The guide · install in Claude Code
Get cited by AI search.
Google AI Mode just rewrote search for the first time in 25 years. The blue links you optimised for aren't where the answer lives anymore — and the AI summary only cites 3–5 sources. This skill audits your site against Google's own AI-optimisation guidelines and tells you exactly what to fix: a scored report, ranked fixes, no fluff.
The full setup
🔒 Drop your email to reveal the setup
Get the full setup, free.
Drop your email and the skill plus the full 4-step setup unlock instantly, along with every other guide and resource in the library. One email a week, unsubscribe anytime.
What you get
100-point AI search score
Crawlability, structured data, content shape, citation signals, freshness — every category Google's AI-optimisation guide names, scored 0–100, with the lowest-cost fixes ranked first.Entity & Offer extractability (+15)
A second score on top of the base 100. Does AI actually know who runs the site and what they sell? Person/Org schema, sameAs links, named offers, and originality signals — all checked.Exact fixes, not vibes
Every deduction comes with the file/line to fix, a before/after snippet, and the impact tier. You can hand the report to your dev and they'll know what to do.The stack — what runs it
Free, local, no third-party APIs:
Claude CodeRuns the skill. The free plan is fine — the audit is local.PlaywrightAuto-installs on first run. Scans the live URL in a headless browser.100-pt base GEOCrawlability, structured data, content shape, citation signals, freshness.Entity & Offer (15 pt)A layer on top — does AI actually know who you are and what you sell.
Setup — 4 steps
If you've used Claude Code before, this is straightforward. If you haven't, start with the content brain guide to get Claude Code installed first, then come back here.
01 · Download the skill
Grab the geo-audit-skill.zip from the resource library. It contains SKILL.md, README.md, the scoring rubric, and the audit script — plus a geo-playbook.pdf that explains why each fix matters.
02 · Drop it into your Claude Code skills folder
Unzip the file, then move the geo-audit/ folder into your Claude Code skills directory:
Restart Claude Code so it picks up the new skill. You can confirm it loaded with /skills list.
03 · Run your first audit
In any Claude Code chat, just ask:
Claude will install Playwright if needed (one-time, ~30s), open the page in a headless browser, extract every signal, score it against the rubric, and write a report file in your project root.
04 · Read the report → fix the top items first
The report ranks every issue by impact tier. Three things to do with it:
- Fix tier-1 issues first — they unlock the biggest score jumps. Usually H1 structure, schema, and crawler access.
- Re-run after each batch — verify the score actually moved. If it didn't, the fix isn't doing what you think.
- Use the playbook PDF as the manual — it explains why each fix matters and the AI-engine behaviour behind it.
Pro move: run the audit on a competitor ranking ahead of you in AI Overviews. The delta between your scores tells you exactly where to invest.
Three ways to use it
Pre-launch audit
Run it on a staging URL before you ship. Fix the top-impact gaps first, then re-run until you're above 80.Competitor scan
Point it at the page that's outranking you in AI Overviews. See exactly what they did and copy the structure — not the content.Quarterly health check
Rerun every quarter. AI-search guidelines drift; your score will too. The playbook tells you what's new.Things people ask first
Why is this different from regular SEO?
Regular SEO optimises for the 10 blue links. AI search optimises for being cited as a source in the AI summary at the top. Different signals matter: structured Q&A, entity definition, named offers, freshness, and crawlability for Googlebot and AI crawlers (GPTBot, ClaudeBot, etc). The playbook covers all 26 fixes.
Do I need to be technical?
No. The skill runs end-to-end in Claude Code — drop in a URL, get a scored report. Most fixes are content/structure changes (rewording H1s as questions, adding schema, etc.) that you or your dev can do in an afternoon.
Does the audit cost anything to run?
Nothing on top of your existing Claude plan. The skill uses Playwright locally to scan the page; no API calls, no third-party services. Runs in ~30 seconds per URL.
What does the 15-pt layer actually check?
Whether AI can identify the person/brand behind the site and surface their offers. Person/Organisation schema, sameAs identity links, named offers, entity disambiguation, and bylined original voice. Google's 100-pt framework doesn't check any of that — but it's what makes you citable as a source vs. just another blog post.
Original guide by Artem Novitckii. More from him on Instagram, in the Skool community, and across his resource library.