GEO Audit

Find out whether AI can read and cite your site

The GEO Audit analyses your site the way AI answer-engine crawlers see it and gives you a score from 0 to 100 with the list of fixes, ranked by impact.

Schematic illustration, not a real screenshot.

Who it is for

  • Anyone with a business site, an online shop or a blog who wants to appear in the answers of AI assistants such as ChatGPT, Perplexity or Gemini.
  • Agencies and consultants who need to show clients what to change and then prove the improvement.
  • Developers who want a repeatable check before and after every release.

The problem it solves

A site can do well on Google and stay invisible to AI

Answer engines pick the sources they can read, understand and attribute. A robots.txt that blocks an AI crawler, content that only appears with JavaScript, missing structured data or a missing «About us» page are enough to leave you out, and nobody tells you. The GEO Audit makes these signals visible and measurable.

How it works

  1. Enter the domain

    Type the site address or pick one you already added. There is nothing to install.

  2. The analysis runs in a queue

    The service reads robots.txt, llms.txt and the sitemap, then the homepage and up to five internal pages. Follow the progress on the page, which refreshes by itself.

  3. Read the score and the fixes

    You get a score from 0 to 100, the crawler matrix and the list of fixes, with the points each one can win back.

  4. Apply and run again

    Copy the ready-made code, fix the site and run the audit again: the comparison with the previous one shows how much you improved.

What you get

Score and band

A number from 0 to 100 with a band (excellent from 80, good from 60, partial from 40, poor below 40) and the breakdown: average of the pages minus the site penalties.

AI crawler matrix

Twelve AI crawlers, including GPTBot, ClaudeBot and PerplexityBot, showing whether your robots.txt lets them in or blocks them. Six are highlighted as decisive.

Site signals

llms.txt file, declared entity through structured data (Organization or WebSite), sitemap, «About us» and «Contact» pages, HTTPS and server response time.

Per-page checks

Amount of text, FAQ or HowTo data, author, lists and tables, question-style headings, date, an opening that answers right away, noindex, content that depends on JavaScript, canonical, heading hierarchy, valid schema and internal links.

Fixes ranked by impact

Each fix shows how many pages it affects and the points it can win back, with the score you would get by applying them all. The estimate is indicative.

Ready-to-copy code

A starter llms.txt, robots.txt rules to reopen blocked crawlers, Organization JSON-LD and a FAQPage skeleton, plus snippets for meta robots, canonical and sitemap. They only appear for the checks the site fails.

Prompt for your coding assistant

A text to paste into your AI development assistant (such as Claude Code) to apply the fixes, downloadable as Markdown.

History and comparison

Trend of the latest audits of the same site and comparison between two audits. If the calculation method changed between them, the page says so.

The searches you appear for

The top 25 real Search Console searches of the last 28 days (clicks, impressions, CTR, average position), the keywords you track in the Rank tracker and the AI prompts where the site was cited.

Competitor benchmark

The same audit on one or more competitors (up to five, depending on the plan), side by side, with the list of where they are ahead. It respects their robots.txt and opt-out requests.

Scheduled audits and alerts

Weekly or monthly audits with an alert if the score drops beyond the threshold you choose, if an important crawler gets blocked or if llms.txt disappears, plus a Monday summary. Nothing is sent before 9 a.m.

Citation test (optional)

Ask an AI engine 5 to 10 industry questions and see whether your site is cited. The cost is shown beforehand and a confirmation is required.

Illustrative example

A services website with a robots.txt that is too closed

A services company has a robots.txt that, because of an old plugin, blocks two AI crawlers, and has no llms.txt file. The first audit puts the block at the top of the fixes and suggests the «Allow» lines to add. After the change, running the audit again and comparing it with the previous one shows the higher score.

What it does not do and the limits to know

  • It measures whether the site is readable and citable: it does not guarantee being cited, because AI engines choose sources by their own criteria.
  • It analyses the homepage and up to five internal pages, as a crawler sees them: it does not run JavaScript and does not see pages behind a login.
  • The recoverable points are a linear estimate: the real score is rounded and never exceeds 100.
  • Search Console searches only appear if you connect Google and the domain is a verified property on your account; the data is about two days behind.
  • The citation test, when it does not use web search, shows what the AI model knows and not live citations: the page says so clearly.
  • Each user can run up to 10 audits per hour.

Frequently asked questions

What is the difference between GEO Audit and an SEO audit?

An SEO audit looks at how search engines crawl and evaluate the site. The GEO Audit checks related technical signals from the point of view of AI answer engines: whether their crawlers can get in, whether the content can be extracted and whether the site declares who it is. They complement each other.

Do I have to install anything on my site?

No. The analysis only reads public pages, as a crawler would. To apply the fixes you need to change the site or have whoever manages it do so.

Does running an audit have a cost?

The GEO audit is structural and does not use paid third-party services. A cost only appears for the optional citation test, which shows the estimate and asks for confirmation before starting.

Can I analyse a client’s or a competitor’s site?

You can analyse any public domain, but Search Console and the Rank tracker only show data for the sites you connected yourself. For competitors there is a dedicated comparison, which respects robots.txt and opt-out requests.

What is the llms.txt file?

It is a Markdown file published at /llms.txt that summarises the site and points to the main content for anyone reading it with a language model. It is an emerging standard and not every engine uses it: that is why it weighs little in the score (5 points out of 100).

How often should I run it again?

After every major change and at regular intervals: you can schedule weekly or monthly audits and get an alert if something gets worse.

Ready to try it on your site?

Create the free account: analyses that have a cost always show the estimate and only start after your confirmation.