes price irrelevant.**
A specialist commands higher fees than a generalist. A heart surgeon is not
compared to a general practitioner on price. The narrow niche that feels risky
is almost always more profitable than the broad positioning that feels safe.

**3. Emotional response drives purchasing; logic justifies afterward.**
"I bought the Porsche for safety and reliability." Sell to the emotion first.
Copy that leads with facts and specifications is selling to the wrong brain.

**4. Sell to existing pain, not future prevention.**
People pay far more for a cure than for prevention. Someone with a splitting
headache pays double without shopping around. Target the pain your prospect is
carrying today — not the risk they might face tomorrow.

**5. Selling features turns prospects into price shoppers.**
Once a prospect can compare you on a feature, they compare you on price.
Outcomes — "12 hours saved per week" — cannot be commodity-shopped the way
features can. Always translate features into results.

**6. Confusion kills sales silently.**
A confused prospect does not ask for clarification — they leave. Choose clarity
over cleverness in every element of your message.

---

## Examples

### Example 1: Wedding photographer (commodity service, emotional niche)

**Context:** Photographer serves four markets. PVP analysis selected family
portrait photography as primary. Avatar: Sarah, 34, new parent. Dominant
emotion: anticipation + anxiety about not capturing the moment perfectly.

**Weak USP (fails logo-swap):**
"Professional photography for families who want beautiful memories." — Any
photographer can claim this. Fails.

**Strong USP (passes):**
"Family portraits so vivid your kids will fight over who gets to hang them —
guaranteed or the session is free."
— Specific emotional outcome (kids fighting over prints = the photo is loved),
with a guarantee that removes risk. Could not belong to a competitor without
copying the exact guarantee.

**Elevator pitch:**
"You know how most family portrait photographers give you a USB drive of 300
photos and you pick through them yourself, then order prints from a box store
that look nothing like the preview? Well, what I do is guide the whole
experience from the session through to framed prints on your wall. In fact, my
last client told me her daughter cried when she saw her portrait and asked if
she could have it in her bedroom."

**Motivator selected:** Love (protecting and celebrating family)

**Headline drafts:**
1. "The Family Portrait Your Kids Will Actually Ask to Hang on Their Wall"
2. "Warning: This Photo Session May Cause Arguments Over Who Gets the Best Shot"
3. "For [City] Families Who Refuse to Settle for Generic Studio Portraits"
4. "What If Your Family Portrait Was So Good Your Kids Showed It Off at School?"

---

### Example 2: SaaS project management tool for agencies

**Context:** Digital agencies with 5–20 person teams. Avatar: agency owner
overwhelmed by missed deadlines. Dominant emotion: fear of losing clients.

**Weak USP (fails logo-swap):**
"Project management software that keeps your team organized and on track." —
Every competitor says this. Fails.

**Strong USP (passes):**
"The project management platform built exclusively for agencies that can't
afford to miss another client deadline — with automated client status updates
so you stop being the bottleneck."
— Specificity (agencies only, automated updates, bottleneck pain) prevents
any competitor from claiming this without copying the exact position.

**Elevator pitch:**
"You know how agency owners spend half their day answering client emails asking
'where are we on the project?' — while their team is actually making progress?
Well, what we do is send automated plain-English status updates to your clients
every Friday so you never have to write that email again. In fact, one of our
customers went from spending 8 hours a week on client communication to under
30 minutes — without a single client noticing the change."

**Motivator selected:** Fear (losing clients due to communication failures)

**Headline drafts:**
1. "Agency Owners: How Many Clients Have You Lost to Missed Deadlines This Year?"
2. "The Hidden Cost of Being Your Agency's Human Status Update System"
3. "Finally: Client Communication That Runs Itself — Even When You're in a Sprint"
4. "For Agencies Tired of Doing Great Work and Still Getting Blamed for Poor
   Communication"

---

### Example 3: Independent bookkeeper for small retail businesses

**Context:** Retail business owners with 1–5 staff. Avatar: shop owner doing
her own books at 11 PM. Dominant emotion: guilt + fear of getting it wrong.

**Weak USP (fails logo-swap):**
"Accurate, reliable bookkeeping for small businesses at affordable rates." —
Generic. Fails.

**Strong USP (passes):**
"Monthly bookkeeping for retail owners — done by the 5th of every month,
with a plain-English one-page summary you can actually read, or your money back."
— The deadline (5th of month), the deliverable format (one-page plain-English),
and the guarantee together create a position no generic competitor holds.

**Elevator pitch:**
"You know how retail shop owners end up doing their own books at 11 PM on a
Sunday because they can't trust that their bookkeeper understands their
inventory situation? Well, what I do is handle everything and send you a
one-page summary every month — in plain English, no jargon — by the 5th.
In fact, one of my clients told me it was the first time in three years she
felt like she actually understood her own numbers."

**Motivator selected:** Guilt + Fear

**Headline drafts:**
1. "Are You Still Doing Your Own Bookkeeping at 11 PM on Sundays?"
2. "What Would You Do With 6 Extra Hours Every Month?"
3. "Free Report: The 5 Bookkeeping Mistakes Most Retail Owners Make at Tax Time"

---

## References

- `target-market-selection-pvp-index/SKILL.md` — Dependency. Read before this
  skill if the primary target market has not been selected.
- `.meta/book-profile.json` — Full book metadata and chapter mappings
- `.meta/research/hunter-report.md` — sk-03 entry: messaging and USP content
  identified across the chapter
- Source: Ch 2 "Crafting Your Message," pp 59–79 (USP, elevator pitch) and
  pp 92–102 (emotional motivators, headlines, copywriting), Allan Dib

## License

This skill is licensed under [CC-BY-SA-4.0](https://creativecommons.org/licenses/by-sa/4.0/).
Source: [BookForge](https://github.com/bookforge-ai/bookforge-skills) — The 1-Page Marketing Plan by Allan Dib.

## Related BookForge Skills

Install related skills from ClawhHub:
- `clawhub install bookforge-target-market-selection-pvp-index`

Or install the full book set from GitHub: [bookforge-skills](https://github.com/bookforge-ai/bookforge-skills)

---
name: lead-nurture-sequence-design
description: "Use this skill to design a complete lead nurture system for a small
  business. Triggers when a user wants to set up a nurture sequence, email sequence,
  drip campaign, CRM follow-up system, shock and awe package, marketing calendar,
  or lead nurturing automation. Also activates for 'fill square 5', 'square #5 of
  the 1-Page Marketing Plan', 'how do I follow up with leads', 'CRM setup', 'email
  marketing', 'repeat customers', 'Joe Girard', 'drip campaign', 'marketing
  calendar', 'event-triggered automations', 'the money is in the follow-up',
  'shock and awe package', 'physical mail marketing', 'nurture leads', 'prospect
  database', or 'I captured leads but don't know what to do with them'."
version: 1.0.0
homepage: https://github.com/bookforge-ai/bookforge-skills/tree/main/books/the-1-page-marketing-plan/skills/lead-nurture-sequence-design
metadata: {"openclaw":{"emoji":"📚","homepage":"https://github.com/bookforge-ai/bookforge-skills"}}
status: published
source-books:
  - id: the-1-page-marketing-plan
    title: "The 1-Page Marketing Plan"
    authors: ["Allan Dib"]
    chapters: [5]
tags:
  - marketing
  - lead-nurture
  - email-marketing
  - crm
  - automation
  - small-business
depends-on:
  - lead-capture-ethical-bribe-design
execution:
  tier: 1
  mode: full
  inputs:
    - type: document
      description: >
        Business description, target market, captured lead list or CRM platform
        in use, and any existing email sequences or follow-up process. The lead
        capture system (from lead-capture-ethical-bribe-design) should already
        be in place, or the user should describe their current lead sources.
  tools-required: [Read, Write]
  tools-optional: []
  mcps-required: []
  environment: >
    Document set — business description, target market, and lead capture
    context. No code execution required.
discovery:
  goal: >
    Help the user build a complete lead nurture infrastructure: CRM selection,
    a 5-part 30-day email sequence, a shock and awe package (for inbound
    inquiries), a 5-cadence marketing calendar, an event-trigger map, and
    role assignments. Produces a nurture-sequence.md document with all
    components ready to implement.
  tasks:
    - "Inventory current nurture state and identify gaps"
    - "Confirm or select CRM platform"
    - "Design 5-part email sequence over 30 days (value-first)"
    - "Plan shock and awe package contents and delivery trigger"
    - "Build marketing calendar (daily / weekly / monthly / quarterly / annual)"
    - "Map event triggers to automated responses"
    - "Assign roles: Entrepreneur, Specialist, Manager"
    - "Write nurture-sequence.md with all outputs"
  audience:
    roles:
      - small-business-owner
      - solopreneur
      - entrepreneur
      - freelancer
      - startup-founder
    experience: beginner-to-intermediate
  when_to_use:
    triggers:
      - "User has captured leads but has no follow-up system"
      - "User is filling square #5 of the 1-Page Marketing Plan"
      - "User wants to set up automated email sequences or drip campaigns"
      - "User wants to design a shock and awe package for inbound inquiries"
      - "User wants to build a marketing calendar with recurring activities"
      - "User wants to map event-triggered automations in their CRM"
    prerequisites:
      - "Lead capture system in place (use lead-capture-ethical-bribe-design first)"
    not_for:
      - "Enterprise marketing automation teams with dedicated platforms already running"
      - "Businesses with no lead list — build that first with lead-capture-ethical-bribe-design"
  environment:
    codebase_required: false
    codebase_helpful: false
    works_offline: true
  quality:
    scores:
      with_skill: 100
      baseline: 35.7
      delta: 64.3
    tested_at: "2026-04-09"
    eval_count: 1
    assertion_count: 14
    iterations_needed: 1
---

# Lead Nurture Sequence Design

A structured process for small business owners to turn a captured lead list into
a revenue-generating relationship system. Based on the principle that 50% of
salespeople give up after one contact, 65% after two, and 79.8% after three —
meaning the business that keeps showing up with value wins by default. Produces
a complete nurture-sequence.md with CRM choice, email sequence outline, shock and
awe package plan, a 5-cadence marketing calendar, an event-trigger map, and role
assignments. Required content for square #5 of the 1-Page Marketing Plan canvas.

---

## When to Use

Use this skill immediately after lead capture is set up. The job of square #5 is
to convert interested prospects — who are not yet ready to buy — into raving fan
customers through consistent, value-first contact over time.

Also use it when:
- Leads are being captured but no one is following up consistently
- Email sequences exist but are pitch-heavy and getting ignored
- The business has no marketing calendar — every week starts from scratch
- A shock and awe package for inbound inquiries has not been designed
- The business is growing but marketing feels like random acts rather than a system

**Dependency check:** If leads are not yet being captured, invoke
`lead-capture-ethical-bribe-design` first — OR ask the user to describe their
current lead sources (business card collection, website forms, referrals, etc.)
before proceeding. This skill requires a list to nurture.

---

## Context & Input Gathering

### Required (must have before proceeding)
- Target market: who the business serves and their top pain points
- Current lead sources: how leads arrive (website form, events, referrals, ads)
- CRM platform in use (or willingness to choose one)

### Observable / inferrable
- Content assets available (blog posts, videos, guides, testimonials)
- Whether inbound inquiries are high enough volume to justify a physical shock
  and awe package
- Whether a Manager-type role exists or needs to be filled or outsourced

### Defaults (apply if not provided)
- CRM: recommend ActiveCampaign (automation), ConvertKit (content-first), or
  HubSpot free tier (all-in-one) based on business type
- Email sequence: default to 5-part, 30-day cadence (Days 1, 3, 7, 14, 30)
- Shock and awe package: recommend for any business with inbound inquiries and
  customer lifetime value above $500
- Marketing calendar: apply the 5-cadence default (daily/weekly/monthly/
  quarterly/annual) from the book verbatim

### Sufficiency check
Proceed when you can name: (1) the target market and their top pain point,
(2) how leads currently arrive, (3) a CRM or email platform to anchor the system.

---

## Process

### Step 1: Inventory the current nurture state

Ask the user to describe what happens after a lead is captured today. Map the
answer against the five infrastructure components: CRM, email sequences, physical
mail, shock and awe package, and event-triggered automations.

For each component, mark: Exists / Partial / Missing.

**WHY:** Most small businesses have one or two pieces in place (usually an email
platform and an occasional newsletter) but none of the components are connected
into a system. The inventory reveals which gaps are causing the most revenue
leakage. A business with no CRM cannot automate anything — fix that first.
A business with a CRM but no sequences is nurturing manually (or not at all).
The inventory sets the build order.

**Anti-pattern to watch for:** A single follow-up call after the lead capture,
followed by nothing. This is the most common broken pattern — and statistically,
79.8% of competitors stop at three contacts. Persistence beyond three contacts
puts you in the top 20.2% of all follow-up effort.

### Step 2: Confirm or select the CRM platform

The CRM is the nerve center of the entire nurture system. Every sequence, trigger,
contact record, and calendar task runs through it. Without a CRM, none of the
steps below can be automated or tracked.

**Selection criteria:**

| Platform | Best for | Automation depth |
|----------|----------|-----------------|
| ActiveCampaign | Service businesses, complex sequences | High |
| ConvertKit | Content creators, solopreneurs | Medium |
| HubSpot (free) | Small teams wanting all-in-one | Medium |
| Mailchimp | Simple lists, budget-constrained start | Low |
| Keap (Infusionsoft) | High-volume, complex funnels | Very high |

Recommendation rule: if customer lifetime value is above $1,000, invest in
ActiveCampaign or Keap. If under $500, ConvertKit or Mailchimp is sufficient
to start.

**WHY:** The CRM is not a cost — it is the infrastructure that makes every other
marketing activity compound over time. A business without a CRM is doing
marketing by hand, which means marketing stops the moment the owner stops.
Automation is what makes the system recur without owner involvement.

### Step 3: Design the email sequence (5-part, 30-day default)

Build a value-first email sequence triggered immediately when a new lead enters
the CRM. The default structure is 5 emails over 30 days.

**The 90/10 rule:** 90% of emails should deliver value (education, insight,
relevant content). Only 10% should make an offer. Pitch-heavy sequences erode
trust and increase unsubscribes.

**Default 5-part sequence structure:**

| Day | Email purpose | Tone |
|-----|--------------|------|
| 1 | Welcome + deliver the ethical bribe | Warm, helpful |
| 3 | Insight or tip relevant to their pain point | Educational |
| 7 | Case study or story of a problem like theirs solved | Story-driven |
| 14 | Address the most common objection or fear | Trust-building |
| 30 | Soft offer or invitation (not a hard sell) | Conversational |

**WHY:** The sequence is spaced to match the typical buying cycle, not the
business's impatience. A prospect who requested a free guide on Day 1 is not
ready to buy on Day 2. The sequence builds familiarity, establishes authority,
and keeps the business top-of-mind so that when the prospect IS ready to buy,
they think of this business first — not a competitor.

**Subject line principle:** The subject line determines whether the email gets
opened. It must be specific, relevant, and curiosity-inducing. A good test: would
you open it if it arrived in your inbox from a stranger?

**Length principle:** Length is irrelevant. Relevance is the criterion. A 1,000-
word email that is interesting will be read. A 200-word email that is boring
will not.

### Step 4: Plan the shock and awe package

A shock and awe package is a physical box mailed to high-probability inbound
prospects — people who have specifically reached out to inquire about the
business's product or service.

**When to use:** Triggered by an inbound sales inquiry (not applied to every
lead). It is the response to "send me more information" — the business sends a
physical package via a tracked courier service rather than a PDF link.

**Delivery method:** Send via FedEx or a tracked courier. A FedEx box on a desk
gets opened. A PDF email gets ignored. No one ignores a package with their name
on it.

**Contents to include (select 4–6 relevant to the business):**

| Item | Purpose |
|------|---------|
| A book (ideally one you authored) | Positions as expert; books are rarely discarded |
| DVD or USB with video introduction | Shows personality; demonstrates problem-solving |
| Testimonials (video, audio, or printed) | Social proof at the moment of highest interest |
| Media clippings or feature articles | Third-party authority |
| Brochures or sales letters | Specific offer details |
| Independent report or white paper | Demonstrates expertise |
| Product sample or gift card | Tangible value, motivates trial |
| Handwritten note | Personal touch; cuts through digital noise |

**Three jobs the package must do:**
1. Give unexpected value — the prospect should think "wow" not "brochure"
2. Position the business as the expert authority — not just another option
3. Move the prospect further down the buying cycle than they would otherwise be

**Budget check:** Calculate customer lifetime value first. If a new customer is
worth $5,000, a $50 shock and awe package is a 1% acquisition investment. The
numbers must make sense, but for most service businesses, they do.

**WHY:** Most businesses respond to inbound inquiries with the cheapest possible
reply: a link, a PDF, or a call. That is a same-same or crappy impression. A
shock and awe package creates a mind-blowingly amazing first impression at the
highest-intent moment in the buying process. Competitors almost never do this —
making it a durable competitive advantage.

