Skip to main content
ClaudeWave
Skill596 repo starsupdated 3d ago

convert-proof

|

Install in Claude Code
Copy
git clone --depth 1 https://github.com/FerroxLabs/wayland /tmp/convert-proof && cp -r /tmp/convert-proof/resources/bundled-extensions/business-conversion/skills/convert-proof ~/.claude/skills/convert-proof
Then start a new Claude Code session; the skill loads automatically.

SKILL.md

# Convert Proof - Conversational Proof

> *"Proof in Donahoe copy never announces itself. It doesn't feel like 'evidence.' It feels like conversation - like you're just telling them what happened, what you saw, what the numbers said."* - The Donahoe Method, Framework 2

This skill turns claims into believable, conversational proof using the four-step pattern (claim → evidence → reason → move on) and the five-type proof library (Anecdote, Receipt, Drive-By, Namecheck, Because).

The output is proof woven through copy - not a "proof section" with a logo bar.

## When to Use

Trigger phrases: "add proof to this", "conversational proof", "donahoe proof", "prove this claim", "back this up", "testimonial as conversation", "drop the receipts", "skin the claim", `/convert proof <claim or section>`.

Use whenever copy contains a claim that needs to be believed but currently reads as assertion. Works at any granularity - a single sentence, a single bullet, a section, or a full draft.

Do NOT use for:
- Designing the visual proof layout (logo bars, video grids, stat blocks) - that's `convert-proof-stack`
- Sourcing testimonials from customers (research task, not copy task)
- Auditing existing proof for Method-fit (use `convert-audit`)
- Generating bullets from scratch (use `convert-bullets` - though they import proof patterns)

## Inputs

Required:
1. **The claim(s)** - what does the copy assert that needs to be believed? "This system generates leads while you sleep." / "Most copywriting courses are wasting your time." / "I made $14k in 11 days." Each claim becomes a proof unit.
2. **Available evidence** - anecdotes (real customer stories), receipts (numbers, dates, screenshots), credentials (resume items the seller has), names (people willing to be quoted), mechanisms (the *why* behind the result).

Optional:
- **Asset type** and where the claim sits in the flow - affects which proof type leads
- **Temperature** of the reader (from `convert-temperature`) - Cool audiences need Because-heavy proof; Boiling audiences need single Receipt
- **Existing voice samples** - match tone
- `out_path` - caller-controlled output. Defaults via `build_report_path`.

## The Pattern (full method)

Every proof unit moves through four steps:

### 1. Make the claim - casually, like it's obvious

Don't announce the proof is coming. Just state the thing.

> *"This system generates leads while you sleep."*

The casualness IS part of the proof. Confidence comes from people who don't need to prep you for the claim.

### 2. Drop the evidence - wrapped in story or specific detail

Never a credential dump. Never a formal "proof section." The evidence comes through naturally, in the form a friend would tell another friend.

> *"Mike - a plumber from Ohio who'd never run an online ad in his life - set this up on a Thursday and woke up Saturday to 47 leads in his inbox. He actually called me because he thought something was broken."*

The evidence works because of the specificity (Mike, Ohio, plumber, Thursday→Saturday, 47, called me, thought it was broken). Each detail is a fingerprint of truth.

### 3. Give the reason - explain WHY this result is possible

People need the logic to justify the emotional decision they've already made. Give them the argument they'll use when someone asks *"why did you buy that?"*

> *"It works because the targeting is built on actual purchase behavior, not the interest-based nonsense that most platforms default to."*

The Reason makes the result feel logical, not miraculous. Skipping the Reason is why miracle-result copy reads as scammy even when the results are real.

### 4. Move on - don't oversell the proof

State it and keep going. Confidence means not needing to hammer the point.

> *"But the leads are actually the easy part. What matters is what happens next..."*

The move-on doubles as a transition into the next section - chute momentum (see `convert-chute`).

## The Five Types of Proof (priority order)

### 1. The Anecdote

The most natural proof in conversation. Nobody tells their friends *"studies show."* They say *"my buddy tried this and..."* Stories are how humans have transferred credible information for thousands of years.

Specifications:
- A specific person (named or vividly described)
- A specific moment in time
- A specific outcome
- A small detail that makes it feel real (the thing they did wrong, the moment they realized, the thing they said)

Example:
> *"Mike - a plumber from Ohio who'd never run an online ad in his life - set this up on a Thursday and woke up Saturday to 47 leads in his inbox. He actually called me because he thought something was broken."*

### 2. The Receipt

Numbers. Dates. Exact amounts. **Specificity IS credibility.** Not *"I made money"* but *"I pulled $14,327 in 11 days from a list of 2,200 people."* The reader's brain automatically does the math - and specific numbers feel true in a way that round numbers never do.

Specifications:
- Non-round numbers ($14,327 not $15,000)
- Real dates or durations (11 days, last March, Q2)
- Counts, not estimates (2,200 not "thousands")
- Optionally: source ("from the campaign dashboard, screenshot below")

Example:
> *"I pulled $14,327 in 11 days from a list of 2,200 people - most of whom hadn't bought from me in over a year."*

### 3. The Drive-By

Your credentials mentioned in passing, never announced. *"...which I figured out after running campaigns for about 15 years"* hits ten times harder than a bio block that says *"With 15 years of industry experience."*

The Drive-By works because it signals you don't NEED the credential to prop you up. You just have it.

Specifications:
- Slipped into a sentence with another purpose
- Never the *headline* of a sentence - always the supporting clause
- Casual phrasing ("I've been doing this since 2008", not "industry veteran since 2008")

Example:
> *"...which is the same hypothesis I tested at my last agency, where we ran about $40 million in spend across e-comm clients