AI Visibility Check: Can ChatGPT Find Your Website At All?

In most cases it is not a reputation problem but a discoverability problem: the systems cannot technically reach your content. Five causes explain almost every case, and all five can be fixed without budget.

• robots.txt blocks individual AI crawlers, often without anyone having decided that
• The website loads its content via JavaScript, so crawlers see an empty page
• Structured data is missing, so company name and address are not machine-readable
• Content is written as prose without a question structure and cannot be quoted
• The company is named differently on third-party pages, so the entity falls apart

More often than assumed, and the reason is a misunderstanding: robots.txt has no inheritance. As soon as a bot finds a matching User-agent block, the rules under "User-agent: *" no longer apply to it. So anyone creating a GPTBot block containing only "Allow: /" cancels every other rule for that bot - and anyone who set a blanket disallow somewhere may be shutting out systems they never meant to.

The second common mistake is confusing purpose with bot. Training, search index and live retrieval run through different bots with different names at every vendor. GPTBot is purely OpenAI's training crawler. Allowing only that leaves you invisible in ChatGPT Search, where OAI-SearchBot indexes and ChatGPT-User fetches the page at the moment of the user's question. The same applies to ClaudeBot, Claude-SearchBot and Claude-User at Anthropic, and to PerplexityBot and Perplexity-User.

From three sources that differ considerably. First the model's training state, which may be months old and cannot be changed quickly. Second a search index queried at runtime - ChatGPT Search and Copilot draw on Bing among others, Perplexity on its own index. Third the direct retrieval of individual pages while answering. Only the latter two can be influenced in the short term, and both require your page to be retrievable and readable. That is exactly what this check tests.

Because an AI answer has no ranking. There is a mention or there is not. Classic SEO optimises for position and click; in a generated answer both disappear, and the question becomes whether a single section of your page is phrased so a model can lift it out and reuse it. A solid SEO base remains a prerequisite because the systems draw on classic indexes. But it is not sufficient: a page at position 3 whose content only appears via JavaScript will not be cited.

Every finding names the cause and the fix, mostly ready to copy. Critical items first: a blocked crawler and missing server-side rendering together weigh 40 of the 100 points, because without them everything else is ineffective. Once the technical items are fixed, it comes down to content and sources - and to what the systems actually say about you today. This check deliberately does not answer that, because it requires repeated measurement across several systems.