/market-copyCompatible with Claude, ChatGPT, Gemini, and Claude Code.
Market Copy
Audits your copywriting and proposes before/after rewrites.
What the skill does
In detail.
Analyzes any page's copy: headlines, subheads, CTAs, unhandled objections. For each weakness it proposes a concrete rewrite with current and optimized versions side by side, and explains why it converts better.
When to use it
Your page isn't converting and you suspect the words are the problem.
Usage example
/market-copy https://mysite.com/offer1 · The simplest way
Copy and paste into any AI chat
This button copies the whole skill. Paste it into Claude, ChatGPT, or Gemini, then describe what you need. No installation required.
In a chat, Word or Excel files can't always be generated: you'll receive the document as structured text instead, and Claude will flag it.
Installation
3 steps to use it in Claude.ai
Download the ZIP
Click the button above. For free skills, leave your email to get the link; for premium ones, pay via Stripe then grab the link.
Open Claude.ai and create the skill
On claude.ai, open your profile (bottom left) → Customize → Skills tab → + button → Create skill. Upload the .zip you just downloaded.
Toggle it on and use the skill
The skill shows up in your list — flip the toggle. In any conversation, type the command (e.g. /market-quick followed by a URL).
Using Claude Code (terminal or app)?
The ZIP works there too: unzip it into ~/.claude/skills/ (Mac/Linux) or C:\Users\YourName\.claude\skills\ (Windows), then restart Claude Code.
In depth
How this skill really works
Market Copy analyzes a page's existing copy — headlines, subheads, CTAs, unhandled objections — then scores the whole thing on a 5-dimension rubric (clarity, persuasion, specificity, emotion, action, each out of 10, for a score out of 100) and proposes concrete rewrites presented as before/after. It first detects the page type (8 types: homepage, landing page, pricing, about, product page, feature page, blog post, contact/demo), because each has its own copy priorities and its own documented writing structure.
The skill also analyzes the existing voice across 5 axes (formality, emotion, complexity, humor, authority, each on a 1-5 scale) so any generated copy stays anchored to the brand's tone, unless that tone is clearly ineffective. When a MARKETING-AUDIT.md or BUYER-PERSONA.md file already exists in the project, it additionally checks the copy's alignment against 6 persona elements (the pain that triggers a search, pre-purchase barriers, myths, objections, past bad experiences, power ideas) for a score out of 60.
Every recommendation ships with a concrete before/after and its reasoning, never a vague "improve your headline" note. Output combines a condensed terminal summary and a full report written to COPY-SUGGESTIONS.md.
When to use it
- When a page isn't converting and the problem is likely in the words, not the design.
- When concrete rewrites of headline, subhead, CTA, and meta description are expected, not just a diagnosis.
- Before relaunching a paid traffic campaign to a page whose copy has never been audited.
- When a BUYER-PERSONA.md or MARKETING-AUDIT.md file already exists and the copy needs checking against the real persona.
- To generate a swipe file of alternative headlines, subheads, CTAs, and meta descriptions ranked by estimated effectiveness.
When not to use it
- Not for a fast first look under 30 lines — that's /market-quick.
- Not for a full CRO audit including forms, page speed, and visual hierarchy — that's /market-landing.
- Not for building a brand voice profile from scratch — the skill analyzes existing voice, it doesn't define one; use /market-brand for that.
The method
The 5-dimension scoring rubric
The SKILL.md documents a full copy-scoring rubric across 5 dimensions, each out of 10, for a score out of 50 (doubled to a 0-100 scale). It's this rubric — never a "sounds good" impression — that drives the diagnosis.
- Clarity
- Would a 12-year-old understand what you do? No jargon, no fluff.
- Persuasion
- Does the copy move the reader toward action? Does it handle objections?
- Specificity
- Does it use concrete numbers, outcomes, and timeframes vs. vague claims?
- Emotion
- Does it connect with the reader's pain, desires, identity, or aspirations?
- Action
- Are CTAs clear, compelling, strategically placed, and low-friction?
Steps
Fetch and parse the page
Extract the primary headline, subhead, hero copy, section headlines, body paragraphs, every CTA's text, navigation, footer, meta title/description, and social-proof elements.
Detect page type and voice profile
Identify which of the 8 page types this is, then score the existing voice across 5 axes (formality, emotion, complexity, humor, authority).
Score the headline and the full copy
Evaluate the primary headline on 5 criteria (clarity, specificity, relevance, differentiation, emotion), then score the full copy on the rubric's 5 dimensions (out of 50, doubled to 100).
Check persona alignment
If a persona or marketing-audit file exists, score the copy against 6 persona elements (out of 60); otherwise infer the persona from the page itself.
Generate the rewrites
Produce 5-10 alternative headlines via the PAS, AIDA, Before-After-Bridge, and 4U frameworks, optimize every CTA, then write at least 5 before/after pairs (headline, subhead, CTA, one body paragraph, meta description).
Render the report
Display a condensed terminal summary and write the full report to COPY-SUGGESTIONS.md (summary, voice profile, scoring rubric, value proposition canvas, recommendations, swipe file, priorities).
Rules
- Always detect the page type before scoring — each type carries its own copy priorities.
- Always document the existing voice profile and anchor generated copy to it, unless that tone is clearly ineffective.
- Use MARKETING-AUDIT.md or BUYER-PERSONA.md data when it exists instead of inferring the persona from the page alone.
- Produce at least 5 before/after pairs covering headline, subhead, CTA, one body paragraph, and the meta description.
- Write the full report to COPY-SUGGESTIONS.md in addition to the condensed terminal summary.
- Write every CTA in first person, with the value made explicit and risk reduced, never a bare "Submit".
Mistakes to avoid
- Applying the same copy structure to a homepage and a pricing page, when each page type has its own documented structure.
- Replacing the existing tone with a generic style without first scoring the brand voice or justifying the change.
- Giving vague feedback ("improve your headline") instead of one of the 4 named frameworks (PAS, AIDA, Before-After-Bridge, 4U).
- Ignoring a BUYER-PERSONA.md file already present in the project and inventing a generic persona instead.
- Failing to supply a concrete before/after when that's the basic unit of every recommendation this skill makes.
What this skill doesn't do
- Does not crawl the whole site — only the page(s) the user targets.
- Does not run a full CRO audit (forms, page speed, visual hierarchy, A/B tests) — that's /market-landing.
- Does not deliver a fast scorecard under 30 lines — that's /market-quick's role; this skill produces a full report.
- Does not build a brand voice profile from scratch — it analyzes the voice already present on the page.
Skill contents — 1 fichiers · 13.1 KB(SKILL.md)
Download for free