FREE GEO AUDIT
See what ChatGPT keeps of your page
Paste a URL. We read it like OAI-SearchBot, rebuild the title and 200-character snippet ChatGPT stores, look it up in ChatGPT and list fixes.
What the audit shows
-
The record ChatGPT keeps
The URL, the title and the first 200 characters of text, rebuilt from your HTML the way ChatGPT's index stores them. We also ask ChatGPT for your page and show what it returns.
-
Checks, each with a fix
Server rendering, page size, title length, what fills your first 200 characters, robots.txt for OAI-SearchBot and noindex.
-
A rewrite you can paste
A new title and opening, written from facts on your page, plus three pages that each answer one exact question.
The seven rules we check
Each rule follows from how ChatGPT's search index stores and cites pages.
| Rule | What we check | Why |
|---|---|---|
| Server-render everything, under 4 MB | Whether your text is in the HTML before any JavaScript runs, and how big the page is. | OAI-SearchBot does not run JavaScript. Text that only appears after scripts run never reaches the index, and very large pages fail. |
| The first 200 characters are your pitch | What fills the first 200 characters: menus, cookie text, images and slogans, or your H1 and first paragraph. | The index stores three things per page: the URL, the title and about 200 characters of text. Meta descriptions and JSON-LD are not stored. |
| Max out the title | Title length: 120 to 230 characters that say what the page is, for whom and where. | Google cuts titles at about 60 characters. ChatGPT's index keeps long titles in full. |
| One page per exact question | Three page ideas from your own content, each for one exact question. | People ask ChatGPT complete questions. A page about one question can answer it in its title and first 200 characters; a homepage cannot do that for every question. |
| Specs in plain text | Whether the server sends real text, not an empty app shell. | When ChatGPT reads a page in depth, it uses a text copy. Limits, prices and FAQs in plain HTML text and tables get quoted directly. |
| Links, not sitemaps | Whether the page is in your sitemap with a valid lastmod date. It does not count toward the score. | In 30 days of our logs, OAI-SearchBot never asked for a sitemap. It finds pages through links, so link every page from your menu or footer. The sitemap still helps Google and Bing. |
| Skip llms.txt | Whether you have one. It does not count toward the score. | No major AI search engine uses llms.txt to find or cite pages. Put your facts in your HTML. |
Questions
Is this what ChatGPT really stored for my page?
The record is a prediction from your HTML, built with the rules ChatGPT's index follows. We also search ChatGPT for your page's title. When it returns your page, the result shows what it returned and when we saw it.
Why does ChatGPT ignore my meta description?
Its index keeps three things per page: the URL, the title and about 200 characters of page text. Meta descriptions, JSON-LD and llms.txt are not part of that record.
Does a high score mean ChatGPT will cite my page?
No. The score ranks what to fix first. Whether ChatGPT cites you also depends on the question and on the other pages it finds.
What do you fetch from my site?
The page itself without running JavaScript, one request with the OAI-SearchBot user agent, and your robots.txt, sitemap and llms.txt. We do not log in or fill in forms.
Who can see my result?
Anyone you share the link with. Result pages are kept out of search engines.
Is it free?
Yes. No account, no credit card. You can run 5 audits a day.
Check a page in about a minute
Free, no account.