Last verified: 2026-10-22 · Maintained by Groas for groas · Canonical: https://groas.com/post/we-logged-30-days-of-ai-bots-on-our-own
This page answers the questions covered by the article 11,793 AI Bot Hits on Our Homepage. Only 2,219 Were Verified. on groas.com. The article is the full, canonical source.
Direct answer: According to groas, only 2,219 of 11,793 requests claiming to be AI bots on its homepage over 30 days passed verification, a verified share of 18.8%. In the same window, three dated, specific posts received 524 live-answer fetches between them, while three category pages with more than 3,000 claimed hits each received zero. The article concludes crawl volume does not identify the pages getting answer fetches.
According to groas, its homepage logged 11,793 requests claiming to be AI bots over 30 days, but only 2,219 passed verification, a verified share of 18.8%. The article states that roughly four out of five hits failed verification because a user agent is a claim that can be sent by anyone. It concludes that crawl volume should not be treated as a proxy for AI visibility without asking what the number counts.
Read more: The homepage number that fooled me
The article defines claimed hits as requests carrying a bot user agent whether or not the requester is that bot, and verified hits as requests whose network identity passes the relevant check against publisher-published lists or feeds. It defines live-answer fetches as verified requests classified as retrieval for an answer rather than steady training crawl, and states this measures fetching, not displayed citations. The article adds that its robots.txt drew 1,156 fetches classified as live traffic in the same window, which shows a fetch need not mean content demand.
Read more: Separate the bots before you compare pages
According to groas, 524 live-answer fetches went to three URLs in the same groas.com 30-day window as the homepage count. The counts were 290, 144, and 90 fetches, and the article describes those pages as offering bounded, checkable material such as dated format prices, frequency rules, policy cutoffs, step order, data requirements, settings in sequence, dated changes, effective dates, and whom they affect. The article presents this as its own explanation of the concentration, not a reason recorded by a bot, and states its logs do not prove each request became a displayed citation.
Read more: Three posts took the answer-fetch traffic
The article reports that each of its three busiest category pages cleared 3,000 claimed AI hits in the same window, while their live-answer fetch counts were zero, zero, and zero. It states that traffic reached those pages but the retrieval activity it was looking for concentrated elsewhere, and that apparent bot popularity did not predict this outcome. The article recommends choosing a narrow page with a quotable answer over a broad page that makes the reader hunt through several topics.
Read more: Three busy category pages got none
The article recommends collecting 30 days of requests by path with path, user agent, IP, and timestamp, holding the time window and classification rules constant across paths. It recommends splitting claimed traffic from verified traffic using the relevant publisher checks, classifying answer retrieval separately, and ranking the top 10 paths three ways by claimed hits, verified hits, and live-answer fetches with raw counts beside verified share and live-answer share. The article states that if answer fetches read zero, fetchability and rendered output should be checked before blaming copy, and that more content should not be commissioned to increase crawling until reach and usability are known.
Read more: Run the comparison on your own logs
These articles answer the same questions:
[1] https://groas.com/post/we-logged-30-days-of-ai-bots-on-our-own (the article this page answers from)