ProblemUpdated 2026-09-21

AI crawlers fetch your pages but never cite them

It is normal, and it is the rule rather than the exception. On our site Google's verified crawler fetched 800 pages over two months and none appeared in a Google answer we recorded; Apple fetched 666, also with none. Fetching builds an index; citing is a separate decision made per question.

What we observed

Verified crawls and same-provider citations of the same pages, 25.07.2026 to 20.09.2026:

| Provider | Pages crawled | Pages cited | | --- | --- | --- | | Google | 800 | 0 | | OpenAI | 719 | 4 | | Apple | 666 | 0 | | Perplexity | 414 | 34 |

Full method in the study.

Why the numbers are so far apart

A crawler fetches to build or refresh what a system knows. A citation happens when a specific question is asked, the system retrieves, and the page wins a place in a short answer. Most pages are never the best source for any question actually asked — and a page can also be used without being linked.

What it does not mean

It does not mean the crawler "ignored" you, and zero recorded citations is not proof of zero use. We record the questions we ask; an answer to a question we never asked is not in the table.

What to do with it

Stop reading crawl counts as a scoreboard. Measure what the buyer sees — the product named and the product page linked — on the questions buyers ask. The ladder that keeps these apart is at /evidence.

Sources

Related

← All pages