### Step 5: Build the marketing calendar

A marketing calendar enforces consistency by treating marketing activities as
non-negotiable scheduled commitments, the same way tax deadlines are
non-negotiable. Set activities at five cadences:

| Cadence | Activity | Owner |
|---------|----------|-------|
| Daily | Check social media for mentions; respond appropriately | Manager |
| Weekly | Write one blog post; send link in email broadcast to list | Specialist |
| Monthly | Mail customers and prospects a printed newsletter or postcard | Manager |
| Quarterly | Send reactivation letter to past customers who haven't purchased recently | Specialist |
| Annually | Send all customers a gift basket thanking them for their business | Entrepreneur approves, Manager executes |

**Customize the table** for the business's specific channels and resources. The
cadences above are the book's defaults — adapt them, but do not reduce the
number of cadences. All five are required for a functioning marketing calendar.

**WHY:** Without a calendar, marketing happens when the business owner "has time"
— which means it happens rarely. The calendar creates the forcing mechanism.
Joe Girard sent 13,000 cards per month, every month, for over a decade. He did
not do it when he felt inspired. He did it because it was his system.

### Step 6: Map event-triggered automations

In addition to the scheduled calendar, certain business events should trigger
specific marketing responses automatically. Map each trigger to its response:

| Trigger event | Automated / assigned response |
|--------------|-------------------------------|
| Prospect met at networking event | Enter CRM; add to monthly newsletter list |
| Inbound sales inquiry received | Send shock and awe package via courier |
| New email subscriber from blog or ad | Start 5-part video or email series (30 days) |
| Customer complaint resolved | Send handwritten apology note + $100 gift voucher |
| Customer makes first purchase | Send welcome gift or onboarding sequence |
| Prospect has not responded in 90 days | Reactivation email: new offer or fresh angle |

Add business-specific triggers as appropriate (e.g., annual contract renewal,
seasonal buying patterns, product launch).

**WHY:** Triggers eliminate decision fatigue. When a specific event happens, the
response is predetermined and automated. The business does not have to remember
to follow up — the CRM does it. Event-triggered automations are the highest-
leverage element of the nurture system because they fire at the moments of
highest prospect attention (an inquiry, a complaint, a first purchase).

### Step 7: Assign roles — Entrepreneur, Specialist, Manager

Every recurring marketing activity needs an owner. The three types are:

- **Entrepreneur (makes it up):** Designs the strategy, chooses campaigns,
  writes the vision for what the nurture system should accomplish.
- **Specialist (makes it real):** Executes specific tasks — writes email copy,
  designs the shock and awe package, builds CRM automations.
- **Manager (makes it recur):** Ensures the calendar activities happen on
  schedule, monitors CRM health, checks that sequences are running.

**The missing role in most small businesses is the Manager.** The entrepreneur
had the idea. The specialist set up the system. But without a Manager, newsletters
don't go out, shock and awe packages don't get sent, and the calendar exists only
as a document.

For sole operators: outsource or part-time hire for the Manager role first.
Even 5–10 hours per week of admin support to run the marketing calendar is
more valuable than another tool or campaign.

**WHY:** The marketing system fails not because the ideas are wrong but because
the operational execution is inconsistent. A marketing infrastructure that runs
80% of the time beats a perfect strategy that runs 10% of the time. The Manager
role is the operational backbone that turns strategy into revenue.

### Step 8: Write nurture-sequence.md

Save the complete output as `nurture-sequence.md` in the user's working directory.
The document must contain all six components: CRM choice, email sequence outline,
shock and awe package plan, marketing calendar table, event-trigger map, and
role assignments.

**WHY:** Square #5 of the 1-Page Marketing Plan must be documented. Without a
written system, the nurture infrastructure lives only in intention — and intention
does not follow up with leads.

---

## Inputs

| Input | Format | Required |
|-------|--------|----------|
| Target market profile | text / .md | Yes |
| Current lead sources and volume | text | Yes |
| CRM or email platform in use | tool name | Recommended |
| Content assets available (guides, videos, testimonials) | list | Recommended |
| Customer lifetime value | dollar amount | Recommended (for shock and awe decision) |
| Existing email sequences or campaigns | text / file | Optional |

---

## Outputs

Primary output: `nurture-sequence.md`

```markdown
# Lead Nurture Sequence: [Business Name]

## CRM Choice
**Platform:** [Name]
**Reason:** [One sentence on why this fits]

## Email Sequence (5-part, 30 days)
| Day | Subject | Purpose |
|-----|---------|---------|
| 1   | [Subject line] | Welcome + deliver ethical bribe |
| 3   | [Subject line] | Value / insight |
| 7   | [Subject line] | Case study / story |
| 14  | [Subject line] | Address top objection |
| 30  | [Subject line] | Soft offer |

## Shock and Awe Package
**Trigger:** Inbound sales inquiry
**Delivery:** FedEx / tracked courier
**Contents:**
- [ ] [Item 1 — e.g., book]
- [ ] [Item 2 — e.g., testimonial compilation]
- [ ] [Item 3 — e.g., handwritten note]
- [ ] [Item 4 — e.g., sample or gift card]
**Estimated cost per package:** $[X]
**Customer lifetime value:** $[Y] → package is [X/Y]% of acquisition cost

## Marketing Calendar
| Cadence | Activity | Owner |
|---------|----------|-------|
| Daily | [social monitoring activity] | [role] |
| Weekly | [blog + email broadcast] | [role] |
| Monthly | [printed newsletter / postcard] | [role] |
| Quarterly | [reactivation letter] | [role] |
| Annually | [customer gift / thank-you] | [role] |

## Event-Trigger Map
| Trigger | Response | Automated? |
|---------|----------|-----------|
| Networking prospect met | CRM entry + newsletter | Manual → CRM |
| Inbound inquiry | Shock and awe package | Manual → fulfilled |
| New email subscriber | 5-part email series | Automated |
| Complaint resolved | Handwritten note + voucher | Manual |
| [Business-specific trigger] | [Response] | [Y/N] |

## Role Assignments
- **Entrepreneur:** [Name / Owner] — designs campaigns, approves calendar
- **Specialist:** [Name / contractor] — writes copy, builds CRM flows
- **Manager:** [Name / VA / outsourced] — runs the calendar, checks sequences

_Square #5 of the 1-Page Marketing Plan canvas: filled._
```

---

## Key Principles

**1. The money is in the follow-up.**
50% of salespeople give up after one contact. 65% give up after two. 79.8% give
up after three. At Contact #4, you are already ahead of 89.8% of all competitors.
At Contact #9, a prospect who is finally ready to buy will call you — you are the
only person who has stayed in touch. Persistence is not pestering; it is
positioning.

**2. Value first, offer rarely.**
90% of nurture communications should deliver genuine value: insights, tips, case
studies, industry news. Only 10% should contain an offer. Inverted ratios — where
every email is a pitch — train the prospect to ignore your messages. A prospect
who looks forward to your emails will buy from you when they are ready.

**3. Physical mail cuts through digital noise.**
Email inboxes are crowded. A hand-addressed envelope or a FedEx package on a desk
commands attention that a PDF cannot. Joe Girard built the world's greatest sales
record with monthly greeting cards — hand-addressed, varying envelope colors,
always with the message "I like you." By the end of his career, two-thirds of
his 13,001 car sales were to repeat customers who had to set appointments to buy
from him. Physical mail, used deliberately, is a premium channel.

**4. The Manager role is critical — and usually missing.**
Most small business marketing systems fail not from bad strategy but from missing
operational execution. The entrepreneur had the idea. The specialist built the
tools. But without someone whose job is to make it recur — send the newsletter,
pack the shock and awe box, enter the business card, check the CRM — the system
sits idle. Identify and fill this role before building a more sophisticated system.

**5. Event-triggered automations are the highest-leverage moves.**
Scheduled calendar activities are important for ongoing presence. But triggered
automations fire at the moments of highest prospect attention — an inbound
inquiry, a resolved complaint, a first purchase. These moments are the most
persuasive because the prospect is already engaged. Automating the right response
to these triggers is worth more than any amount of cold outreach.

---

## Examples

### Example 1: Joe Girard — Monthly Greeting Cards (verbatim from source)

Joe Girard is listed in the *Guinness World Records* as "the world's greatest
salesman." Between 1963 and 1978, he sold 13,001 cars at a Chevrolet dealership —
more retail big-ticket items, one at a time, than any other salesperson in
recorded history. His stats: 13,001 cars total, 18 on his best day, 174 in his
best month, 1,425 in his best year. He sold more cars by himself than 95% of all
dealerships in North America.

His core nurture system: a personalized greeting card mailed every month to his
entire customer list. In January, it was a Happy New Year card. In February, a
Valentine's Day card. The message inside was always the same: "I like you." He
would vary the size and color of the envelope — this was critical to bypassing
the postal equivalent of spam filters, where people stand over the trash can and
discard anything that looks like an ad or junk mail. By the end of his career, he
was sending 13,000 cards per month and needed to hire an assistant.

By the time he was a decade into his career, almost two-thirds of his sales were
to repeat customers. It got to the point where customers had to set appointments
in advance to come in and buy from him — contrast that with other car salespeople
who stood around waiting and hoping for walk-in traffic.

**The lesson:** Consistent, personal, low-pitch contact over years builds a
pipeline that sells itself. The content of the card was not about cars. It was
about the relationship.

---

### Example 2: Shock and Awe Package — Professional Services Firm

**Trigger:** A prospective client emails asking about accounting services.

**Standard response (same-same):** Email back a PDF brochure of services and a
link to the website.

**Shock and awe response:**
- FedEx package arrives within 48 hours containing:
  - A copy of a relevant book on small business finance (positions firm as expert)
  - A printed compilation of 5 client testimonials (one from a similar industry)
  - A one-page media clipping from a local business publication featuring the firm
  - A handwritten note: "Thank you for reaching out. Inside is everything you
    need to know about how we work and why our clients stay with us for years."
  - A gift card to a local coffee shop: "Have a coffee on us while you read."

**Result:** The prospect is not evaluating three accounting firms from their inbox.
They have a physical box on their desk from one firm that went out of its way to
impress them before a single dollar was spent. At a customer lifetime value of
$8,000/year, a $60 package is 0.75% of first-year value.

---

### Example 3: Local Physiotherapy Clinic — Full Nurture System

**CRM:** ActiveCampaign (patient records + email automation)

**Email sequence (new inquiry, 30 days):**
- Day 1: Welcome + link to free stretching guide (ethical bribe delivered)
- Day 3: "The most common mistake people make after a sports injury"
- Day 7: Patient story — back pain resolved in 6 sessions
- Day 14: FAQ — "How many sessions will I need? What should I expect?"
- Day 30: "Book your first appointment — this week only, we have 3 openings"

**Shock and awe package:** Triggered for inbound referrals from GPs or surgeons
(high-value patients). Contents: clinic brochure, testimonials, a printed copy
of a patient recovery guide, and a handwritten note from the lead physiotherapist.

**Marketing calendar:**
- Daily: monitor Google reviews and Facebook mentions; reply within 2 hours
- Weekly: publish one injury-prevention tip on the blog; send to email list
- Monthly: mail a postcard to lapsed patients ("It's been 6 months — how is your
  recovery going?")
- Quarterly: send a reactivation letter to patients not seen in 12+ months
- Annually: send a small gift basket to top 20 referring GPs

**Event triggers:**
- New patient books first appointment → welcome email sequence starts
- Patient cancels appointment → reschedule email + 15% discount on next booking
- 5-star Google review posted → thank-you email + referral ask

**Roles:** Owner (Entrepreneur) designs campaigns. Admin coordinator (Manager)
runs the calendar. Outsourced copywriter (Specialist) writes email content.

---

## References

- Research summary: `.meta/research/lead-nurture-sequence-design.md`
- Hunter report: `.meta/research/hunter-report.md` (sk-07 entry)
- Dependency: `skills/lead-capture-ethical-bribe-design/SKILL.md`
- Previous skill in sequence: `lead-capture-ethical-bribe-design` (square #4)
- Canvas meta-skill: `marketing-plan-canvas` (square #5 slot)
- Book profile: `.meta/book-profile.json`

## License

This skill is licensed under [CC-BY-SA-4.0](https://creativecommons.org/licenses/by-sa/4.0/).
Source: [BookForge](https://github.com/bookforge-ai/bookforge-skills) — The 1-Page Marketing Plan by Allan Dib.

## Related BookForge Skills

Install related skills from ClawhHub:
- `clawhub install bookforge-lead-capture-ethical-bribe-design`

Or install the full book set from GitHub: [bookforge-skills](https://github.com/bookforge-ai/bookforge-skills)

---
name: lead-capture-ethical-bribe-design
description: "Use this skill to design a lead-capture ad strategy using an
  ethical bribe — a high-value free offer that self-selects high-probability
  prospects. Triggers when a user wants to capture leads, create a lead magnet,
  design a lead-generation ad, stop selling directly from ads, build a free
  report or free guide offer, set up gated content or a content upgrade, design
  a CRM capture plan, or fill square #4 of the 1-Page Marketing Plan. Also
  activates for 'ethical bribe', 'hunting vs farming', 'I keep running ads but
  nobody buys', 'how do I get people into my database', '3% buyer', 'addressable
  market', '1,233%', 'lead-gen not direct-sell', 'landing page offer', or
  'capture leads from advertising'."
version: 1.0.0
homepage: https://github.com/bookforge-ai/bookforge-skills/tree/main/books/the-1-page-marketing-plan/skills/lead-capture-ethical-bribe-design
metadata: {"openclaw":{"emoji":"📚","homepage":"https://github.com/bookforge-ai/bookforge-skills"}}
status: published
source-books:
  - id: the-1-page-marketing-plan
    title: "The 1-Page Marketing Plan"
    authors: ["Allan Dib"]
    chapters: [4]
tags:
  - marketing
  - lead-generation
  - lead-magnet
  - advertising
  - small-business
depends-on:
  - target-market-selection-pvp-index
execution:
  tier: 1
  mode: full
  inputs:
    - type: document
      description: >
        Business description, the target market (from target-market-selection-pvp-index
        or provided directly), and any existing ad copy or marketing materials.
  tools-required: [Read, Write]
  tools-optional: []
  mcps-required: []
  environment: >
    Document set — business description, target market profile, and optionally
    existing ads. No code execution required.
discovery:
  goal: >
    Help the user shift from direct-sell advertising (reaching 3% of their
    addressable market) to lead-generation advertising (reaching 40%), by
    designing an ethical bribe concept, a lead-gen ad headline, a minimal capture
    form, and a CRM landing plan. Produces a concrete lead-capture.md document.
  tasks:
    - "Confirm target market (invoke dependency or gather directly)"
    - "Identify the target market's top pain point or burning question"
    - "Brainstorm ethical bribe concepts using the Seven Mistakes formula"
    - "Select the strongest concept and write the lead-gen ad headline"
    - "Design a minimal capture form (name + email or name + address only)"
    - "Plan the CRM landing: what happens after the form is submitted"
    - "Compute the addressable market shift: 3% → 40% = 1,233% improvement"
    - "Write lead-capture.md with all outputs"
  audience:
    roles:
      - small-business-owner
      - solopreneur
      - entrepreneur
      - freelancer
      - startup-founder
    experience: beginner-to-intermediate
  when_to_use:
    triggers:
      - "User is running ads that sell directly and getting poor results"
      - "User wants to build a prospect database for follow-up"
      - "User needs a lead magnet or ethical bribe concept"
      - "User is filling square #4 of the 1-Page Marketing Plan"
      - "User wants to shift from hunting to farming"
    prerequisites:
      - "Target market must be known (use target-market-selection-pvp-index first)"
    not_for:
      - "Enterprise demand generation with marketing automation teams"
      - "Businesses that already have a functioning lead-gen funnel and are happy with it"
  environment:
    codebase_required: false
    codebase_helpful: false
    works_offline: true
  quality:
    scores:
      with_skill: 92.9
      baseline: 21.4
      delta: 71.4
    tested_at: "2026-04-09"
    eval_count: 1
    assertion_count: 14
    iterations_needed: 1
---

# Lead Capture with Ethical Bribe Design

A structured process for small business owners to stop selling from ads and
start building a database of high-probability prospects. Uses the 3/7/30/30/30
market readiness model to show why lead-gen ads reach 40% of the addressable
market vs 3% for direct-sell ads — a 1,233% improvement in advertising
effectiveness. Produces an ethical bribe concept, lead-gen ad copy, a minimal
capture form, and a CRM landing plan. Required content for square #4 of the
1-Page Marketing Plan canvas.

---

## When to Use

Use this skill after target market selection and before designing a nurture
sequence. It sits in the middle of the marketing funnel: the job of square #4
is to capture interested prospects who are not yet ready to buy, so they can be
nurtured until they are.

Also use it when:
- Ads are running but generating few or no responses
- The business has no prospect database — every month starts from zero
- Marketing feels like feast-or-famine ("hunting" mode)
- The business is about to launch a new advertising campaign

Do NOT use this skill as a substitute for target market selection. If the target
market is not yet defined, invoke `target-market-selection-pvp-index` first, or
ask the user to describe their target market directly before proceeding. Every
step of this skill requires knowing exactly who the ethical bribe is for.

