Home / Tools / AI visibility checker
Free tool · GEOPaste your company's URL and in 20 seconds you'll know whether AI crawlers can get into your site, whether you have llms.txt and structured data, and whether your pages answer where AI reads. You get a score out of 100 and the three things to fix first. It doesn't check whether AI already cites you: that's measured by asking the engines, and it's the first step of the initial GEO audit.
Checks whether AI can read you. Not whether it cites you.
This check tells you whether AI can read you. Whether it already cites you, and who it cites instead, is the first measurement in the initial GEO audit: ten questions from your field, across the major AI engines.
Request your initial GEO auditEach one comes back as pass, partial or fail, with what we found and what to do about it. The points reflect how much each signal matters for a model to read you and pull out an answer.
We request your page with the User-Agent of GPTBot, OAI-SearchBot, ClaudeBot and PerplexityBot and compare the response with what a browser gets. Many CDNs and firewalls return 403 to these crawlers without any trace of it in robots.txt.Source: crawler documentation from OpenAI, Anthropic and Perplexity
We read your robots.txt and evaluate the rules for those four crawlers plus Google-Extended and CCBot, which only feed model training. Blocking the training crawlers costs half the points; blocking the ones that answer questions costs all of them.Source: RFC 9309, the robots.txt standard
We request /llms.txt. It passes if it exists, is plain text and contains at least one link. No engine has publicly confirmed using it for citations; we check it because it's a plain text file you can publish right away, and it's the proposed way to tell a model which pages matter.Source: the llmstxt.org proposal
We parse the page's JSON-LD blocks. It passes if you declare Organization plus a page entity: Article with author and date, WebPage, WebSite or Service. Organization on its own, or a block with a syntax error, counts as partial.Source: schema.org and Google's structured data guide
We take the first paragraph after the H1. It passes at 40 to 300 characters if it doesn't open with a question or with "in this article". This one is our own rule of thumb: the answer has to sit where the model reads first.Source: house rule, explained in SEO for AI: how to get ChatGPT to cite you
HTTPS, status 200, no noindex, a canonical that points to the URL itself and a title under 70 characters. Without these Google can't read you either, and GEO is built on top of SEO.Source: Google Search Central
The only way to know is to ask the engines. In the initial GEO audit we do it with ten fixed questions from your field, run across the major AI engines.
The checker reads structure: where the answer sits, what you declare, who can get in. It doesn't judge what you say or who links to you.
Some CDNs let GPTBot's User-Agent through and cut off OpenAI's official IP ranges instead. The check runs from our server, not from OpenAI's IPs, so it can't see that kind of block.
Many CDNs and firewalls block crawlers they identify as AI by default. It doesn't show up in robots.txt; you only see it by requesting the page with the bot's User-Agent, which is what this check does.
No. It means AI can read you and extract answers. Whether it cites you also depends on your authority and on answering better than the sources it already uses. That's measured with the 10-question protocol of the initial GEO audit.
Yes, any public URL. WordPress, Shopify or a custom build: the checks are the same.
Send you the report and, if you agree, information about GEO. You can unsubscribe from any email. The details are in the privacy policy.
To understand what a model looks at before citing a source, read SEO for AI: how to get ChatGPT to cite you and the comparison GEO vs SEO.
The checker tells you whether AI can read you. The initial GEO audit tells you whether it already cites you, who it cites instead, and which pages to change first. Itemized quote within 48 hours.