Official skill · Marketing

Landing page copy review

v1.0.2 · Updated 4 November 2026

Reviews landing page copy with 6 concrete tests (headline, so-what, opposite, proof, CTA, one job per section) and returns ranked line edits. Use it when A landing or home page is about to go live.

Works in Claude, ChatGPT, and coding agents. On Team and Business, the official skill in your pod updates itself when we improve it.

When to use it

• A landing or home page is about to go live
• A page gets traffic but few sign-ups and you suspect the words

What you give it

• The page copy or URL
• Target reader and the one action (optional)
• Traffic source (optional)

What you get

A verdict and up to 10 line edits ranked must-fix, should-fix, polish, each with the test it failed.

How it works

1. Run the standalone headline test Most visitors scan, so the top of the page has to carry the main point on its own (NN/g's inverted pyramid: most important information first). Read only the headline and subhead. Can a stranger say what it is, who it's for, and why it beats what they do now? If they can't say what it is, that is the top finding, above everything else.
2. Climb the so-what ladder on every benefit Ask "so what?" until the answer is something the reader feels (time, money, risk, status). Stop at the first rung a reader would care about; if the copy stops one rung lower ("AI-powered"), write the higher rung as the edit.
3. Apply the opposite test to adjectives (our default) If no competitor would claim the opposite ("fast", "easy", "powerful", "secure"), the word carries no information. Replace it with a number, a comparison, or a named example. If there is nothing concrete to say, cut the word.
4. Check proof sits next to each claim A claim needs a number, a customer name, a quote, a screenshot, or a demo within the same section. Proof three sections later doesn't count. Flag claims with no proof anywhere as "add proof or soften".
5. Check every CTA Starts with a verb, says what happens next ("Start free trial", "Get the checklist"), not "Submit" or "Learn more". NN/g found a bare "Learn more" leaves users unsure where it goes and loses meaning for screen-reader users. Our default: one primary CTA wording repeated across the page; a second, lower-commitment CTA is allowed once.
6. One job per section (our default) Name each section's job in 3 words (explain, prove, handle objection, ask). Two jobs in one section, or two sections with the same job: merge or cut. Check the biggest objection for this buyer (price, switching cost, trust) is answered somewhere above the last CTA.
7. Rank and write the edits Must-fix: blocks understanding or the action (headline, CTA, missing proof on the main claim). Should-fix: weakens persuasion. Polish: wording. Max 10 edits; each one quotes the current line and gives the replacement. Never invent proof: write "[add: customer number]" where proof is needed.

Share it with your team

Add it to an AI Skill Pod and everyone on the team, contractors included, uses the same version from their own AI. Read the best ways to share AI skills with a team

Add to my team's pod

Changelog

• v1.0.2 · 4 November 2026 · Example: the input says sections 2 and 4 and the pricing block were not shared, and the quote-wording check moves into the should-fix edit, so no polish item lacks a current line.
• v1.0.1 · 4 November 2026 · Example: the suggested headline uses only the input's facts (swaps in minutes, from the section 3 quote) instead of a feature the input never mentions.
• v1.0.0 · 4 November 2026 · First release.