---

## Context & Input Gathering

### Required (must have before proceeding)
- Target market: who the business serves (segment name + key pain points)
- Business description: what the business does, what it sells

### Observable / inferrable
- Top prospect pain points (often apparent from the business description)
- Appropriate ethical bribe format (depends on industry and delivery preference)

### Defaults (apply if not provided)
- If the target market is unknown: invoke `target-market-selection-pvp-index`
  OR ask: "Who is your ideal customer, and what is the biggest problem or
  question they have before they buy from you?"
- If the user has no CRM: default recommendation is to start with a simple
  email service (Mailchimp, ConvertKit, or equivalent) — a spreadsheet is
  better than nothing but is not a system

### Sufficiency check
You have enough to proceed when you can name: (1) the target market segment,
(2) their top pain point or burning question before they buy, and (3) the
business's area of expertise. Everything else is generated in the steps below.

---

## Process

### Step 1: Confirm target market and map the top pain point

Review or gather the target market profile. Then identify the single most
pressing pain point, fear, or question a prospect in this market has BEFORE
they make a purchase decision.

**WHY:** The ethical bribe must be irresistibly relevant to the prospect's
current situation, not a generic freebie. The pain point or pre-purchase
question is the hook. A prospect who is about to make a stressful or high-stakes
decision (e.g., hiring a contractor, choosing a photographer, selecting
software) has specific fears and information gaps — the ethical bribe addresses
exactly those.

Ask: "What is the one thing your ideal prospect is most worried about, confused
about, or afraid of making a mistake on — before they buy?"

### Step 2: Apply the 3/7/30/30/30 market readiness model

Explain the model to the user and use it to reframe their advertising goal.

At any given time in a target market:

| Segment | Size | Readiness |
|---------|------|-----------|
| Ready to buy NOW | 3% | Will respond to direct-sell ads |
| Open to buying (not actively searching) | 7% | Will respond to lead-gen ads |
| Interested but not right now | 30% | Will respond to lead-gen ads |
| Not interested | 30% | Cannot be reached profitably |
| Would not buy even if free | 30% | Irrelevant — never target |

Direct-sell ads (ads that say "buy now", "call today", "get a quote") compete
for the 3% only. Lead-gen ads capture the 3% + 7% + 30% = **40%**.

40% ÷ 3% = 13.33x = **1,233% improvement** in advertising effectiveness.

**WHY:** Most business owners have been told that advertising is about making
sales. It is not. Advertising is about finding interested people and getting
them to raise their hand. The 3/7/30/30/30 model makes this concrete and
calculable. Seeing the math — not 10% better, not 50% better, but 1,233% better
— is what creates the mindset shift from hunting to farming.

### Step 3: Brainstorm ethical bribe concepts

Generate 3–5 ethical bribe concepts for the target market. An ethical bribe is
free content of genuine value to the prospect that causes them to self-identify
as a high-probability buyer.

**Formats:** free report, checklist, video series, guide, quiz result, tool,
template, mini-course, or DVD/physical item (where appropriate).

**The Seven Mistakes formula:** The most proven headline structure is:

> "Free [Format] Reveals the [N] Costly Mistakes to Avoid When [Doing the
> thing the prospect is about to do]"

This works because:
1. "Mistakes to avoid" triggers loss-aversion (more powerful than gain-framing)
2. "When [doing the thing]" qualifies the prospect — only people about to do
   this will care
3. The word "free" removes the friction of a purchase decision
4. The number makes it feel specific and bounded (not an overwhelming read)

Generate at least three headline candidates using this formula or close
variations. Evaluate each on: (a) relevance to the target pain point, (b)
specificity of the qualifying phrase, (c) how strongly it self-selects the
right prospect.

**WHY:** The bribe must be high-value to the *prospect*, not impressive to the
*business owner*. A 50-page whitepaper on company history is not a bribe. A
"7 Questions to Ask Any Plumber Before You Let Them Touch Your Pipes" checklist
is a bribe — it solves a real problem the prospect has right now.

### Step 4: Select the strongest concept and write the lead-gen ad

Choose the ethical bribe concept with the highest prospect relevance and
self-selection power. Then write a complete lead-gen ad unit:

**Headline:** The Seven Mistakes formula or equivalent (from Step 3).

**Body copy (3–5 sentences):**
- Name the prospect's pain or fear in their own language
- Introduce the ethical bribe as the solution to that specific worry
- State exactly what they will learn or receive
- Call to action: "Request your free [format] at [URL/phone/address]"

**WHY:** The ad has one job: generate a lead, not make a sale. Every word of
the ad should move the reader toward requesting the bribe — not toward
evaluating the business's credentials, pricing, or features. Those belong in
the nurture sequence, not the ad.

**Anti-pattern to avoid:** Do not include the price, a special offer, or a
"call us now" direct-sell call to action. That collapses back to targeting only
the 3%.

### Step 5: Design the minimal capture form

Specify the fields on the lead capture form. The rule is: capture the minimum
data needed to follow up, and nothing more.

**Default minimum:**
- Name (first name only is fine)
- Email address (for digital follow-up)

**Alternative minimum (for physical delivery):**
- Name + mailing address (if sending a physical bribe like a book or DVD)

**Optional (add only if operationally required):**
- Phone number (adds friction; only include if the follow-up requires calling)
- Business name / company (B2B only, if segmentation requires it)

**WHY:** Every additional field reduces response rate. The goal at this stage is
volume of interested prospects in the database, not completeness of their
profile. Profile can be enriched through the nurture sequence (see
`lead-nurture-sequence-design`). A prospect who gives their name and email is
in your database; a prospect who abandons a long form is lost forever.

### Step 6: Plan the CRM landing

Map what happens after the form is submitted:

1. **Confirmation page:** Thank the prospect. Tell them when and how to expect
   the ethical bribe (e.g., "Check your email — your free report arrives in
   the next 5 minutes").
2. **Bribe delivery:** Automated email sends the ethical bribe immediately
   (PDF attachment, download link, or shipping confirmation for physical items).
3. **CRM entry:** Prospect's name + email is logged to the CRM with a source
   tag (which campaign/ad generated this lead) and a timestamp.
4. **Nurture trigger:** The CRM starts the prospect on a follow-up sequence
   (to be designed with `lead-nurture-sequence-design`).

If no CRM exists yet: capture name + email to a spreadsheet or basic email
service as a starting point. The database is the marketing asset — even a
simple list is better than no list.

**WHY:** The CRM is the goldmine. The ethical bribe is the mining tool. Without
a system that captures and stores lead information, every ad campaign is
one-shot: prospects who respond but don't buy today are lost. With a database,
the prospect who was not ready in January might convert in June — and you still
have them.

### Step 7: Compute the addressable market shift

For the user's specific context, make the 1,233% improvement tangible:

```
Current ad budget: $[X]
Prospects reached by ad: [N]
Direct-sell addressable market: 3% of N = [0.03 × N] high-probability prospects
Lead-gen addressable market:   40% of N = [0.40 × N] high-probability prospects
Marketing spend per valid prospect (direct-sell): $[X] ÷ [0.03N] = $[Y1]
Marketing spend per valid prospect (lead-gen):    $[X] ÷ [0.40N] = $[Y2]
Improvement: [Y1] ÷ [Y2] = 13.33x = 1,233%
```

**WHY:** Abstract percentages are easy to dismiss. Plugging in real numbers
makes the opportunity cost of direct-sell advertising undeniable. A business
spending $2,000/month on direct-sell ads that reach 2,000 people is spending
$33.33 per valid prospect (60 prospects). The same $2,000 in lead-gen reaches
800 valid prospects at $2.50 each — with 13x more firepower per prospect.

### Step 8: Write lead-capture.md

Save the full output as `lead-capture.md` in the user's working directory.

**WHY:** Square #4 of the 1-Page Marketing Plan must be documented. Without a
written ethical bribe concept and ad copy, execution reverts to "we'll do it
later" — the most common reason small businesses never build a prospect database.

---

## Inputs

| Input | Format | Required |
|-------|--------|----------|
| Target market profile | text / .md (from PVP skill) | Yes |
| Business description | text | Yes |
| Existing ad copy or campaigns | text / image | Optional |
| Ad budget (for Step 7 calculation) | dollar amount | Recommended |
| CRM or email platform in use | tool name | Recommended |

---

## Outputs

Primary output: `lead-capture.md`

```markdown
# Lead Capture Plan: [Business Name]

## Target Market
[One paragraph: who they are, their top pain point]

## Ethical Bribe
**Concept:** [Name of the bribe — e.g., "Free Guide: 7 Costly Mistakes..."]
**Format:** [Report / Checklist / Video series / etc.]
**Delivery:** [Email PDF / Physical mail / Download link]

## Lead-Gen Ad Copy
**Headline:** [Full headline using Seven Mistakes formula or equivalent]
**Body:** [3–5 sentence ad copy]
**Call to action:** [Exact CTA text and response mechanism]

## Capture Form Fields
- [ ] First name
- [ ] Email address
- [ ] [Optional: phone / company if justified]

## CRM Landing Plan
1. Confirmation page message: [text]
2. Bribe delivery: [automated email / physical ship / instant download]
3. CRM tag: [campaign source + date]
4. Nurture trigger: [name of sequence or "to be designed"]

## Market Readiness Calculation
| Segment | % | Count (of [N] reached) |
|---------|---|------------------------|
| Ready to buy NOW | 3% | [0.03N] |
| Open to buying | 7% | [0.07N] |
| Interested, not now | 30% | [0.30N] |
| **Lead-gen total** | **40%** | **[0.40N]** |
| Direct-sell total | 3% | [0.03N] |
| **Improvement** | | **1,233%** |

_Square #4 of the 1-Page Marketing Plan canvas: filled._
```

---

## Key Principles

**1. The goal of your ad is to generate a lead, not make a sale.**
Conflating these two goals is the most costly mistake in small business
advertising. The ad exists to cause interested people to raise their hand.
The sale happens later, through the nurture sequence, when the prospect is
ready. Trying to close from the ad collapses the 40% addressable market back
to 3%.

**2. The ethical bribe must be genuinely valuable, not a gimmick.**
A weak bribe (a discount coupon, a brochure dressed as a "free report") does
not self-select high-probability prospects — it attracts bargain hunters. The
bribe must deliver real information the prospect wants: what to look out for,
what mistakes to avoid, what questions to ask. Its value positions the business
as an educator and expert before any sales conversation begins.

**3. Minimal capture fields preserve response rate.**
The database is built one name at a time. Every extra form field loses
prospects. Capture name and email. Build the rest of the profile through
follow-up. A long form is a conversion killer.

**4. Authority positioning over salesperson positioning.**
When a business educates its prospects through free content, it is no longer
questioned — it is trusted. The prospect who reads a free guide arrives at the
sales conversation already believing the business is the expert. A salesperson
has to earn trust through the pitch; an educator has already earned it before
the pitch begins.

**5. The database is the asset, not the ad.**
An ad campaign is temporary. The prospect database is permanent. Every name
captured today is a future sale — not just the next sale. Businesses that
treat the database as their primary asset run marketing infrastructure; everyone
else runs "random acts of marketing" that cost more than they produce.

---

## Examples

### Example 1: Wedding Photographer (canonical book example)

**Target market:** Engaged couples planning a wedding.

**Top pain point:** Fear of making an expensive, irreversible mistake when
choosing a photographer for the most important day of their life.

**Ethical bribe:** A free DVD showcasing the photographer's work AND educating
couples on what to look for.

**Headline (verbatim from the book):**
> "Free DVD Reveals the Seven Costly Mistakes to Avoid When Choosing a
> Photographer for Your Wedding Day."

**Why it works:** Anyone requesting this DVD has self-identified as someone
planning a wedding and evaluating photographers. The list of requesters is
a database of high-probability prospects — every single one is a potential
customer, unlike a general ad audience where 97% will never buy.

**Capture form:** Name + mailing address (for DVD delivery).

**CRM landing:** DVD ships within 48 hours. Lead enters CRM tagged
"wedding-dvd-[campaign]". Follow-up sequence starts day of delivery.

---

### Example 2: Independent IT Services Firm (B2B service)

**Target market:** Small business owners (10–50 employees) who manage their
own IT and are worried about data security and downtime.

**Top pain point:** Fear of a ransomware attack or data loss that could shut
down the business, combined with uncertainty about whether their current setup
is actually protected.

**Ethical bribe:** A free security checklist that lets business owners audit
their own IT setup in 15 minutes.

**Headline:**
> "Free Checklist: The 9 Critical Security Gaps Most Small Businesses
> Have — and How to Spot Them Before Hackers Do."

**Ad body:**
> "Most small business owners assume their IT is 'good enough' — until
> the day a ransomware attack proves otherwise. This free 9-point
> checklist takes 15 minutes and shows you exactly where your biggest
> vulnerabilities are. No tech jargon. No sales pitch. Just a clear,
> honest audit you can do yourself. Download it free at [URL]."

**Capture form:** First name + work email.

**CRM landing:** Instant PDF download. Lead tagged "security-checklist-[ad]".
Follow-up email Day 1: "Did you find any gaps?" opens the conversation without
selling.

**Market shift:** If the ad reaches 500 local businesses: direct-sell would
reach 15 ready buyers; lead-gen reaches 200 interested prospects (1,233%
improvement). At $500/month ad spend: $33/prospect vs $2.50/prospect.

---

### Example 3: Local Kitchen Renovation Showroom (retail/trades)

**Target market:** Homeowners planning a kitchen renovation in the next 6–18
months.

**Top pain point:** Anxiety about cost blow-outs, contractor reliability, and
making design decisions they will regret in an expensive project.

**Ethical bribe:** A free planning guide that walks homeowners through the
renovation process before they speak to any contractor.

**Headline:**
> "Free Guide: 6 Mistakes Homeowners Make That Turn a $20,000 Kitchen
> Renovation Into a $35,000 Nightmare — and How to Avoid Every One."

**Ad body:**
> "Planning a kitchen renovation? Before you call a single contractor,
> read this. Our free 12-page guide reveals the six budget-blowout
> mistakes we see homeowners make every week — and the exact questions
> to ask any contractor before you sign anything. Request your free
> copy at [URL] or call [number]."

**Capture form:** First name + email (or first name + address if mailed).

**CRM landing:** PDF emailed immediately. Lead tagged "kitchen-guide-[month]".
Follow-up sequence: 3 educational emails over 14 days, then invitation to visit
the showroom.

---

## References

- Research summary: `.meta/research/lead-capture-ethical-bribe-design.md`
- Hunter report: `.meta/research/hunter-report.md` (sk-06 entry)
- Dependency: `skills/target-market-selection-pvp-index/SKILL.md`
- Next skill in sequence: `lead-nurture-sequence-design` (square #5)
- Canvas meta-skill: `marketing-plan-canvas` (square #4 slot)
- Book profile: `.meta/book-profile.json`

## License

This skill is licensed under [CC-BY-SA-4.0](https://creativecommons.org/licenses/by-sa/4.0/).
Source: [BookForge](https://github.com/bookforge-ai/bookforge-skills) — The 1-Page Marketing Plan by Allan Dib.

## Related BookForge Skills

Install related skills from ClawhHub:
- `clawhub install bookforge-target-market-selection-pvp-index`

Or install the full book set from GitHub: [bookforge-skills](https://github.com/bookforge-ai/bookforge-skills)

---
name: irresistible-offer-builder
description: "Use this skill to construct a complete, irresistible offer using
  an 8-element checklist (Value, Language, Reason Why, Value Stack, Upsells,
  Payment Plan, Outrageous Guarantee, Scarcity). Triggers when a user wants to
  build an offer, design an irresistible offer, stop lazy discounting, fix a
  '10% off' offer, create a value stack, add an outrageous guarantee, design
  scarcity marketing, build a payment plan for a high-ticket product, improve
  offer conversion, write a compelling offer, design upsells, answer
  'why-should-I-buy', or fill square #2 of the 1-Page Marketing Plan. Also
  activates for 'my offers aren't converting', 'I just discount to get
  clients', 'how do I make my offer stand out', 'I keep competing on price',
  or 'what should I include in my offer'."
version: 1.0.0
homepage: https://github.com/bookforge-ai/bookforge-skills/tree/main/books/the-1-page-marketing-plan/skills/irresistible-offer-builder
metadata: {"openclaw":{"emoji":"📚","homepage":"https://github.com/bookforge-ai/bookforge-skills"}}
status: published
source-books:
  - id: the-1-page-marketing-plan
    title: "The 1-Page Marketing Plan"
    authors: ["Allan Dib"]
    chapters: [2]
tags:
  - marketing
  - offer-design
  - pricing
  - conversion
  - small-business
depends-on:
  - target-market-selection-pvp-index
execution:
  tier: 1
  mode: full
  inputs:
    - type: document
      description: >
        Business description including the specific product or service to build
        the offer around, the target market (ideally from target-market-selection-pvp-index),
        and any existing offer or pricing structure to improve.
  tools-required: [Read, Write]
  tools-optional: []
  mcps-required: []
  environment: >
    Document set — business description and offer inputs in markdown.
    No code execution required.
discovery:
  goal: >
    Help small business owners construct a complete, differentiated offer document
    covering all 8 required elements — escaping the lazy discounting trap and
    producing an offer.md ready to use in marketing campaigns.
  tasks:
    - "Run two diagnostic questions to identify the right product/service"
    - "Define the A-to-B transformation (Value)"
    - "Collect and apply target market jargon (Language)"
    - "Draft reason-why justification"
    - "Design value stack of bonuses"
    - "Identify upsell opportunity"
    - "Structure payment plan if high-ticket"
    - "Draft outrageous guarantee statement"
    - "Design credible scarcity element"
    - "Audit against 10% off anti-pattern"
    - "Write offer.md with all 8 elements filled"
  audience:
    roles:
      - small-business-owner
      - solopreneur
      - entrepreneur
      - freelancer
      - startup-founder
      - consultant
    experience: beginner-to-intermediate
  when_to_use:
    triggers:
      - "User wants to build or redesign a marketing offer"
      - "User is relying on price discounts to close sales"
      - "User's current offer blends in with competitors"
      - "User is filling square #2 of the 1-Page Marketing Plan"
      - "User asks how to increase conversion without reducing price"
    prerequisites:
      - "Target market is defined (run target-market-selection-pvp-index first)"
    not_for:
      - "Enterprise product pricing strategy with large teams (use dedicated pricing frameworks)"
      - "Businesses without a defined target market — run target-market-selection-pvp-index first"
  environment:
    codebase_required: false
    codebase_helpful: false
    works_offline: true
  quality:
    scores:
      with_skill: 100
      baseline: 7.1
      delta: 92.9
    tested_at: "2026-04-09"
    eval_count: 1
    assertion_count: 14
    iterations_needed: 1
---

# Irresistible Offer Builder

A structured process for small business owners to escape the "10% off" commodity trap
and construct a compelling, differentiated offer that reverses buying risk, creates
urgency, and gives prospects an obvious reason to choose you now. Produces a complete
`offer.md` document covering all 8 required elements — ready to use in ads, landing
pages, and sales conversations.

