Why ChatGPT and Perplexity Might Never Cite Your Website

The uncomfortable starting point

A site can rank on page one of Google and still never get mentioned by ChatGPT or Perplexity for the exact same topic. These are related systems, not the same system, and the reasons a page gets cited by one don't automatically transfer to the other. Understanding why usually comes down to a short, specific list of genuinely common, checkable causes.

Reason one: the AI crawler is blocked, sometimes without anyone realizing it

GPTBot, ClaudeBot, Google-Extended, and PerplexityBot each have their own distinct, separately-named user-agent in robots.txt — completely separate from the crawlers that power ordinary search indexing. A site can be fully open to Googlebot for regular search while simultaneously blocking every AI crawler by name, sometimes as leftover boilerplate from a security plugin or a robots.txt copied from another site years ago, with nobody ever specifically deciding to make that trade-off. This is worth checking directly and first, since it's the single most common reason a page that's otherwise perfectly fine never gets considered at all.

Reason two: the content technically exists but isn't actually extractable

If the real content only appears after JavaScript runs — and the specific crawler in question doesn't execute it — what actually gets fetched is an empty shell page with no usable text. A page can look completely normal in a regular browser and still be functionally blank to a crawler that never executes the script that renders the real content. This is a common, genuinely invisible failure mode, since visiting the page yourself in a browser never reveals it.

Reason three: the content doesn't actually answer anything directly

AI systems tend to favor content that states a clear, direct answer plainly, then supports it — not content that builds toward a conclusion gradually across several paragraphs of setup. A page can be well-written, accurate, and genuinely useful to a human reader and still be a poor citation candidate simply because there's no single, clean, extractable statement a model can confidently lift and attribute.

Reason four: there's no genuine signal of authority or trustworthiness

Author information, publish and update dates, and citations to credible external sources all function as trust signals — for both traditional rankings and for a generative system trying to decide whether a source is reliable enough to actually cite by name. A page with none of these signals, even if the content itself happens to be accurate, gives a system less confidence to attribute a claim to it specifically.

Reason five: something more specific and popular already dominates that exact topic

Sometimes there's no real technical problem — a handful of large, well-established, frequently-cited sources already dominate a given topic so thoroughly that a smaller, newer site's genuinely good content simply doesn't surface as often, the same underlying dynamic that makes ranking on page one of Google difficult for a highly competitive keyword. Not every citation gap has a technical fix.

How to actually check where a site stands

Robots.txt is the fastest thing to rule out — open it directly and check for AI crawler user-agents by name, since this alone explains a meaningful share of real cases. From there, the most direct test is simply asking: query ChatGPT and Perplexity with the actual questions a page is meant to answer, and see what does and doesn't get cited. That's genuinely the only reliable way to know how a specific model behaves toward specific content right now — no indirect check fully substitutes for actually asking.

What actually to do about it

Fix the technical blockers first — robots.txt and JavaScript-dependent rendering — since those prevent consideration entirely, regardless of content quality. Then focus on making the content itself more directly citable: a clear, early, extractable answer to the actual question, genuine specificity over vague generalities, and real trust signals like clear authorship and dates. None of this guarantees a citation on any particular query, but it removes the barriers that make one impossible in the first place.