Bots tab

Every other part of AgentMention measures the output of AI systems. The Bots tab measures the input: which crawlers actually fetch your pages.

Two things are needed before it shows anything. The tab has to be enabled for your account — if it is not in your sidebar, it is not enabled. And your site has to be forwarding request data, which is the bot analytics integration.

What is counted

Requests to your site that come from identified crawlers, grouped into three categories:

• AI — crawlers operated by AI assistants and answer engines
• Search — traditional search-engine crawlers
• Other — everything else identified as automated

A visit is a single request. A crawler fetching forty pages registers forty visits.

The views

Summary. Total visits for the period by category, with the change against the previous period.

Timeline. Visits per day, so you can line up a spike with a publication or a site change. The shortest window is two days — a single-day bar chart says nothing about direction.

Bots table. One row per crawler platform, with its share of traffic. Clicking a platform drills into that crawler alone: its timeline and the pages it fetched.

Pages view. Which URLs crawlers request most. This is the practical half of the tab.

Reading it

AI crawler visits are a precondition, not an outcome. A crawler fetching your pages does not mean an assistant cites you; a crawler never fetching them means it cannot. Read this tab together with Citations & Sources: pages that are crawled but never cited have a content problem, pages that are never crawled have an access problem.

Check what they fetch. If crawlers spend their visits on your blog archive and never touch your product pages, that is what assistants will know you for.

Zero AI crawler traffic usually means access, not disinterest. Start with the Robots.txt audit — a disallow rule is the most common cause — and confirm the integration is reporting at all.

Spikes follow publication. A burst after a launch is normal. A sustained drop to zero for one platform, while others continue, points at a change on your side: a rule, a firewall, a CDN setting.

Caveats

• Only traffic your integration forwards is counted. If it covers one hostname or excludes static paths, the numbers describe that scope.
• Identification depends on how a crawler presents itself. Traffic that does not identify itself lands in "Other" or is not counted.
• This is your own server data. It is independent of assistant answers and available immediately once the integration runs — it does not wait for a collection cycle.

Related

• Bot analytics integration — setting it up
• Robots.txt audit — whether AI crawlers are allowed in
• Technical checks — whether a fetched page is readable