Filling lazy offers with structure is the single highest-leverage improvement most
small businesses can make. Dib: "Taking the lazy, ill-thought-out road of '10% off'
or similar crappy offers is akin to throwing your marketing dollars in the trash."

---

## When to Use

Use this skill AFTER identifying your target market (see `target-market-selection-pvp-index`)
and BEFORE writing advertising copy, building landing pages, or running campaigns.
The offer is what all marketing points to — it must be irresistible before any media spend.

Also use it when:
- A business's current offer is "a discount" or "competitive pricing" — with nothing
  else differentiating it
- Conversion rates are low despite adequate traffic
- Prospects keep asking "how much?" before any other question (a sign of commodity
  positioning)
- Closing requires excessive persuasion or price negotiation

Do NOT use this skill to replace target market selection. If the target market is
undefined, run `target-market-selection-pvp-index` first — offer language and pain
points depend entirely on knowing exactly who you are talking to.

---

## Context & Input Gathering

### Required (must have before proceeding)
- Target market: who the offer is for (segment + pain points)
- Product or service to build the offer around
- Current pricing or price range

### Observable / inferrable
- Industry terminology and jargon (can be elicited in Step 3)
- Competitive landscape (can be derived from user description)

### Defaults (apply if not provided)
- If no target market is defined: prompt "Run target-market-selection-pvp-index first,
  or describe your ideal buyer — their biggest frustration and what they most want"
- If user has multiple products: run the two diagnostic questions to select one
- If pricing is unknown: estimate from context; refine in Step 6 (Payment Plan)

### Sufficiency check
You have enough to proceed when you know: who the buyer is, what pain they are
experiencing now, and which product/service addresses it. Everything else is built
in-session.

---

## Process

### Step 1: Run the two diagnostic questions

Ask the user:

1. "Of all your products and services, which one do you have the most confidence
   delivering? Which problem are you sure you can solve — to the point where you'd
   stake your reputation on it if you only got paid on results?"
2. "Of all your products and services, which one do you most enjoy delivering?"

Select the product or service that satisfies both. If they conflict, default to
Question 1 (confidence > enjoyment when building a first irresistible offer).

**WHY:** The offer must be built around something you can guarantee with conviction.
Doubt in delivery undermines the guarantee and the claim. Your strongest service
also produces the most credible value stack and guarantee.

---

### Step 2: Define the A-to-B transformation (Value)

For the selected product/service, define:
- **Point A:** Where is the customer now? (the pain, the problem, the frustration)
- **Point B:** Where will they be after? (the specific, concrete outcome)
- **The bridge:** What you provide to move them from A to B

Test: Could you guarantee this transformation? If not, sharpen the outcome until
you could stand behind it.

**WHY:** This is the crux of the entire offer. Dib says this is the "most valuable
thing you can do for your customer." Without a clear A-to-B, everything else in the
offer floats — the language, the guarantee, the scarcity all depend on a specific,
believable transformation. Vague transformations ("we help businesses grow") produce
vague offers that get ignored.

---

### Step 3: Collect target market language (Language)

Ask the user: "What words, phrases, or jargon does your target market use to describe
their problem and desired outcome? Think of how they would describe the situation to
a friend — not how you would describe your service."

If the user is unsure, prompt: "What do your customers complain about most? What words
do they use when they first contact you?"

Translate the offer into that language. Strike feature/spec terminology. Replace with
outcome and pain-relief language.

Examples:
- BMX riders: "endos," "sick wheelies," "bunny hops" — not frame geometry specs
- Golf: "hooks," "slices," "handicaps" — not club shaft flexibility ratings
- IT clients: "computer goes down," "can't access files" — not "network infrastructure"

**WHY:** Customers buy based on recognition — "this is exactly my problem." If your
offer uses your internal product language, it signals that you understand your product
but not your buyer. Jargon-matched language creates immediate resonance and makes the
offer feel tailor-made rather than generic.

---

### Step 4: Draft the reason why

Every strong offer must explain why it is available. Without a stated reason, prospects
assume there is a catch and become skeptical.

Draft a single sentence: "We're able to offer [the offer] because [genuine reason]."

Real reasons that work:
- Clearing old inventory / end-of-season stock
- Launching a new service and need case studies/testimonials
- Moving location or warehouse
- Anniversary / milestone
- Capacity available this month (genuine)
- Cost reduction from a new supplier/process passed on to clients

**WHY:** Dib offered a service at half competitor price and customers kept calling to
ask "what's the catch?" A clear reason why disarms skepticism instantly. Do NOT
fabricate reasons — a false reason destroys trust more than having none.

---

### Step 5: Design the value stack (bonuses)

List complementary bonuses that increase the perceived value of the core offer.
Aim for total perceived value of bonuses to exceed the price of the core offer.

Structure:
- Core offer: [main product/service + outcome]
- Bonus 1: [complementary item — name it, value it]
- Bonus 2: [complementary item — name it, value it]
- Bonus 3: [complementary item — name it, value it]
- Total value: $X | Your price today: $Y

**WHY:** Value stacking makes saying no feel irrational — the buyer gets more than
they pay for. Bonuses do not need to be expensive to produce; they need to be
genuinely useful to the target market.

---

### Step 6: Identify the upsell

At the moment of purchase, when the prospect has said yes, offer one complementary
high-margin product or service.

Upsell formula: "Most of our [offer] clients also add [complementary item] because
[specific reason it solves the next obvious problem]."

Examples: the fries with the burger, the extended warranty, the setup service with
the software, the monthly maintenance contract with the installation.

**WHY:** The prospect has already agreed to spend — this is the highest-conversion
moment for an additional offer. A $200 add-on feels small after a $2,000 core
agreement (contrast principle). Upsells increase revenue per transaction without
acquiring a new customer.

---

### Step 7: Structure the payment plan (if high-ticket)

If the offer price is $500 or higher, present an installment option.

Payment plan formula: [N] easy payments of $[installment amount]
- Set installments so that N × installment > full price (covers financing cost)
- Full-pay option: "Or save $X by paying in full today"

Example: $5,000 → "12 easy payments of $497" (total: $5,964). Lump-sum payers
receive a "$964 savings" framing.

**WHY:** People budget monthly. A $5,000 ask feels like a major decision;
$497/month feels like a line item. The same purchase is psychologically smaller
when expressed as an installment. Making lump-sum payment the "savings" option
also incentivizes full-pay without removing the installment path for those who
need it. For genuinely low-ticket items (under $100), payment plans add friction
without benefit — skip this element.

---

### Step 8: Draft the outrageous guarantee

Write a guarantee that totally reverses the risk of doing business with you.
"Satisfaction guaranteed" is the floor — it is weak and ineffective. The guarantee
must be specific and stakes-based.

Guarantee structure:
"If [specific deliverable outcome] does not happen by [specific timeframe], we will
[specific consequence — refund, redo, pay penalty]."

For deep guarantee mechanics and verbatim scripts: see `sales-conversion-trust-system`.

**WHY:** Prospects have been disappointed before. They do not trust claims by default.
An outrageous guarantee signals that you are willing to bear the risk of your own
delivery — which only someone who is confident in their results can honestly offer.
Risk reversal converts fence-sitters by removing the downside. "Satisfaction
guaranteed" fails because it is vague, expected, and easily disputed.

---

### Step 9: Design the scarcity element

Every offer needs a credible reason why the prospect must respond now. Scarcity
must be genuine — never fabricated.

Credible scarcity types:
- **Limited supply:** "Only 3 slots available at this price"
- **Limited time:** "Offer closes [specific date]"
- **Limited resources:** "We can only take 10 new clients this quarter to maintain
  service quality"
- **Running countdown:** Countdown timer on landing page, or deadline in email

**WHY:** People respond more strongly to fear of loss than to prospect of gain.
Without a reason to act now, prospects defer — and deferral is a lost sale. The
key constraint is credibility: scarcity must have its own reason why (genuine
capacity, genuine deadline, genuine stock limit). Fake scarcity is obvious to
prospects and destroys trust. Genuine scarcity creates urgency without manipulation.

---

### Step 10: Audit against the 10% off anti-pattern

Before writing offer.md, check each element against this failure-mode list:

| Anti-pattern | Failure sign | Fix |
|---|---|---|
| Lazy discount | Offer is "X% off" with nothing else | Add 7 remaining elements |
| Competitor copy | Offer is structurally identical to nearest competitor | Differentiate on Value Stack + Guarantee |
| No reason why | No explanation for why the offer is available | Draft Step 4 reason |
| Weak guarantee | "Satisfaction guaranteed" only | Rewrite with specific outcome + timeframe |
| No scarcity | No deadline, no limit, no urgency trigger | Add Step 9 scarcity element |
| Product language | Offer uses specs, features, technical terms | Rewrite in Step 3 jargon |
| Prevention pitch | Offer sells future benefit, not present pain | Reframe around existing pain (Step 2) |

**WHY:** Most offers fail 3–5 of these checks. Finding them here costs nothing;
discovering them after a campaign launch costs budget and momentum.

---

### Step 11: Write offer.md

Compile all 8 elements into the final offer document (see Outputs section for template).

**WHY:** Without a written offer, the message degrades as it passes through ads,
landing pages, and sales scripts. The document is the source of truth that keeps
all channels saying the same thing.

---

## Inputs

| Input | Format | Required |
|---|---|---|
| Target market (segment + pain points) | text / from target-market.md | Yes |
| Product or service to build the offer around | text | Yes |
| Current price or price range | rough estimate | Yes |
| Existing offer or marketing copy (if any) | text | Recommended |
| Known customer objections | text | Recommended |
| Existing testimonials or results | text | Recommended |

---

## Outputs

Primary output: `offer.md`

```markdown
# Offer: [Business Name] — [One-line offer headline]

## Diagnostic

**Product/service selected:** [Name]
**Confidence basis:** [Why you'd stake your reputation on this]
**Point A (current state):** [Specific pain/problem the buyer is in NOW]
**Point B (desired state):** [Specific, concrete outcome after delivery]

---

## The 8-Element Offer

### 1. Value Statement
[One to two sentences: what transformation is being delivered, expressed as
Point A → Point B in concrete terms]

### 2. Offer Headline (in target market's language)
[A headline using the buyer's own jargon — not product terminology]

Supporting copy: [2–3 sentences using emotionally resonant, pain-targeted language]

### 3. Reason Why
[One sentence explaining why this offer is available at this price/terms right now]

### 4. Value Stack
| What you get | Estimated value |
|---|---|
| [Core offer: name the outcome, not the service] | $[value] |
| [Bonus 1 name + what it does for them] | $[value] |
| [Bonus 2 name + what it does for them] | $[value] |
| [Bonus 3 name + what it does for them] | $[value] |
| **Total value** | **$[total]** |

**Your investment today:** $[offer price] (or see payment plan below)

### 5. Upsell
"When you [core offer], most clients also add [upsell name] because [specific
next-pain it solves]. Add it today for $[price]."

### 6. Payment Plan
**Option A:** [N] easy payments of $[installment] — total $[installment × N]
**Option B:** Pay in full today — $[full price] (save $[difference])

_[Skip this section if under $500 total]_

### 7. Outrageous Guarantee
"If [specific outcome] does not [happen / be delivered] by [specific timeframe],
we will [specific consequence: full refund / redo at no charge / pay penalty of $X].
No questions asked."

_For deeper guarantee scripting: see sales-conversion-trust-system_

### 8. Scarcity
[State the limit and the genuine reason for it]

Example: "We are only taking [N] new clients this [month/quarter] because
[genuine capacity reason]. [Deadline or counter if applicable.]"

---

## Offer Summary (one paragraph for ads/sales scripts)

[Combine all 8 elements into a single paragraph that could be read aloud in
30 seconds or used as a Facebook ad body. Should contain: transformation,
headline language, reason why, value stack summary, guarantee, and deadline.]

---

_Square #2 of the 1-Page Marketing Plan canvas: filled._
```

---

## Key Principles

**1. Pain over prevention — always.**
People do not buy prevention; they buy cure. Targeting a prospect already in pain
eliminates price sensitivity, shortens the sales cycle, and produces better
customer satisfaction (the pain was real, the relief is appreciated). Always
identify the pain the prospect is in NOW, not a future risk they might want to
avoid.

**2. Reason why is non-negotiable.**
A strong offer without a stated reason why creates suspicion, not desire. Prospects
think: "If this is such a great deal, what's the catch?" The reason why disarms
this immediately and allows the value to land. Even a simple, obvious reason ("we
have capacity this month") is better than none.

**3. Value stack beats discount every time.**
Discounting trains buyers to wait for sales and positions you as a commodity. A
value stack keeps the price intact and makes the offer feel like a windfall.
Zero-cost bonuses (a checklist, a guide, a consult call) that are genuinely useful
can convert a hesitant prospect without touching the margin.

**4. Contrast principle amplifies upsells.**
A $300 upsell added to a $3,000 core sale feels negligible by comparison. Present
the upsell at the moment of purchase — not before, not after — to maximize the
contrast effect.

**5. Installment pricing reframes affordability.**
The same $5,000 purchase presented as 12 × $497 closes more often because people
think in monthly budgets. Price is rarely the actual objection — cashflow is.
Payment plans solve cashflow without reducing price.

**6. Scarcity must be credible to work.**
Fake countdowns and fabricated limits are transparent and destroy trust. Genuine
scarcity (real capacity limits, real deadlines, real stock) creates urgency without
manipulation. If you have no genuine scarcity, create it by design: limit intake,
set enrollment windows, cap cohort sizes.

**7. "Satisfaction guaranteed" is the floor, not the ceiling.**
Generic guarantees do nothing because everyone offers them. A specific, stakes-based
guarantee differentiates you and converts fence-sitters. If you can't make an
outrageous guarantee, sharpen the offer until you can.

---

## Examples

### Example 1: Business coach selling a $2,000 coaching program

**Diagnostic:**
- Confident delivering: 90-day revenue growth program for freelancers
- Pain (Point A): Freelancer earning $3K–$4K/month, inconsistent, working weekends
- Outcome (Point B): $7K+/month consistent income, with systems to maintain it without weekends

**The 8 elements:**

1. **Value:** "Go from $3K months to $7K months — with a system you can repeat without
   burning yourself out."
2. **Language:** "Stop chasing invoices. Stop discounting to close. Stop working
   weekends on projects you hate."
3. **Reason why:** "I'm opening 5 new spots this quarter while I test a new group
   format — at 40% below what I'll charge when it's proven."
4. **Value stack:** 90-day coaching ($2,000 value) + Pricing Script Library ($400
   value) + Client Qualification Toolkit ($300 value) + Weekly group Q&A calls
   ($600 value) = $3,300 total value. Price: $2,000.
5. **Upsell:** "Most clients add a done-with-you proposal template package at checkout
   — $300, saves 10+ hours on your first new client pitch."
6. **Payment plan:** 3 payments of $740 (total $2,220) or $2,000 paid in full (save $220).
7. **Guarantee:** "If you follow the program and don't hit $6,000 in a single month
   within 90 days, I'll continue coaching you at no charge until you do."
8. **Scarcity:** "5 spots only — I cap intake to protect the group's quality. 3 already
   filled as of this week."

---

### Example 2: Physical product — ergonomic keyboard ($149)

**Diagnostic:**
- Product: Ergonomic split keyboard
- Pain (Point A): Developer with wrist pain after 6+ hour coding sessions
- Outcome (Point B): Code for 8 hours without wrist pain, without changing workflow

**The 8 elements:**

1. **Value:** "Type all day without wrist pain — without relearning how to type."
2. **Language:** "Built for engineers who live in their terminal. Zero adjustment
   period. Plug in. Code."
3. **Reason why:** "We overproduced the Q1 batch and are clearing stock before
   the updated model ships in 60 days."
4. **Value stack:** Keyboard ($149 value) + Carrying case ($29 value) + Wrist
   placement guide PDF ($15 value) + 90-day adjustment support via email ($30 value)
   = $223 value. Price: $149.
5. **Upsell:** "Add the matching wrist rest set for $29 — used by 68% of our buyers
   and the #1 thing people wish they'd ordered at the same time."
6. **Payment plan:** [Not applicable — under $500]
7. **Guarantee:** "If your wrists still hurt after 30 days of daily use, return it
   for a full refund. We pay return shipping. No questions."
8. **Scarcity:** "84 units remaining from this batch. No restock until August."

