Home/AI Visibility Check
Free tool · about 4 minutes · nothing stored
Do the assistants know your product exists?
Buyers increasingly start inside an assistant rather than a search box. If ChatGPT, Perplexity, Gemini, Claude and Copilot have never encountered your product, you are absent from that conversation and there is no ranking report that will tell you. This gives you the twelve prompts that find out, and a sheet to score the answers.
Everything happens in your browser. There are no API calls, no accounts and no cost, because you paste the prompts into the assistants yourself. That is deliberate: reading the actual answers with your own eyes tells you far more than a number would.
How to read what comes back
Two different things are being measured here and it helps to keep them apart.
Discovery: are you found by someone who does not know you exist?
The unbranded prompts, numbered one to six, are the ones that matter commercially. A buyer asking what the best tools in your category are has no idea who you are, and the assistant answers from what it has encountered about your category across the open web. Being absent from all six is the normal result for a product that has not done any of this work, and it is entirely fixable.
Accuracy: when it does know you, does it get you right?
The branded prompts, seven to twelve, test something different and often more alarming. Assistants routinely describe products with outdated pricing, features that were removed, or a positioning the company abandoned two years ago. This happens because the sources the model encountered are old, and the fix is to make current, structured, unambiguous information about yourself easy to find and hard to misread.
Run it in a fresh session, and run it more than once
Use a signed-out or temporary chat wherever you can, because personalisation and memory will quietly show you a friendlier picture than a stranger gets. Answers also vary between runs, so a single absence is weak evidence and three absences across three engines is not.
One check before you conclude anything. Open your
own robots.txt and confirm you are not blocking the retrieval crawlers, which are
the ones that decide whether citation is possible at all: OAI-SearchBot,
ChatGPT-User, PerplexityBot, Claude-SearchBot and
Bingbot. A blanket disallow, or a staging rule that reached production, makes
citation impossible regardless of anything else you do.
Worth being exact about one thing most articles on this get wrong:
Google-Extended governs Gemini training and grounding. It does not control whether
you appear in AI Overviews or AI Mode, both of which ride on ordinary Googlebot access. If you
have blocked Googlebot, that is the problem; blocking Google-Extended is a separate decision
with different consequences. The readiness check reads
your robots.txt and tells you which of these actually applies to you.
The bigger picture
AI visibility is one of eight dimensions
Being invisible to assistants is one way a launch underperforms. The full Launch Readiness Score covers the other seven as well, in about seven minutes.