Free tool
Eight checks on whether ChatGPT, Claude, Perplexity, and Gemini can fetch your site, work out what your company is, and find something worth quoting. Every verdict comes with the evidence it was based on.
The list is short on purpose. Everything here is deterministic, provable from bytes we fetched, and fixable in an afternoon. Nothing here is scored by a model, so two runs on an unchanged site give the same answer.
GPTBot, ClaudeBot, PerplexityBot and Google-Extended have to reach your pages before any of them can quote you.
The file assistants read to orient themselves: what you are, and which of your pages answer what.
sameAs ties your name to an entity the engines already know. Without it you are a string, not a company.
Question-shaped markup on the pages that answer questions is the format an answer engine can lift verbatim.
These are the three strings an engine quotes back when it names you at all.
Most crawlers do not run your app. Text that only appears after hydration is text the engines never saw.
It is where assistants look to describe who you are, and the answer they give when they cannot find one is usually wrong.
Assistants skip products whose price they cannot find, and recommend the ones whose price they can.
It is not a site audit. There is a whole category of tool that will hand you 200 findings about heading hierarchy and image alt text, and if that is what you need, Screaming Frog does it better than we would.
It also cannot tell you the two things that most often cost a brand its citations, because neither lives on your own site: whether you have profiles on the directories and review sites the engines lean on, and whether anyone independent has written about your category with you in it. Those are off-page, and no crawler pointed at your domain will ever find them. Passing all eight of these and still being invisible is common, and it means the problem was never your markup.
Next
Whether an answer engine can fetch your pages, understand what your company is, and find something quotable. It is separate from whether it recommends you, which depends on what other sites say about your category. Readiness is the part you control entirely.
It fetches your homepage, your robots.txt, your llms.txt, and looks for an about and a pricing page, then runs eight deterministic checks over what came back. No AI is involved in grading, which is why every verdict comes with the evidence behind it. These are the same checks CueScout runs for paying customers.
No, and deliberately not. Title lengths, alt text, and readability scores are what Screaming Frog and Ahrefs already do well. This checks the narrower set of things that decide whether an AI engine can use your site as a source, which conventional SEO tools mostly do not look at.
Because assistants answering "what does X cost" need a number they can repeat. "Contact us for pricing" gives them nothing, so they tend to recommend the competitor whose price they can state. This is one of the more annoying findings to act on and one of the more reliable ones.
A plain markdown file at /llms.txt that tells assistants what your product is and which of your pages are worth reading. It is a convention rather than a standard, adoption is uneven, and no engine requires it. It costs an hour to write, which is why it is on the list.
Because readiness is necessary and not sufficient. Passing all eight means the engines can read you. Whether they recommend you depends on independent coverage: reviews, comparisons, and community threads where other people name you. That is the harder half, and the visibility checker is where you measure it.
No. The public checker runs against whatever URL you type and returns the result. Nothing is written to a database. If you want the checks tracked over time, that is the paid product, which runs against a site you have added to your own account.