agentleFS
ChykalophiaChykalophia/cro-research-reporting/skills/conversion-copywriting/SKILL.md

conversion-copywriting

Skill1 starsChanged 29 days agoMITView on GitHub

This skill should be used when writing or rewriting customer-facing copy: product pages, landing pages, hero/headline sections, CTAs and button microcopy, email bodies, and ad copy. Trigger phrases include "write copy for", "rewrite this", "make this punchier/clearer", "headline for", "CTA for", "hero section", "product description", "ad copy", or "this copy is weak". Use it whenever copy quality affects conversion, even if the user does not name a framework.

2 other projects use a version of this.

Adopted by

Most recently taken up by

The file

We split this skill into the rules it gives an agent

7 rules. About 0 are common, 4 are distinctive, and 3 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈4

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 3

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    Write copy that earns the next scroll, the next click, the sale.

    line 17
  2. Very specific:

    Voice: confident, warm, non-hype: No corporate-speak, no exclamation-point inflation.

    line 24
  3. Distinctive:

    No one, human or model, reliably predicts the winner: GPT-4 picked the winning Upworthy headline 38–40% of the time vs 33% by chance (Ye et …

    line 29
  4. Very specific:

    Every headline is checked against Useful, Urgent, Unique, Ultra-specific.

    line 41
  5. Very specific:

    Pull customer language first: Use the voc-research language bank (or mine reviews/survey answers) for the exact phrases customers use for …

    line 45
  6. Distinctive:

    Add proof (social proof, specifics) and risk-reversal near the CTA.

    line 49
  7. Distinctive:

    Don't invent proof, testimonials, or stats.

    line 62

≈ “Common” is an estimate: at least 1 in 5 of the 350,495 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

Conversion Copywriting

Write copy that earns the next scroll, the next click, the sale. Clarity over cleverness; desire first, proof second; every word earns its place.

Principles

  1. Clarity is the conversion lever. A reader who has to decode a sentence leaves. Simple words, short sentences, one idea per line.
  2. Lead with desire/benefit, justify with proof. Open on the outcome the reader wants; bring features, specs, and credentials in as the rational backup.
  3. Specific beats vague. Numbers, names, concrete detail. "Recovered $340K in 90 days" beats "drives results."
  4. One job per asset. A page, an email, an ad each has a single primary action. Cut anything that competes with it.
  5. Voice: confident, warm, non-hype. No corporate-speak, no exclamation-point inflation. (Chykalophia voice rules, including no em dashes, in references/voice-and-headlines.md.)

What the evidence says about AI and copy

  • AI-written copy and creative can match or beat human work in live tests: fully AI-generated display ads lifted CTR up to 19% over expert ads, while AI edits to human ads did not help (Lee, Ghose, Todri & Adamopoulos, NYU Stern 2025).
  • Telling people an ad is AI-generated cut CTR by 31.5% (same NYU study). Don't add "AI-written" framing to the copy itself; follow any platform or legal disclosure rules that apply.
  • No one, human or model, reliably predicts the winner: GPT-4 picked the winning Upworthy headline 38–40% of the time vs 33% by chance (Ye et al., Marketing Science 2025). Frameworks and the 4 U's are a quality floor, not a predictor.
  • So the default output is a variant set for testing, written from customers' own words.

Pick the framework to the format (references/copy-frameworks.md)

  • PDP / product page: PAS or FAB for the body, 4 U's headline, proof + risk-reversal blocks.
  • Landing page (one offer): BAB or 4 Ps, or AIDA for the full-page arc.
  • Email: PAS hook, one job per email, 4 Ps for offer emails.
  • Ads: PAS or AIDA, ruthless brevity.
  • Homepage / brand: StoryBrand SB7 (customer is the hero, brand is the guide).
  • Long-form sales / consulting narrative: PASTOR.

Default headline test: the 4 U's

Every headline is checked against Useful, Urgent, Unique, Ultra-specific. Hit at least three; "useful + ultra-specific" is non-negotiable. Details and templated families in references/voice-and-headlines.md.

Workflow

  1. Identify the reader's awareness stage, the one job, and the single most compelling outcome.
  2. Pull customer language first. Use the voc-research language bank (or mine reviews/survey answers) for the exact phrases customers use for the problem, the outcome and their objections. Copy built from these phrases is the default; invented phrasing is the fallback.
  3. Choose the framework for the format; draft to it.
  4. Write 3–5 variants per key element (headline, subhead, CTA, key benefit block). Make each variant differ on one lever (e.g. outcome vs objection-handling vs proof-led), and label the hypothesis it tests and the customer evidence it draws on. Run the 4 U's as a floor on each. Hand the set to experiment-design for sizing; if traffic can't support a test, recommend the variant with the strongest customer evidence and say that is the reason.
  5. Apply the voice rules (cut hedging, buzzwords, em dashes, exclamation points; make verbs strong and nouns specific).
  6. Add proof (social proof, specifics) and risk-reversal near the CTA; end with one clear action.

Reference map

  • references/copy-frameworks.md — AIDA, PAS, BAB, FAB, 4 Ps, PASTOR, StoryBrand SB7, each with a worked example and a "use when" table.
  • references/voice-and-headlines.md — the 4 U's, headline families + CTA formula, microcopy, and the Chykalophia voice/banned-phrase rules.

Engagement context

At the start of a task, check for the engagement brief (default ./cro-engagement.md, created by the getting-started skill / /kickoff). If it exists, read it first and write to the client, ICP, price tier, and especially the brand voice/banned-words it records. If it's missing and the work is more than a one-off, offer to capture one via getting-started.

Related skills

Pair with voc-research (the language bank), experiment-design (testing variants), ecommerce-cro-audit (where copy is a lever and persuasion models live), report-information-design (for on-page structure), and outbound-and-positioning (for cold email + the positioning that copy expresses).

Guardrails

Don't invent proof, testimonials, or stats. Don't claim one variant "will win"; say which evidence each draws on. Don't manufacture false urgency. Keep claims substantiated; if you don't have the number, ask for it rather than fabricating.

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.