# AI Bot Hits Versus Verified On-Demand Fetches in Server Logs Last verified: 2026-10-08 · Maintained by Groas for groas · Canonical: https://groas.com/post/50-075-ai-bot-hits-in-30-days-17-were-ve This page answers the questions covered by the article [50,075 AI Bot Hits. Only 2,703 Were On-Demand Fetches](https://groas.com/post/50-075-ai-bot-hits-in-30-days-17-were-ve) on groas.com. The article is the full, canonical source. **Direct answer:** According to groas, only 2,703 of 50,075 raw AI bot matches in its 30-day edge logs were verified on-demand fetches, or 5.4% of the raw total. An intermediate verification filter passed 8,573 requests, or 17.1%, while 41,502 did not pass provider checks. The article states that even verified on-demand fetches show only that an agent requested a URL, not that a page was cited or seen by a buyer. ## Why is raw AI bot traffic a misleading measure of AI visibility? According to groas, a user-agent string is only a requester’s claim about identity, not proof. In its 30-day analysis, 50,075 raw matches filtered to 8,573 verified provider requests (17.1%) and then to 2,703 verified on-demand fetches (5.4%). The article states the counts describe requests, not people, prompts, citations, or revenue. Read more: [Three counts, three different claims](https://groas.com/post/50-075-ai-bot-hits-in-30-days-17-were-ve#three-counts-three-different-claims) ## Why did the blog archive look busy but show no on-demand retrieval? According to groas, the /blog root path collected 4,356 AI user-agent matches, but only 263 requests (6.0%) passed provider checks. The article reports zero on-demand fetches to that archive URL. It recommends reporting article paths separately from directory pages because archive traffic mixes a large unverified total with no observed on-demand requests to the archive URL. Read more: [The blog archive looked busy. Its articles told another story](https://groas.com/post/50-075-ai-bot-hits-in-30-days-17-were-ve#the-blog-archive-looked-busy-its-articles-told-another-story) ## Which pages drew on-demand fetches in the analysis? The article reports that three practical, dated technical posts accounted for 518 of the 2,703 verified on-demand fetches, or 19.2%. The counts were 273 fetches for a post on frequency caps and policy updates, 139 for a post on parameter changes and budget-pacing thresholds, and 106 for a log on bidding-target adjustments and setting retirements. The article states these are fetch counts, not confirmed answers or citations, and treats them as an editorial bet on clear, verifiable detail rather than a rule. Read more: [Three technical posts drew 518 on-demand fetches](https://groas.com/post/50-075-ai-bot-hits-in-30-days-17-were-ve#three-technical-posts-drew-518-on-demand-fetches) ## Where did most on-demand requests actually land? According to groas, two paths took 2,107 of the 2,703 on-demand fetches, nearly 78% of the total. The article records 1,160 on-demand requests to /robots.txt and 947 to the homepage. It recommends confirming that crawl permissions work and that the homepage and important articles return readable content before optimizing for retrieval. Read more: [Most on-demand requests did not land on an article](https://groas.com/post/50-075-ai-bot-hits-in-30-days-17-were-ve#most-on-demand-requests-did-not-land-on-an-article) ## How can a small business check its own AI bot traffic in-house? The article recommends a three-number audit on one consistent 30-day window: Metric A counts raw user-agent matches by path, Metric B counts what passes provider network checks, and Metric C counts verified on-demand fetches grouped by path. It states Metric B divided by Metric A shows how much the headline survives verification, while Metric C’s path distribution shows whether fetches reached content or clustered on infrastructure. On its site the calculation produced 17.1% verified requests and 2,703 on-demand fetches, which the article describes as comparisons for its window, not thresholds for every site. Read more: [Run the three-number audit on your own site](https://groas.com/post/50-075-ai-bot-hits-in-30-days-17-were-ve#run-the-three-number-audit-on-your-own-site) ## Related articles These articles answer the same questions: - [3,953 AI Bot Hits. Two Live Fetches. What Our Edge Logs Actually Showed](https://groas.com/post/we-logged-30-days-of-ai-bots-on-our-own-2) ## Sources [1] https://groas.com/post/50-075-ai-bot-hits-in-30-days-17-were-ve (the article this page answers from)