---

### Example 3: SaaS — project management tool with free trial (fixing lazy offer)

**Before (lazy offer):** "14-day free trial. No credit card required."

**Diagnostic:**
- Pain (Point A): Agency owner whose team misses deadlines and loses client trust
- Outcome (Point B): Every project delivered on time, client trust restored, repeat business

**The 8 elements applied:**

1. **Value:** "Never miss a client deadline again — or give one painful status update
   call explaining why."
2. **Language:** "Built for agencies running 5+ client projects at once, where one
   slipped deadline costs a retainer."
3. **Reason why:** "We're proving this with new agencies this quarter — so you get
   free onboarding support that won't exist at full price."
4. **Value stack:** 14-day trial ($0) + 1:1 setup call with our team ($200 value)
   + Deadline recovery playbook PDF ($75 value) + 30-day price lock guarantee
   ($X value) = included free.
5. **Upsell:** "Most agencies add client reporting automation at signup — $29/month,
   saves 3 hours per client per month on status reports."
6. **Payment plan:** Annual plan: $59/month billed yearly (vs $79/month monthly).
7. **Guarantee:** "If you use the tool for 14 days and your team still misses a
   deadline, we'll give you 3 months free to fix it — or refund your first paid
   month, no questions."
8. **Scarcity:** "Free onboarding call is available for the next 11 signups only —
   our onboarding team has limited capacity this month."

**Result:** Same free trial — now differentiated by outcome, pain-targeted language, a specific guarantee, genuine value-adds, and credible scarcity.

---

## References

- `hunter-report.md` — sk-04 entry: 8-element offer checklist, density 4
- `.meta/research/irresistible-offer-builder.md` — full research summary with
  all source quotes and anti-pattern list
- `sales-conversion-trust-system` — deep mechanics for guarantee scripting
  (verbatim IT and pest control guarantee scripts, risk-reversal psychology)
- `target-market-selection-pvp-index` — prerequisite: defines who the offer is for
- `book-profile.json` — full book metadata

## License

This skill is licensed under [CC-BY-SA-4.0](https://creativecommons.org/licenses/by-sa/4.0/).
Source: [BookForge](https://github.com/bookforge-ai/bookforge-skills) — The 1-Page Marketing Plan by Allan Dib.

## Related BookForge Skills

Install related skills from ClawhHub:
- `clawhub install bookforge-target-market-selection-pvp-index`

Or install the full book set from GitHub: [bookforge-skills](https://github.com/bookforge-ai/bookforge-skills)

---
name: customer-revenue-quality-audit
description: Classify every customer into one of four archetypes — Tribe, Churners, Vampires, or Snow Leopards — score your customer base using Net Promoter Score (NPS), and produce a segmentation report with a concrete fire/grow decision per customer. Use this skill when you want to run a customer audit, audit customer quality, segment customers, identify bad customers, identify draining customers, fire bad customers, clean up polluted revenue, distinguish Tribe from Churners from Vampires from Snow Leopards, classify customers by loyalty, assess NPS score, improve customer loyalty, remove toxic clients, free up team capacity, identify which customers to grow vs which to cut, run a customer quality review, calculate whether you have concentration risk from a single large customer, answer "should I fire this customer", identify your worst customers, stop subsidizing customers who cost more than they generate, or eliminate revenue that is actively making your business sick.
version: 1.0.0
homepage: https://github.com/bookforge-ai/bookforge-skills/tree/main/books/the-1-page-marketing-plan/skills/customer-revenue-quality-audit
metadata: {"openclaw":{"emoji":"📚","homepage":"https://github.com/bookforge-ai/bookforge-skills"}}
status: published
source-books:
  - id: the-1-page-marketing-plan
    title: "The 1-Page Marketing Plan: Get New Customers, Make More Money, and Stand Out from the Crowd"
    authors: ["Allan Dib"]
    chapters: [8]
tags: [customer-management, segmentation, nps, retention, small-business]
depends-on: []
execution:
  tier: 1
  mode: full
  inputs:
    - type: document
      description: "Customer list or description — who your current customers are, how much they pay, how demanding they are, and any known complaints or referrals"
  tools-required: [Read, Write]
  tools-optional: []
  mcps-required: []
  environment: "Document set. Typical files: customer-list.md or customer-list.csv, support-log.md, revenue-report.md. If no files exist, gather information through conversation."
---

# Customer Revenue Quality Audit

## When to Use

You are a business owner who suspects that not all revenue is equal. Typical triggers:

- Certain customers consume far more staff time than others while paying the same
- You feel drained after dealing with a particular client even though they pay on time
- Your team morale is suffering due to one or two demanding accounts
- You acquired customers through heavy discounting and they are now price-sensitive and disloyal
- One customer makes up a disproportionately large share of total revenue and you worry about losing them
- You are running "eat what you kill" mode and accepting any revenue without evaluating its quality
- A customer complains constantly despite receiving the same service others are happy with
- You want to grow high-value relationships but cannot because problem customers consume all available capacity

Before starting, clarify:
- **Do you have a customer list?** Any format works — spreadsheet, CRM export, a written list.
- **Can you estimate revenue per customer?** Even rough brackets (high/medium/low) are sufficient.
- **Do you have a sense of which customers cause the most friction?** This is the key input for archetype classification.

---

## Context & Input Gathering

Ask the user for the following if not already provided:

1. **Customer inventory** — Name or category of each current customer/account.
2. **Revenue contribution** — Approximate monthly or annual revenue per customer.
3. **Demand level** — How many support requests, calls, or escalations does each customer generate?
4. **Payment behavior** — Do they pay on time, or require constant chasing?
5. **Referral behavior** — Have they referred anyone? Do they speak positively about the business?
6. **Team impact** — Do they deal respectfully with staff, or do they escalate to the owner and cause team stress?
7. **Acquisition method** — Were they acquired through normal marketing, or via heavy discounting or overpromising?

---

## Process

### Phase 1: Archetype Classification

**Step 1 — Apply the four-archetype taxonomy to each customer.**

Every customer in your business falls into one of four categories. Map each customer against the definitions below:

| Archetype | Who They Are | Behavioral Signals | Decision |
|-----------|-------------|-------------------|----------|
| **Tribe** | Raving fans who are actively conspiring for your success | Pay on time, refer others unprompted, respect your team, accept your pricing, treat you as a valued partner | GROW |
| **Churners** | Customers you acquired through overhyping, discounts, or poor fit | Price-sensitive, complain frequently, churn out when the novelty wears off, damage your reputation in the market | FIRE |
| **Vampires** | Customers who can afford you but consume disproportionate resources | Demand access to the owner, terrorize your team, consume 5x average support time while paying average rates, drain morale | FIRE FAST |
| **Snow Leopards** | One large customer who makes up an outsized revenue share | Fun to work with, pay well, but represent concentration risk — rare and almost impossible to replicate | MANAGE (do not build strategy around replicating them) |

WHY: Not all revenue dollars are equal. A dollar from a Churner comes with support costs, churn risk, brand damage, and team morale erosion attached to it. A dollar from a Tribe member comes with referrals, goodwill, and compounding lifetime value attached to it. Treating them the same is what Dib calls "polluted revenue" — revenue that makes your business sick rather than healthy.

**Step 2 — Classify every customer.**

For each customer on your list, assign one archetype. Where ambiguous, use this decision tree:

```
Does this customer consume significantly more time/energy than average?
  YES → Are they pleasant and respectful?
          YES → Snow Leopard (large + nice but probably outsized revenue share)
          NO  → Vampire (affordable but consuming; fire fast)
  NO  → Do they refer others and pay on time?
          YES → Tribe (grow these)
          NO  → Churner (acquired via bad fit; fire)
```

Document each classification in the segmentation table (see Outputs).

---

### Phase 2: NPS Scoring

**Step 3 — Send the NPS survey to all active customers.**

The Net Promoter Score is a formal measurement of customer loyalty. Send every customer a two-question survey:

**Primary question (scored 0–10):**
> "How likely is it that you would recommend our company/product/service to a friend or colleague?"

**Follow-up open-ended question:**
> "What is the main reason for your score?"

Record each response. The open-ended reason is critical — it tells you *why* a customer is a Promoter or Detractor, which informs both firing decisions and Tribe growth strategy.

WHY: The NPS survey gives you a defensible, quantified basis for decisions that might otherwise feel personal or arbitrary. It also surfaces issues you did not know existed from customers who were quietly unhappy but never complained — they simply stopped buying.

**Step 4 — Apply NPS thresholds to segment customers.**

| NPS Score | Label | Archetype Mapping |
|-----------|-------|------------------|
| 9–10 | Promoter | Tribe |
| 7–8 | Passive | Potential Snow Leopard zone; monitor |
| 0–6 | Detractor | Churner or Vampire; prioritize for review |

WHY: The thresholds are not arbitrary. Passives (7–8) are satisfied but not loyal — they will switch providers if given a better offer. Only Promoters (9–10) actively recruit new customers for you. Detractors (0–6) are actively damaging your reputation in the market even if they remain paying customers.

**Step 5 — Calculate your NPS.**

```
NPS = % of Promoters − % of Detractors

Range: −100 (all detractors) to +100 (all promoters)
Benchmark: >0 = good; ≥50 = excellent
```

Example: 60% Promoters, 10% Detractors, 30% Passives → NPS = 60 − 10 = **+50** (excellent)

Record this number. It is your revenue quality baseline. Repeat the NPS survey quarterly to track whether firing Detractors and investing in Tribe members is improving the score over time.

---

### Phase 3: Revenue Quality Analysis

**Step 6 — Estimate true profitability per customer.**

For each Detractor/Churner/Vampire customer, calculate a rough adjusted profit:

```
Adjusted Profit = Revenue − (Direct Costs + Time Cost + Morale Cost)

Time Cost = (Hours per month spent on this customer) × (Owner/staff hourly rate)
Morale Cost = qualitative flag (Low / Medium / High)
```

WHY: Most business owners track gross revenue per customer but ignore the time cost of managing them. A customer paying $3,000/month who requires 20 hours of escalations per month at an effective rate of $200/hour is generating $3,000 − $4,000 = **net loss** before other costs. Running an honest profit and loss per customer is often the moment a business owner realizes they have been subsidizing problem customers.

**Step 7 — Flag Snow Leopard concentration risk.**

If any single customer represents more than 20–25% of total revenue, flag them as a Snow Leopard. Document:
- Their revenue as a percentage of total
- What would happen to the business if they left tomorrow
- The dependency risk this creates

WHY: Snow Leopards are wonderful customers — they are often pleasant, pay well, and are fun to work with. The risk is not in the relationship; it is in the strategy error of treating them as a template for growth. They are exquisite and rare, like the actual animal. Chasing more Snow Leopards is a bad growth strategy. Instead, invest in replicating the Tribe.

---

### Phase 4: Decisions and Actions

**Step 8 — Apply firing rules and produce the action plan.**

For each customer, assign one of four actions:

| Action | Applies To | What It Means |
|--------|-----------|--------------|
| **GROW** | Tribe members | Invest in deepening the relationship — referral programs, exclusive access, appreciation outreach, loyalty rewards |
| **MONITOR** | Passives + Snow Leopards | No immediate action; survey again next quarter; for Snow Leopards, diversify revenue exposure |
| **FIRE (scheduled)** | Churners | Give appropriate notice; refer them elsewhere professionally; do not renew |
| **FIRE (urgent)** | Vampires | Move quickly — every day they remain costs morale and capacity; use the firing script below |

WHY: Firing is not abandonment — it is capacity creation. When you remove a Vampire or Churner from your customer base, you free the time and energy that was being consumed to serve your Tribe better. This creates scarcity (you are selective about who you work with), which increases your perceived value in the market and allows you to attract more Tribe-type customers. Firing problem customers and referring them to competitors is the most elegant strategy available: you solve your problem while gifting competitors with it.

**Step 9 — Write the firing conversation script for each customer being fired.**

Use the template below. Adjust the reason for each customer (Churner vs Vampire may require different framing):

```
CUSTOMER FIRING SCRIPT — [Customer Name]
Date: [date]
Method: [email / phone call / in-person meeting]

---

Opening (acknowledge the relationship):
"[Customer name], I want to thank you for your business over the past [time period].
We have genuinely valued the relationship."

Reason (clear but non-accusatory):
"After reviewing our current capacity and focus, we have made the decision to work
exclusively with [describe your ideal customer type — e.g., 'clients in the enterprise
segment / clients who have been with us for more than 12 months / clients whose projects
align with our core specialty'].

This means we will not be able to continue serving [their company] beyond [date]."

Transition (professional referral):
"We want to make sure you are well taken care of. We recommend [competitor or alternative
provider] as a strong fit for your needs. I am happy to make an introduction if that
would be helpful."

Close (firm and warm):
"Thank you again for the time we have worked together. I wish you and your team every
success."

---
Note: Do NOT negotiate. Do NOT offer a discount to stay. A customer won on price will
be lost on price. The decision is final.
```

WHY: A professional, gracious exit protects your reputation. Fired customers talk. A gracious exit turns a Churner into a neutral — they cannot reasonably complain. A botched firing turns a Churner into an active Detractor who damages your brand in the market.

**Step 10 — Build the Tribe growth plan.**

For each Tribe member identified, document at least one growth action:

| Tribe Member | Current Revenue | Referral Potential | Next Growth Action |
|-------------|----------------|-------------------|-------------------|
| [Name] | [amount] | [High/Med] | [e.g., invite to referral program, offer exclusive early access, send handwritten thank-you, provide additional service bundle] |

WHY: Your Tribe members keep the lights on and promote your business for free. They are the customers who will survive you focusing on firing problem customers — and they are the ones who will notice when you have more time and attention to give them.

---

## Inputs

| Input | Required | Notes |
|-------|----------|-------|
| Customer list with names/accounts | Yes | Spreadsheet, CRM export, or written list |
| Revenue per customer (approximate) | Yes | Even rough brackets work |
| Support/demand level per customer | Yes | Can be estimated if not logged |
| NPS survey results | Recommended | Can be run as part of this skill; requires email access to customers |
| Payment history | Optional | Strengthens Churner classification |

---

## Outputs

This skill produces four documents:

1. **customer-segmentation-report.md** — Full table of all customers with archetype, NPS score, estimated true profit, and action (GROW / MONITOR / FIRE scheduled / FIRE urgent)
2. **firing-scripts/[customer-name]-firing-script.md** — One firing script per customer being exited
3. **tribe-growth-plan.md** — Action plan for each Tribe member with specific next steps
4. **nps-baseline.md** — NPS calculation, Promoter/Passive/Detractor breakdown, and quarterly review schedule

---

### Output Template: Customer Segmentation Report

```markdown
# Customer Segmentation Report
Date: [date]
Business: [business name]
Total active customers: [N]
NPS baseline: [score]

## Segmentation Table

| Customer | Revenue/mo | Archetype | NPS Score | True Profit Est. | Action |
|----------|-----------|-----------|-----------|-----------------|--------|
| [Name]   | $[X]      | Tribe     | 10        | Positive        | GROW   |
| [Name]   | $[X]      | Vampire   | 2         | Net loss        | FIRE (urgent) |
| [Name]   | $[X]      | Churner   | 4         | Marginal        | FIRE (scheduled) |
| [Name]   | $[X]      | Snow Leopard | 8     | Positive but concentrated | MONITOR |

## Revenue Quality Summary
- Tribe revenue: $[X] ([X]% of total)
- Polluted revenue (Churners + Vampires): $[X] ([X]% of total)
- Snow Leopard concentration: $[X] ([X]% of total)
- Adjusted NPS after planned exits: [projected score]

## Actions This Quarter
1. Fire [N] Vampires by [date]
2. Fire [N] Churners by [date]
3. Grow [N] Tribe members with [specific action]
4. Re-survey all Passives in [month]
```

---

## Key Principles

**Not all revenue is equal.** A dollar from a Tribe member and a dollar from a Vampire look identical on a revenue report. They are not. The Vampire dollar comes with hidden costs: staff time, morale erosion, owner distraction, and opportunity cost. Track revenue quality, not just revenue volume.

**A customer won on price will be lost on price.** Churners are almost always acquired through discounting or overpromising. They never become Tribe members. The acquisition method predicts the customer type.

**Firing creates scarcity, and scarcity creates value.** When you fire problem customers and become selective about who you work with, the market perceives you as premium. Selectivity is a positioning signal.

**The right customer is always right.** The cliché "the customer is always right" is dangerous when applied to Vampires and Churners. Reframe it: the *right* customer is always right. Problem customers are not always right — they are always wrong for your business.

**Genuine complaints are different from problem customers.** A customer who raises a legitimate service failure is a valuable intelligence asset — they reveal a gap others experienced silently. Do not confuse a customer with a real grievance (fix them) with a Vampire or Churner (exit them).

**Give the squeaky wheel's oil to the quiet good customer.** Low-value, price-sensitive customers complain the most and demand the most attention. High-value Tribe customers are often silent — they are getting good service and do not need to escalate. Firing problem customers redirects time and energy to the customers who actually deserve it.

---

## Examples

### Example 1: Accounting Firm Fires a Vampire

A small accounting firm had one client — a medium-sized retailer — who paid $4,500/month. However, the client's owner called the firm's principal directly at least twice a week, demanded same-day turnarounds, questioned every bill, and once yelled at a junior accountant. NPS score: 3.

True profit calculation:
- Revenue: $4,500/month
- Direct service cost: $1,200/month
- Time cost: 18 hours/month × $220/hr effective rate = $3,960
- **Adjusted profit: $4,500 − $1,200 − $3,960 = −$660/month (net loss)**

The firm sent the firing script (email, 2 weeks notice, competitor referral included). The next month, the principal reclaimed 18 hours, used them to onboard two new Tribe-type clients at $2,500/month each with minimal ongoing support needs. Net monthly improvement: +$5,160.

### Example 2: E-Commerce Brand Identifies Churners Through Acquisition Audit

An e-commerce brand ran a 50%-off launch promotion to acquire their first 200 customers. Twelve months later, 80% of those customers had never reordered. NPS survey results for the promo cohort: average score of 5.2 (Detractor territory). NPS for customers acquired at full price: 8.6 (Passive/Promoter territory).

Decision: Stop discounting. Segment the promo cohort as Churners. Run a reactivation email sequence — customers who respond with a purchase move to Monitor; those who do not are removed from active marketing budget. This freed 30% of the email marketing budget previously wasted on a Churner cohort.

Lesson: The acquisition method predicted the customer quality. Discounting to acquire customers is a Churner factory.

### Example 3: Consulting Practice Manages a Snow Leopard

A strategy consultant had one client — a large financial services firm — that represented 65% of annual revenue. The client was pleasant, paid promptly, and gave positive feedback (NPS: 9). The consultant loved working with them.

The Snow Leopard risk: if that client's internal champion left the firm or the budget was cut, the consultant's practice would face a 65% revenue cliff overnight.

Decision: Do not fire. Do not over-invest in chasing similar clients (they are rare). Instead: cap hours devoted to this account, diversify by actively marketing to new clients, and set a target of reducing this client's revenue share to below 35% within 18 months — without reducing the relationship quality.

Lesson: Snow Leopards are not problems to solve — they are risks to manage. The risk is strategic concentration, not the client themselves.

---

## References

- Dib, A. (2016). *The 1-Page Marketing Plan*, Chapter 8: Increasing Customer Lifetime Value — "Polluted Revenue and the Unequal Dollar" (pp. 242–247) and "Fire Problem Customers" (pp. 243–247).
- Net Promoter Score methodology: Reichheld, F. (2003). "The One Number You Need to Grow." *Harvard Business Review*. The NPS survey question and scoring thresholds (9–10 Promoters, 7–8 Passives, 0–6 Detractors) are the standard implementation.
- The four-archetype taxonomy (Tribe, Churners, Vampires, Snow Leopards) is Dib's original framework from the book. The "polluted revenue" and "unequal dollar" concepts are core to Chapter 8.

## License

This skill is licensed under [CC-BY-SA-4.0](https://creativecommons.org/licenses/by-sa/4.0/).
Source: [BookForge](https://github.com/bookforge-ai/bookforge-skills) — The 1-Page Marketing Plan by Allan Dib.

## Related BookForge Skills

No direct dependencies. Install the full book set from GitHub.

Or install the full book set from GitHub: [bookforge-skills](https://github.com/bookforge-ai/bookforge-skills)

---
name: customer-lifetime-value-growth
description: Grow Customer Lifetime Value (CLV) using five proven levers — raise prices, upsell at point of purchase, move customers up an ascension ladder, increase purchase frequency through reminders and subscriptions, and win back lapsed customers with a reactivation campaign. Use this skill when you want to increase customer lifetime value, grow CLV, grow revenue from existing customers, cross-sell or upsell, raise prices without losing customers, build a subscription model, increase repeat purchases, add recurring revenue, implement an ascension ladder, reactivate customers, win back customers, fill square 8, increase purchase frequency, run a reactivation campaign, grow revenue without new customer acquisition, apply the 20/80 rule to your customer base, or design a return-visit voucher system.
version: 1.0.0
homepage: https://github.com/bookforge-ai/bookforge-skills/tree/main/books/the-1-page-marketing-plan/skills/customer-lifetime-value-growth
metadata: {"openclaw":{"emoji":"📚","homepage":"https://github.com/bookforge-ai/bookforge-skills"}}
status: published
source-books:
  - id: the-1-page-marketing-plan
    title: "The 1-Page Marketing Plan: Get New Customers, Make More Money, and Stand Out from the Crowd"
    authors: ["Allan Dib"]
    chapters: [8]
tags: [marketing, customer-retention, pricing, upsell, clv, small-business]
depends-on: ["customer-experience-systems-design"]
execution:
  tier: 1
  mode: full
  inputs:
    - type: document
      description: "Business description — what the business sells, approximate average transaction value, rough customer count, and whether a customer list or CRM exists"
  tools-required: [Read, Write]
  tools-optional: []
  mcps-required: []
  environment: "Document set. Typical files: business-description.md, customer-list.csv or CRM export, current-pricing.md (to be created or referenced). Output: clv-growth-plan.md."
---

# Customer Lifetime Value Growth

## When to Use

You are a small business owner who wants to grow revenue without spending more on new customer acquisition. Typical triggers:

- Revenue is flat despite a healthy customer count
- You have not raised prices in over a year
- You make one sale per customer and then lose contact
- You have a lapsed customer list sitting unused in a CRM or spreadsheet
- You want to add recurring revenue to a currently transactional business
- You want to know what "square 8" of the 1-Page Marketing Plan looks like in practice

Before starting, clarify:
- **Dependency check:** If the customer experience baseline is unknown — what the current delivery process looks like and whether customers are being wowed or merely served — invoke `customer-experience-systems-design` first, or ask the user to describe their current customer experience level. CLV growth built on a broken experience will produce churners, not loyalists.
- **What does the business sell?** (product, service, or both; average transaction value)
- **Does a customer list exist?** (CRM, spreadsheet, email list — any record of past buyers)
- **When did prices last change?**

---

## Context & Input Gathering

Ask the user for the following if not already provided:

1. **Business type** — What is sold, how, and at what price point?
2. **Current average transaction value** — What does a typical customer spend per visit/purchase?
3. **Purchase frequency** — How often does a typical customer buy? Monthly, annually, once?
4. **Customer list status** — Does a list of past customers exist? How far back does it go?
5. **Last price change** — When were prices last raised, and by how much?
6. **Current upsell/add-on offers** — Is anything offered at the point of purchase beyond the primary item?
7. **Product/service ladder** — Are there multiple price tiers, or a single offering?

---

## Process

### Phase 1: Baseline Current CLV

**Step 1 — Calculate the current CLV estimate.**

Use this formula to establish a baseline:

```
CLV = Average Transaction Value × Average Purchases per Year × Average Customer Lifespan (years)
```

If exact numbers are unavailable, use reasonable estimates and note the assumption. The goal is a directional baseline, not accounting-grade precision.

WHY: You cannot measure improvement without a baseline. Even a rough CLV estimate reveals which lever has the most leverage — a business with high transaction value but low frequency should prioritize frequency; a business with high frequency but low transaction value should prioritize price or upsell.

Record the baseline in the output document.

---

### Phase 2: Audit Each of the 5 Levers

For each lever, assess the current state (Active / Partial / Not in use) and identify the single highest-impact action.

**Step 2 — Lever 1: Raise Prices**

Assess: When were prices last raised? By how much? Is inflation being absorbed silently?

The most overlooked lever. Most business owners assume customers will leave if prices rise. In practice, customers are far less price-sensitive than assumed — especially when they are positioned correctly and receiving genuine value. A price increase drops straight to the bottom line with no additional cost attached. By contrast, every dollar of new-customer revenue carries acquisition cost.

Key implementation guidance:
- Give customers a reason why: quality improvements, input cost increases, or an expansion of what they receive.
- Customers won on price will be lost on price — losing price-sensitive customers during a price rise is often net-positive.
- Grandfathering option: apply the increase only to new customers; tell existing customers they are locked in at the current rate. This reinforces their loyalty and makes them feel valued rather than penalized.

Action: Identify the last price change date and draft a 10–20% increase with a rationale statement.

**Step 3 — Lever 2: Upsell**

Assess: Is anything offered at the moment of purchase beyond the primary item?

Upselling is the bundling of add-ons with the primary product or service at the point of sale. It works because of the contrast principle: after a customer commits to a primary "expensive" purchase, add-ons feel comparatively cheap. A $5 accessory feels trivial next to a $100 main purchase. A customer who has just bought a suit will add shirts, ties, and belts without resistance — the contrast makes them seem reasonably priced.

"Would you like fries with that?" is responsible for hundreds of millions of dollars in McDonald's revenue. It is a single question asked at one moment.

Customers are at peak receptivity immediately after buying — they are in a buying state of mind. Waiting to approach them again later is a missed opportunity.

Framing: "Most customers who bought X also bought Y" — social norm framing taps the human desire to fit in with normal buying behavior.

Action: Identify the top 2–3 add-on candidates that complement the primary offering at high margin. Draft the upsell offer and the moment it will be presented.

**Step 4 — Lever 3: Ascension**

Assess: Does the business have a standard and a premium tier? Is there a product/service ladder?

Ascension is the process of moving existing customers from lower-tier to higher-priced, higher-margin products and services over time. It must be a constant part of the marketing process — not a one-time pitch. Customers stay on existing products due to inertia. When their current option no longer meets their needs, they shop competitors rather than asking you for an upgrade.

Having only a single pricing option leaves enormous revenue on the table. Rule of thumb: approximately 10% of existing customers would pay ten times more; approximately 1% would pay one hundred times more. A single-option business never finds out.

Product ladder example:
- Entry level: self-paced course ($300)
- Standard: group coaching program ($1,500)
- Premium: one-to-one engagement ($5,000)
- Ultra-high-ticket: private intensive or retainer ($25,000+)

Ultra-high-ticket items also activate the contrast principle — standard products appear more reasonably priced by comparison.

Action: Map the current product/service ladder. Identify the missing tier (most commonly: the premium or ultra-high-ticket option). Draft what that tier would offer and at what price.

**Step 5 — Lever 4: Increase Frequency**

Assess: How often do customers return? Is there any systematic mechanism to bring them back?

Two proven tactics:

**Reminders:** Products and services with a natural expiry or renewal cycle (car servicing, dental checkups, ink cartridges, pet vaccinations) can be re-purchased on autopilot if the business sends a timely reminder by email, SMS, or post. Fully automatable. Sending reminders is not pushy if the product genuinely benefits the customer — not sending reminders is a disservice.

**Vouchers / forced return visits:** Issue a voucher at checkout valid only from the following day onward (not redeemable on the day of issue) with an expiry date. This forces a return visit and attaches loss aversion to not returning — the customer will "waste" the voucher if they do not come back. Completely distinct from discounting: the current transaction is not discounted; a future visit is incentivized.

**Subscriptions:** Convert a one-time purchase into a recurring delivery. The Dollar Shave Club model — cheap disposable razor blades converted to a monthly subscription — is the archetype. Subscription turns off the customer's price-shopping radar; when the need is automatically handled, they stop comparison-shopping. Any consumable or regularly-used service is a candidate.

Action: Identify which frequency tactic best fits the business model. For reminder-eligible businesses: define the reminder interval and channel. For transactional businesses: design the voucher mechanic or assess whether a subscription tier is feasible.

**Step 6 — Lever 5: Win Back Lapsed Customers (Reactivation)**

Assess: Does a lapsed customer list exist? When was it last contacted?

Past customers have already crossed the trust chasm between prospect and customer. They are 21 times more likely to buy again than a cold prospect. The list of past buyers is a gold mine — most of the hard work of establishing trust is already done. Reactivation campaigns produce fast cash from a warm audience.

**3-Step Reactivation Procedure:**

1. Pull the lapsed customer list from the CRM or spreadsheet. Define "lapsed" for this business (no purchase in 6 months? 12 months?). Filter out bad-fit customers you do not want back.
2. Create a strong offer: a gift card, coupon, or free offer with a clear call to action. The offer must be compelling enough to overcome inertia.
3. Contact them. Acknowledge the gap — they haven't heard from you in a while. Ask why they haven't returned. If the reason is a business error: apologize and describe the corrective action taken. If they reactivate: follow up after the next purchase to make them feel valued.

Campaign headline options: "We Miss You" / "Have We Done Something Wrong?"

Run reactivation campaigns on a quarterly cadence as a standing item in the marketing calendar.

---

### Phase 3: Prioritize and Sequence

**Step 7 — Identify the highest-leverage lever for this business.**

Score each lever on two dimensions:
- **Impact** (H/M/L): How much could this realistically move CLV given the business model?
- **Effort** (H/M/L): How much work is required to implement the first version?

Prioritize levers with High Impact and Low-to-Medium Effort. For most small businesses: Raise Prices and Reactivation deliver the fastest bottom-line impact with the lowest effort. Upsell and Frequency systems take slightly longer to design but compound over time. Ascension is the highest-ceiling lever but requires building new products.

WHY: Trying to implement all five levers simultaneously produces none of them well. A sequenced 90-day plan with one lever per month is more likely to produce results than a five-lever launch that stalls at week two.

**Step 8 — Draft the 90-day implementation sequence.**

Assign each prioritized lever to a 30-day window. Include:
- Lever name
- Specific tactic to implement
- Who is responsible
- What "done" looks like
- How success will be measured

---

### Phase 4: Draft the CLV Growth Plan Document

**Step 9 — Write clv-growth-plan.md.**

Structure:

```
# CLV Growth Plan — [Business Name]
Generated: [date]

## Baseline
- Average transaction value: $[X]
- Average purchase frequency: [X] per year
- Average customer lifespan: [X] years
- Estimated current CLV: $[X]

## Lever Audit

| Lever | Current State | Priority | First Action |
|-------|--------------|----------|-------------|
| 1. Raise Prices | [Active/Partial/Not in use] | [H/M/L] | [action] |
| 2. Upsell | [Active/Partial/Not in use] | [H/M/L] | [action] |
| 3. Ascension | [Active/Partial/Not in use] | [H/M/L] | [action] |
| 4. Increase Frequency | [Active/Partial/Not in use] | [H/M/L] | [action] |
| 5. Reactivation | [Active/Partial/Not in use] | [H/M/L] | [action] |

## 90-Day Implementation Sequence

### Month 1: [Top-priority lever]
- Tactic: [specific action]
- Owner: [role]
- Done when: [completion signal]
- Measure: [metric]

### Month 2: [Second lever]
...

### Month 3: [Third lever]
...

## Reactivation Campaign
- Lapsed customer definition: no purchase in [X] months
- List size: [X] contacts
- Offer: [gift card / coupon / free offer]
- Headline: ["We Miss You" / "Have We Done Something Wrong?" / custom]
- Contact channel: [email / SMS / post]
- Follow-up: [post-reactivation action]

## CLV Target
- Target CLV after 12 months: $[X]
- Primary driver: [lever]
```

---

## Inputs

| Input | Required | Notes |
|-------|----------|-------|
| Business description (product/service, price point) | Yes | Short paragraph if no file exists |
| Average transaction value and purchase frequency | Yes | Estimates acceptable for baseline |
| Customer list or CRM export (lapsed customers) | Yes for reactivation | If absent, note reactivation as deferred |
| Current pricing and last price change date | Yes for Lever 1 | |
| Current product/service ladder | Yes for Lever 3 | Can be inferred from conversation |

---

## Outputs

This skill produces one primary document:

1. **clv-growth-plan.md** — CLV baseline, lever-by-lever audit table, 90-day priority sequence, specific tactics per lever, reactivation campaign template, and 12-month CLV target

---

## Key Principles

**The diamond mine is under your feet.** Most businesses leave their existing customer base largely untapped while spending marketing budget on new acquisition. Past customers have already crossed the trust chasm — they are 21 times more likely to buy again. The existing and lapsed customer base is the richest source of profit available.

**Price increases go straight to the bottom line.** Every dollar of new customer revenue carries acquisition cost. A price increase on existing customers does not. Most customers are less price-sensitive than business owners assume. Holding prices constant while inflation rises is a silent pay cut.

**The contrast principle makes upsells work.** After committing to the primary purchase, add-ons feel cheap by comparison. The moment of purchase is peak buying receptivity — waiting to offer more later is a structural missed opportunity.

**A single price option is money left on the table.** Approximately 10% of customers would pay 10x more; approximately 1% would pay 100x more. Without a premium and ultra-high-ticket tier, the business never discovers who they are.

**Subscriptions turn off the price-shopping radar.** When a recurring need is automatically handled, customers stop comparison-shopping. Convenience has a price — most customers accept it.

**Reactivation is fast cash.** Past customers are a warm audience. A simple three-step campaign — pull list, create offer, make contact — produces results from people who already trust the business.

**Focus on the existing base, not just new acquisition.** Two ways to grow a business: get new customers, or make more from existing and past ones. The second path is higher-margin, faster, and requires no new positioning or advertising spend.

---

## Examples

### Example 1: Shoe Store — Voucher Forcing Return Visits (Lever 4)

A specialty shoe store about an hour's drive from the customer's home issued a voucher worth $30 for every $100 spent at checkout. A $300 purchase produced a $90 voucher. The voucher carried an expiry date approximately six months out, but critically: it was only valid from the day after issue — it could not be redeemed immediately.

The customer came home, told her husband about the voucher, and he was dragged back the next day. They spent $200 more. A new $60 voucher was issued at that checkout. The wife — tired of the drive — nearly pleaded with the cashier not to give it to her. The cashier apologised and said it was store policy.

With one tactic, the store had nearly doubled the initial transaction value and created psychological loss aversion around not returning. The voucher is not a discount — it does not reduce the current transaction. It creates a future asset the customer will "waste" by not returning.

**Application for a service business:** A massage clinic issues a "pre-loaded session credit" voucher valid from the following week with a 90-day expiry. The customer has already paid for the session in their mind — not booking forfeits it.

### Example 2: SaaS Tool — Ascension Ladder (Lever 3)

A solo developer sells a productivity tool at a single flat rate of $29/month. The product is profitable but growth is flat. Applying the ascension lever reveals the missing tiers:

- Solo plan: $29/month (existing)
- Team plan: $99/month — 5 seats + team admin features
- Business plan: $249/month — unlimited seats + priority support + API access
- Agency/white-label: $999/month — client workspaces + custom branding

Within 6 months, 12% of existing customers upgraded to Team, and 2% moved to Business. Monthly recurring revenue increased 38% with no new customer acquisition.

WHY this works: customers on the solo plan had expanded use cases that were not being served. They were not being asked to move up — inertia kept them at the entry tier. The ascension campaign surfaced those customers and gave them a path.

### Example 3: Trades Business — Reactivation Campaign (Lever 5)

A residential plumbing company had 847 customers in their CRM who had not booked in over 18 months. They ran a 3-step reactivation:

1. Pulled the list; filtered out 23 addresses flagged as problematic. Remaining list: 824.
2. Offer: "$50 off any service call booked in the next 30 days."
3. Sent an email with the subject line "Have We Done Something Wrong?" — acknowledged the gap and asked for feedback.

Result: 67 customers booked (8.1% conversion). Average service call value $280. Revenue: ~$18,760 from a single email to a dormant list. Cost: one afternoon of setup and a $50 discount per job. Follow-up email sent to all 67 after the job was completed.

---

## References

- Dib, A. (2016). *The 1-Page Marketing Plan*, Chapter 8: Increasing Customer Lifetime Value (pp. 222–247). Growth levers section: pp. 222–236.
- Cialdini, R. B. (1984). *Influence: The Psychology of Persuasion* — Contrast principle referenced on pp. 228–230 of source text. The contrast principle explains why add-ons feel cheap after the primary purchase and why ultra-high-ticket items make standard products appear reasonably priced.
- Dollar Shave Club subscription model referenced p. 233 as the archetype for converting consumable products into recurring revenue subscriptions.
- Dependency: `customer-experience-systems-design` — CLV growth levers built on a poor customer experience accelerate churn rather than retention. Establish experience baseline before implementing growth campaigns.

## License

This skill is licensed under [CC-BY-SA-4.0](https://creativecommons.org/licenses/by-sa/4.0/).
Source: [BookForge](https://github.com/bookforge-ai/bookforge-skills) — The 1-Page Marketing Plan by Allan Dib.

## Related BookForge Skills

Install related skills from ClawhHub:
- `clawhub install bookforge-customer-experience-systems-design`

Or install the full book set from GitHub: [bookforge-skills](https://github.com/bookforge-ai/bookforge-skills)

---
name: customer-experience-systems-design
description: Design and document the four core business systems (Marketing, Sales, Fulfillment, Administration) and create memorable customer experiences through experience design and innovation. Use this skill when you want to systematize your business, build an operations manual, create checklists for recurring tasks, scale without being in the room, pass the fire-yourself test, remove yourself as the bottleneck, eliminate owner-dependency, design customer experience touchpoints, apply innovation to a boring or commodity business, differentiate by experience rather than product, run an E-Myth systems audit, identify the manager role missing from your business, document procedures so staff can run the business without you, build a business asset rather than buying yourself a job, or answer "what would happen if I left for six months?"
version: 1.0.0
homepage: https://github.com/bookforge-ai/bookforge-skills/tree/main/books/the-1-page-marketing-plan/skills/customer-experience-systems-design
metadata: {"openclaw":{"emoji":"📚","homepage":"https://github.com/bookforge-ai/bookforge-skills"}}
status: published
source-books:
  - id: the-1-page-marketing-plan
    title: "The 1-Page Marketing Plan: Get New Customers, Make More Money, and Stand Out from the Crowd"
    authors: ["Allan Dib"]
    chapters: [7]
tags: [business-systems, customer-experience, operations, small-business, checklists, operations-manual, experience-design, systematization]
depends-on: []
execution:
  tier: 1
  mode: full
  inputs:
    - type: document
      description: "Business description — what the business does, current team size, key recurring tasks, and the owner's current role"
  tools-required: [Read, Write]
  tools-optional: []
  mcps-required: []
  environment: "Document set. Typical files: business-description.md, customer-journey.md, operations-manual.md (to be created)."
---

# Customer Experience Systems Design

## When to Use

You are a small business owner who wants to stop being the bottleneck in your own business. Typical triggers:

- You cannot take a vacation without the business suffering
- A key task only works when you personally handle it
- You want to hire or delegate but have no documented process to hand over
- You are growing but quality is inconsistent because delivery depends on individual judgment
- You want to sell your business one day but realize it cannot run without you
- You are competing on price because customers cannot see a meaningful difference in your experience
- You want to create loyalty and referrals, not just one-off transactions

Before starting, clarify:
- **What is the business?** (product/service, industry, team size — solo or staff)
- **What are the main recurring tasks?** (List everything the owner currently does)
- **Is this audit or build?** Audit = map the gap between current state and documented systems. Build = create the first checklist for the highest-priority task.

---

## Context & Input Gathering

Ask the user for the following if not already provided:

1. **Business description** — What does the business sell? How does it deliver value?
2. **Current team** — Is the owner a sole operator? Are there staff?
3. **Pain point** — What breaks or degrades when the owner is absent?
4. **Aspiration** — What does "running without me" look like in 1-2 years?
5. **One recent task** — Describe the last recurring task you handled personally this week. This becomes the first checklist candidate.

---

## Process

### Phase 1: Systems Coverage Audit

**Step 1 — Map the four business systems.**

Every business, regardless of industry or size, requires four systems to function. Map all current activity against them:

| System | Purpose | Typical Tasks |
|--------|---------|--------------|
| **Marketing** | Generate a consistent flow of new leads | Ad creation, content publishing, lead magnet delivery, social media, email list management |
| **Sales** | Nurture leads and convert to paying customers | Follow-up calls, proposal writing, CRM updates, contract signing, onboarding |
| **Fulfillment** | Deliver the product or service in exchange for payment | Service delivery, product assembly, project management, quality checks, client communication |
| **Administration** | Support all other functions | Invoicing, accounts receivable, bookkeeping, HR, reception, scheduling, supplier management |

WHY: Most small businesses over-invest in Fulfillment and Administration while neglecting Marketing and Sales. No one presses a deadline on marketing tasks — so they slip. Mapping reveals the imbalance.

**Step 2 — Score each system's documentation coverage.**

For each of the four systems, rate current documentation on a 0–3 scale:
- 0 = All in the owner's head, nothing written
- 1 = Some informal notes or email threads
- 2 = Partial written procedures, but incomplete
- 3 = Full checklists with steps, ownership, and triggers documented

Fill in the coverage matrix:

| System | Coverage Score (0–3) | Biggest Undocumented Task | Owner-Dependent? |
|--------|---------------------|--------------------------|-----------------|
| Marketing | | | |
| Sales | | | |
| Fulfillment | | | |
| Administration | | | |

**Step 3 — Apply the fire-yourself test.**

Ask: *"If I left this business for six months with no contact, would it thrive, survive, or collapse?"*

- **Thrive** — Systems are documented and staff can run them. The owner is an investor.
- **Survive** — Key tasks muddle through but quality drops. Partial systems exist.
- **Collapse** — The business IS the owner. The know-how lives entirely in one person's head.

If the answer is Survive or Collapse, the owner is the bottleneck. The goal of this skill is to remove that bottleneck through documentation — not by hiring more people, but by extracting the know-how from the owner's head into replicable checklists.

WHY: A business that cannot run without its owner is not a business — it is a self-made prison. It also has no sale value. An investor or acquirer pays for systems, not for the owner's personal effort.

---

### Phase 2: Prioritized Documentation Plan

**Step 4 — Identify the three highest-priority undocumented tasks.**

Prioritize by: (a) frequency (daily/weekly tasks have the most leverage), (b) bottleneck impact (tasks only the owner can currently perform), and (c) revenue impact (tasks that directly touch the customer experience or cash flow).

For each candidate task, capture:
- Task name
- System it belongs to
- How often it recurs
- Current owner (person or role)
- What breaks if it is skipped or done wrong

Output: a ranked documentation backlog — the order in which to build checklists.

**Step 5 — Think at 10x scale.**

Before writing the first checklist, ask: *"If this business were ten times its current size, what roles would exist?"*

List those roles (e.g., bookkeeper, marketing coordinator, sales rep, fulfillment manager). Even as a sole operator, the tasks performed map to these roles. Documenting tasks by role — not by person — makes delegation straightforward when it is time to hire.

WHY: Documenting for the current size creates procedures that feel personal and tied to one person. Documenting for 10x creates a transferable system that scales.

---

### Phase 3: Build the First Checklist

**Step 6 — Write the checklist for Task #1.**

Use this structure for every checklist:

```
CHECKLIST: [Task Name]
System: [Marketing / Sales / Fulfillment / Administration]
Role responsible: [Role name — not person name]
Trigger: [What starts this task — a date, an event, a customer action]
Frequency: [Daily / Weekly / Monthly / On-trigger]

STEPS:
1. [Action] — [Why this step matters or what to watch for]
2. [Action] — [Why]
3. [Action] — [Why]
...

ESCALATION: [What to do if a step fails or produces an unexpected result]
COMPLETION SIGNAL: [How you know the task is done correctly]
```

Example — Accounts Receivable Follow-Up (Administration system):

```
CHECKLIST: Accounts Receivable Follow-Up
System: Administration
Role responsible: Bookkeeper
Trigger: End of week accounts receivable report
Frequency: Weekly

STEPS:
1. Run accounts receivable report from accounting software — establishes which invoices are outstanding
2. For invoices 7–13 days overdue: send a friendly payment reminder email using Template AR-1 — early contact preserves the relationship while prompting action
3. For invoices 14–27 days overdue: call the customer directly to remind them to pay — phone contact signals seriousness without hostility
4. For invoices more than 27 days overdue: forward to debt collection agency using Procedure AR-Debt — beyond this threshold, internal follow-up costs exceed recovery likelihood

ESCALATION: If a customer disputes an invoice amount, escalate to the business owner within 24 hours. Do not proceed to debt collection without owner sign-off.
COMPLETION SIGNAL: All overdue invoices categorized and actioned. Report filed in shared folder with date stamp.
```

WHY checklists over training manuals: Checklists are executable. Training manuals are read once and forgotten. A checklist runs every time a task runs, enforcing the same standard regardless of who performs the step. This is how McDonald's delivers consistent food quality using teenagers — the system does the work, not the individual.

---

### Phase 4: Experience Design and Innovation

**Step 7 — Audit customer touchpoints for memorable moments.**

Map the customer journey through your business (before purchase, during delivery, after delivery). For each touchpoint, ask:
- Is this moment forgettable or memorable?
- Does it serve only the customer's functional need, or does it also create delight?
- Is there a pain point here that we could eliminate — even if competitors accept it as normal?

**Step 8 — Generate innovation opportunities.**

Innovation does not require a new product. Innovation can be applied to how the product is priced, packaged, delivered, supported, or marketed. For each touchpoint where the current experience is functional but forgettable, generate one idea that adds theater, removes friction, or turns a liability into an asset.

Use this prompt for each touchpoint:
*"What would make a customer tell a friend about this moment — not the product itself, but this specific experience?"*

Output: an innovation opportunity list with one idea per touchpoint, ordered by ease of implementation.

WHY: Businesses that compete only on the product end up competing on price. The experience around the product is where differentiation happens and where loyalty is built.

---

## Inputs

| Input | Required | Notes |
|-------|----------|-------|
| Business description (industry, product/service, team) | Yes | Can be a short paragraph if no file exists |
| List of recurring tasks the owner currently handles | Yes | Can be gathered through conversation |
| Customer journey or touchpoint map | Optional | If absent, construct one from the business description |

---

## Outputs

This skill produces up to four documents:

1. **systems-coverage-matrix.md** — The four-system audit table with coverage scores, undocumented tasks, and bottleneck flags
2. **documentation-backlog.md** — Prioritized list of tasks to document, with system classification and rationale
3. **[task-name]-checklist.md** — The first complete checklist, ready to hand to a staff member or outsourcer
4. **innovation-opportunities.md** — List of experience design ideas mapped to customer touchpoints

---

## Key Principles

**Products make money; systems make a fortune.** A business with documented systems is a transferable asset. A business without them is a job — and a job that owns you rather than one you own.

**Sell what they want; deliver what they need.** Customers buy outcomes (ripped abs, not health). Design the delivery experience around the desired outcome, not just the product feature. A fitness instructor sells appearance improvements but delivers health — they must bridge the gap or customers will not use the service and will blame the product.

**Document for a business ten times your current size.** Think in roles, not people. Every sole operator performs multiple roles simultaneously. Documenting by role makes the system scalable and transferable.

**Checklists over training.** Checklists run every time a task runs. Training is done once. Checklists are the executable layer of your operations manual.

**Tell them all the trouble you go to.** The effort behind your product or service is part of the value. Guinness turned a pouring time of 119.5 seconds from a liability into a marketing campaign. The backstory of your quality is marketing material — do not let it go unnoticed.

**Innovation lives in the experience, not the product.** Any business — including a blender manufacturer — can innovate in how it delivers, packages, or markets its product. The question is not "is my product innovative?" but "is my customer's experience with my product memorable?"

---

## Examples

### Example 1: Blendtec — Experience Innovation for a Commodity Product

Blendtec makes kitchen blenders. The product is ordinary. The experience they created was extraordinary: a YouTube series called "Will It Blend?" featuring a scientist blending iPhones, golf balls, and iPads. Hundreds of millions of views. Near-zero production cost relative to reach.

The lesson: You do not have to invent a new product. You have to invent a reason for people to talk about your product. What can you demonstrate, showcase, or do with your product that would make someone share it?

**Application:** For a plumbing company, this might be a YouTube series showing the most disgusting pipe blockages they have ever fixed — and how they solved them. Gross, memorable, shareable, and specific to the product.

### Example 2: Restaurant with Pickup and Drop-Off Service

A restaurant in a competitive suburb added a complimentary pickup and drop-off service for dinner bookings within a 5km radius. Customers no longer had to worry about driving under the influence. The restaurant benefited by selling more alcohol — their highest-margin product. The customer friction (drink or drive) became a selling point.

The lesson: Identify the thing that stops your customer from fully enjoying your product. Remove it. The removal often creates a new revenue opportunity.

**Application:** For a photography studio, this might be a "worry-free booking" package that includes location scouting, weather contingency rescheduling at no extra cost, and a print-guarantee promise. The barriers to booking become the reason to book.

### Example 3: Accounts Receivable Follow-Up — Administration System

A service business where the owner personally chased every late invoice applied the 4-step checklist above. The owner handed the checklist to a part-time bookkeeper. Late payment collection time dropped by 40%. The owner freed 3 hours per week. The bookkeeper had a clear escalation path and did not need to judge when to call versus email.

This is what systematization does: it moves the judgment from the person into the procedure, so the person executes rather than decides from scratch every time.

### Example 4: Consulting Firm — Sales System Documentation

A solo management consultant was the only person who could respond to inbound inquiries because all the qualification questions lived in their head. They created a Sales system checklist:

```
CHECKLIST: Inbound Lead Qualification
System: Sales
Role responsible: Sales Coordinator
Trigger: New inquiry form submission or inbound call
Frequency: On-trigger

STEPS:
1. Send acknowledgment email within 2 hours using Template S-1 — confirms receipt, sets expectations, differentiates from competitors who take days
2. Review inquiry against Ideal Client Profile (budget >$10k, decision-maker is the contact, project starts within 90 days) — filters time-wasters early
3. If qualified: book a 30-minute discovery call using the calendar link in Template S-2
4. If not qualified: send polite decline with referral to alternative resource using Template S-3 — preserves goodwill, prevents scope creep
5. Log outcome in CRM with tag [qualified] or [declined] and reason code

ESCALATION: If the prospect requests a proposal before a discovery call, escalate to the consultant — do not proceed without a scoping conversation.
COMPLETION SIGNAL: Lead status updated in CRM. Next action scheduled or closed.
```

The consultant hired a part-time coordinator to run this checklist. They reclaimed 6 hours per week and converted more leads because response time dropped from 2 days to 2 hours.

---

## References

- Dib, A. (2016). *The 1-Page Marketing Plan*, Chapter 7: Delivering a World-Class Experience (pp. 191–221).
- Gerber, M. (1995). *The E-Myth Revisited* — foundational text on the Entrepreneur/Specialist/Manager role triad and why small businesses fail to systematize.
- The three-role framework (Entrepreneur: makes it up; Specialist: makes it real; Manager: makes it recur) — all three are required for a functioning business. Most small businesses have the first two but are missing the Manager role. Documented systems are how the Manager role gets installed in the business.

## License

This skill is licensed under [CC-BY-SA-4.0](https://creativecommons.org/licenses/by-sa/4.0/).
Source: [BookForge](https://github.com/bookforge-ai/bookforge-skills) — The 1-Page Marketing Plan by Allan Dib.

## Related BookForge Skills

No direct dependencies. Install the full book set from GitHub.

Or install the full book set from GitHub: [bookforge-skills](https://github.com/bookforge-ai/bookforge-skills)

---
name: advertising-media-roi-framework
id: advertising-media-roi-framework
title: Advertising Media & ROI Framework
description: >
  Use when selecting advertising media channels, calculating Customer Acquisition Cost (CAC) per channel,
  deciding whether to stop, measure, or scale each channel, checking for dangerous single-point-of-failure
  lead source dependencies, or building a diversified media plan with per-channel tracking setup and
  ROI decision rules. Triggers: "advertising media", "ROI calculation", "CAC", "customer acquisition cost",
  "where to advertise", "marketing channels", "lead sources", "stop losing money on ads", "scale marketing",
  "single point of failure marketing", "diversify ad channels", "track marketing ROI", "media plan",
  "fill square 3", "which channels should I use", "is my advertising working".
version: 1.0.0
homepage: https://github.com/bookforge-ai/bookforge-skills/tree/main/books/the-1-page-marketing-plan/skills/advertising-media-roi-framework
metadata: {"openclaw":{"emoji":"📚","homepage":"https://github.com/bookforge-ai/bookforge-skills"}}
status: published
source-books:
  - title: "The 1-Page Marketing Plan"
    author: Allan Dib
    chapters: [3]
source-book: the-1-page-marketing-plan
source-chapters: [3]
quality:
  scores:
    with_skill: 92.9
    baseline: 7.1
    delta: 85.7
  tested_at: "2026-04-09"
  eval_count: 1
  assertion_count: 14
  iterations_needed: 1
depends-on:
  - target-market-selection-pvp-index
  - marketing-message-and-usp-crafting
  - lead-capture-ethical-bribe-design
tags:
  - marketing
  - advertising
  - media-selection
  - roi
  - cac
  - small-business
---

# Advertising Media & ROI Framework

## When to Use

Use this skill when a business owner needs to:
- Decide which advertising channels to use to reach their target market
- Calculate whether current ad spend is generating positive or negative return
- Determine whether to stop, fix measurement on, or aggressively scale a channel
- Audit their lead source portfolio for dangerous single-source dependency
- Build a structured media plan with per-channel budgets, tracking mechanisms, and decision rules

This is square 3 of the 1-Page Marketing Plan canvas: **What media will you use to reach your target market?**

Do not confuse media selection with market selection (square 1) or message crafting (square 2). All three must be right for a campaign to succeed.

---

## Context & Input Gathering

Before building the media plan, confirm these inputs are known. If any are missing, resolve them first.

**Required inputs:**

| Input | Source |
|-------|--------|
| Target market definition | IF unknown → invoke `target-market-selection-pvp-index` OR ask: "Who is your ideal customer — describe them specifically." |
| Marketing message / Unique Selling Proposition (USP) | IF unknown → invoke `marketing-message-and-usp-crafting` OR ask: "What is the core message you send to prospects?" |
| Lead capture offer (ethical bribe) | IF unknown → invoke `lead-capture-ethical-bribe-design` OR ask: "What do you offer prospects in exchange for their contact details?" |
| Profit per sale (front-end) | Ask: "What is your average profit per transaction on the first sale?" |
| Customer Lifetime Value (CLV) estimate | Ask: "What does a typical customer spend with you over their entire relationship? (rough estimate is fine)" |
| Current advertising channels + monthly spend | Ask: "List every channel you currently advertise on and how much you spend per month on each." |
| Tracking currently in place | Ask: "How do you currently track where your leads and customers come from?" |

**Why gather all inputs first:** The CAC calculation requires profit-per-sale and CLV to make stop/scale decisions. Skipping this step leads to decisions based on vanity metrics (response rate, open rate) rather than actual ROI.

---

## Process

### Step 1 — Audit Existing Channels

For each channel the business currently uses:

1. Record: channel name, monthly ad spend, leads generated, customers acquired.
2. Calculate CAC per channel (formula below).
3. Compare CAC to profit-per-sale and CLV.
4. Apply the 3-Scenario Decision Rule.
5. Flag any channel with no tracking in place as Priority Fix.

**Why audit before adding new channels:** Scaling broken channels wastes money. Cutting losers frees budget for winners and new tests.

---

### Step 2 — Calculate Customer Acquisition Cost (CAC)

**The CAC Formula:**

```
CAC = Total Campaign Cost ÷ Number of Customers Acquired
```

**Worked example (direct mail):**
- Send 100 letters at a total print + postage cost of $300
- 10 people respond (10% response rate)
- 2 of those 10 become customers (20% closure rate)
- Customers acquired = 2
- **CAC = $300 ÷ 2 = $150**

**Decision:**
- If profit per sale = $600 → net gain = $600 − $150 = **+$450 per customer** → WINNING
- If profit per sale = $100 → net loss = $100 − $150 = **−$50 per customer** → LOSING

**Important:** Response rate and conversion rate are intermediate metrics only. They help calculate CAC but are not the goal. The only numbers that matter are **CAC** and **CLV**.

**The CLV exception:** If front-end profit is less than CAC but CLV is confirmed high (e.g., subscription businesses, repeat-purchase businesses), accepting a front-end loss is a deliberate strategy — not a failure. Do not apply this exception until CLV is measured, not estimated.

---

### Step 3 — Apply the 3-Scenario Decision Rule

For every channel, exactly one of three situations is true:

| Scenario | Condition | Action |
|----------|-----------|--------|
| **FAIL** | Profit per sale < CAC (negative ROI), consistently | **STOP.** Change the channel, the message, or the offer. Do not continue spending on a confirmed loser. |
| **UNMEASURED** | No tracking in place; ROI unknown | **FIX MEASUREMENT IMMEDIATELY.** With tools readily available (toll-free numbers, UTM codes, coupon codes, dedicated landing pages, call tracking), not measuring is inexcusable. Treat this as urgent. |
| **WIN** | Profit per sale > CAC (positive ROI), consistently | **SCALE AGGRESSIVELY. No budget cap.** Winning marketing is not an expense — it is a legal money printing press. Capping spend on a winner is like refusing to buy $100 bills at $80. |

**Why no budget cap on winners:** Setting a fixed "marketing budget" implies either (a) marketing is not working or (b) you do not know if it is working. If it works, the only sensible limit is operational capacity — and even that signals it is time to raise prices.

**Testing phase exception:** During initial testing, spend conservatively. Fail fast and fail cheap. Test headline, offer, positioning, and channel variables one at a time. Once a winner emerges, remove the cap.

---

### Step 4 — Check 5-Source Diversification

Count the number of distinct, active lead sources generating customers.

**Rule: Minimum 5 different lead sources.**

**Why 5 is the floor:** One is the most dangerous number in any business. Single-source dependency creates fragility:

- **Google Slap** (2000s–2010s): Google increased pay-per-click costs 4–10x overnight for certain categories, halting campaigns immediately.
- **SEO algorithm change**: Businesses that relied entirely on organic search rankings lost their lead flow overnight when Google updated its algorithm.
- **Fax broadcasting**: Outlawed in the United States. Every business relying solely on this channel went broke.
- **Single large customer departing**: Revenue collapses with no pipeline to replace it.

If fewer than 5 sources exist, identify candidates for new channels to test (see Step 5).

**Prefer paid media for most sources.** Paid media is reliable (if you pay, the ad runs) and forces ROI discipline (if it does not work, you cut it). Free or "organic" methods such as word of mouth have no measurement pressure — time wasted on ineffective free channels carries an opportunity cost that translates to real money.

**Own your marketing assets.** Social media platforms can change their rules at any time. An email subscriber list you own is more valuable than 10x the followers on a platform you do not own.

---

### Step 5 — Identify New Channel Candidates

Match candidate channels to the target market profile:

1. **Where does the target market spend time?** (Online communities, industry publications, local venues, social platforms, physical locations)
2. **What media do they trust?** (Peer referrals, trade publications, search, direct mail, radio)
3. **What channels are competitors NOT using?** (Under-competition means lower cost and higher attention)
4. **What channels allow direct response measurement?** (Prefer channels where you can attach a unique phone number, URL, or code)

Rank candidates by: (a) target market reach, (b) trackability, (c) estimated cost per lead, (d) time to first result.

Select the top 2–3 to test. Do not test more than 3 new channels simultaneously — it obscures what is working.

---

### Step 6 — Set Up Tracking Mechanisms Per Channel

For each channel (existing and new), assign at least one tracking mechanism before spending:

| Tracking Method | Best For |
|-----------------|----------|
| Unique toll-free number per campaign | Direct mail, print, radio, TV |
| UTM parameters on URLs | Online ads, email, social |
| Dedicated landing page per source | Online ads, QR codes in print |
| Coupon or promo codes | Any channel |
| Call tracking software | Phone-heavy businesses |
| "How did you hear about us?" (at point of sale) | Backup for all channels |

**Why tracking is non-negotiable:** Without tracking, Scenario 2 (UNMEASURED) applies to every channel. You cannot cut losers or scale winners. The technology exists and is inexpensive. Not using it is a choice to operate blind.

---

### Step 7 — Draft the Media Plan Document

Produce `media-plan.md` using the structure below.

---

## Inputs

- Target market profile (from `target-market-selection-pvp-index`)
- Marketing message / USP (from `marketing-message-and-usp-crafting`)
- Lead capture offer (from `lead-capture-ethical-bribe-design`)
- Profit per sale (front-end)
- Customer Lifetime Value (CLV)
- Current channel list with spend data
- Current tracking status per channel

---

## Outputs

### Primary: `media-plan.md`

```markdown
# Media Plan — [Business Name]
**Date:** [Date]
**Target Market:** [From square 1]
**Core Message:** [From square 2]
**Lead Capture Offer:** [From square 4]
**Profit Per Sale (Front-End):** $[X]
**Customer Lifetime Value (CLV):** $[X]

---

## Channel Decision Table

| Channel | Monthly Spend | Customers/Mo | CAC | Profit/Sale | Decision | Tracking Method |
|---------|--------------|-------------|-----|-------------|----------|-----------------|
| [e.g. Google Ads] | $[X] | [X] | $[X] | $[X] | SCALE / STOP / MEASURE | UTM + landing page |
| [e.g. Direct Mail] | $[X] | [X] | $[X] | $[X] | SCALE / STOP / MEASURE | Unique phone # |
| [e.g. Email List]  | $[X] | [X] | $[X] | $[X] | SCALE / STOP / MEASURE | Tracked links |
| ... | | | | | | |

**Total active channels:** [X] / 5 minimum required
**Diversification status:** [ADEQUATE / AT RISK — add [X] more sources]

---

## New Channel Tests (next 90 days)

| Channel | Rationale | Test Budget | Tracking Setup | Success Threshold (CAC) |
|---------|-----------|-------------|----------------|-------------------------|
| [Channel] | [Why this market is there] | $[X] | [Method] | CAC < $[X] |

---

## Decision Rules Summary

- Any channel with CAC < profit/sale for 3+ consecutive months → **Scale, no cap**
- Any channel with CAC > profit/sale for 2+ consecutive months → **Stop. Reallocate.**
- Any channel with no tracking → **Fix tracking within 2 weeks or pause spend**
- Total lead sources < 5 → **Add one new test channel per quarter until threshold met**
```

---

## Key Principles

**cac-vs-profit-decision:** The only question that matters for any advertising channel is: does CAC exceed or fall below profit per sale (adjusted for CLV)? Response rates, open rates, click-through rates are intermediate metrics — useful only for diagnosing why CAC is high or low.

**no-budget-cap-on-winning-channels:** Setting a marketing budget on a channel with confirmed positive ROI is equivalent to refusing discounted money. The only legitimate budget limit during active campaigns is operational capacity to fulfill demand — which itself signals an opportunity to raise prices.

**5-source-diversification:** Relying on fewer than 5 lead sources makes the business brittle. Any external change — algorithm updates, regulatory changes, platform policy shifts, cost spikes — can eliminate a single-source business's lead flow overnight.

**paid-over-free-media:** Paid media is preferred for most of the 5 sources because (a) it is reliable — paying for placement means the ad runs, and (b) it forces ROI accountability — when money is at stake, measurement follows. Free channels such as word of mouth carry hidden time costs and no inherent measurement pressure.

**tracking-is-non-negotiable:** Not measuring where leads and sales come from is the mark of an amateur. Modern tracking tools (toll-free numbers, UTM codes, call tracking, dedicated landing pages) are inexpensive and readily available. Running any campaign without tracking is a deliberate choice to waste money.

**reject-arbitrary-marketing-budget:** Treating marketing as a fixed expense category implies it either does not work or is not being measured. Marketing that works is an investment with measurable return, not an expense with a ceiling.

**front-end-back-end-distinction:** Profit per sale on first purchase (front-end) may be lower than CAC in high-CLV businesses. Accepting a front-end loss is only valid when CLV is confirmed through actual measurement — not assumption.

---

## Examples

### Example 1: Direct Mail CAC Calculation (Verbatim)

A plumbing company sends a direct mail campaign:
- 100 letters sent; print + postage cost = $300
- 10 recipients respond (10% response rate)
- 2 of the 10 responders become customers (20% closure rate)
- **CAC = $300 ÷ 2 = $150**

If the profit per job is $600: net per customer = $600 − $150 = **+$450. Scale.**
If the profit per job is $100: net per customer = $100 − $150 = **−$50. Stop and rethink.**

The response rate (10%) is irrelevant in isolation. A 0.5% response rate that yields a CAC of $80 against $600 profit per sale is far better than a 15% response rate that yields a CAC of $200.

---

### Example 2: Google Slap — Single-Source Failure

A software reseller built its entire lead flow on Google pay-per-click (PPC) ads. Monthly spend: $5,000. CAC was positive, business growing.

Google changed its ad classification rules. Overnight, the cost per click increased 5x. Monthly spend to maintain the same lead volume jumped to $25,000 — above the business's operating capacity.

The business had no alternative lead sources. It halted its Google campaign while investigating, and lead flow dropped to zero for 6 weeks. Revenue collapsed.

**Prevention:** Had the business maintained 5 lead sources (e.g., Google PPC + email list + direct mail + SEO content + referral program), one source failing would have reduced lead flow by roughly 20%, not 100%.

---

### Example 3: Multi-Channel Plan for a Local Accounting Firm

**Target market:** Small business owners (1–10 employees) in a single metro area.
**Profit per engagement:** $1,200/year average
**CLV:** $4,800 (4-year average tenure)
**CAC threshold for front-end:** $1,200 (break even on first year)
**CAC threshold with CLV:** $2,000 (still profitable over tenure)

| Channel | Monthly Spend | CAC | Decision | Tracking |
|---------|--------------|-----|----------|----------|
| Google Ads (local) | $800 | $900 | SCALE | UTM + dedicated landing page |
| Direct mail (local biz list) | $400 | $1,100 | MEASURE (3 months) | Unique phone # |
| LinkedIn outreach | $0 (time) | Unknown | MEASURE | CRM tagging |
| Referral program | $200 | $400 | SCALE | Referral codes |
| Email newsletter | $50 | $600 | SCALE | Tracked links |

Total sources: 5. Diversification: ADEQUATE.

Next test: local radio (30-day test, $600 budget, unique phone number, CAC threshold $1,200).

---

## References

- Source: "The 1-Page Marketing Plan" by Allan Dib, Chapter 3 "Reaching Prospects with Advertising Media," pp. 103–126
- Related skills: `target-market-selection-pvp-index` (square 1), `marketing-message-and-usp-crafting` (square 2), `lead-capture-ethical-bribe-design` (square 4), `customer-lifetime-value-growth` (CLV expansion), `marketing-metrics-dashboard` (ongoing measurement)
- The 3M framework: Market (Ch. 1) + Message (Ch. 2) + Media (Ch. 3) must all be correct for a campaign to succeed. Failure on any one causes campaign failure regardless of the other two.

## License

This skill is licensed under [CC-BY-SA-4.0](https://creativecommons.org/licenses/by-sa/4.0/).
Source: [BookForge](https://github.com/bookforge-ai/bookforge-skills) — The 1-Page Marketing Plan by Allan Dib.

## Related BookForge Skills

Install related skills from ClawhHub:
- `clawhub install bookforge-target-market-selection-pvp-index`
- `clawhub install bookforge-marketing-message-and-usp-crafting`
- `clawhub install bookforge-lead-capture-ethical-bribe-design`

Or install the full book set from GitHub: [bookforge-skills](https://github.com/bookforge-ai/bookforge-skills)