Investigate Changes in AI Bot Traffic
Use Hardal agent, platform, category, and page reports to investigate a rise or fall in recorded AI bot visits.
A change in AI bot visits can reflect a different agent mix, a few repeatedly requested paths, a broader collection scope, or a change in site access. Start by locating the change before deciding what caused it.
Match the comparison
Compare complete periods with the same source, timezone, filters, and AI Focus setting. Record collection deployments, website releases, and changes to CDN coverage alongside the dates. A newly instrumented route can add recorded visits without a corresponding increase in underlying demand.
Check the metric definition in AI Visibility. AI-only visit counts and all-agent counts describe different populations; turning AI Focus off does not change every scorecard.
Locate the contributor
Start with Agents
Compare the visit trend and group it by Agent Name, then Agent Platform or Agent Category. Identify whether the change is broad or concentrated.
Inspect recent examples
Use Last Events to check the time, agent, and page path around the change. Treat these rows as examples rather than the entire reporting period.
Check the affected pages
Open Pages, find relevant paths, and expand their rows. Use the page-path chart to compare their trends.
Choose the next check
| Observed pattern | Next investigation |
|---|---|
| One agent accounts for most of the increase | Review its documented role and the paths it requested |
| One path accounts for most visits | Check the page's release, redirects, and request pattern |
| Recorded activity changes after a collection deployment | Compare integration scope and original request metadata |
| Several important paths lose recorded activity | Review collection coverage and site/CDN access evidence |
Use your server or CDN logs for response status, access challenges, or performance evidence. The AI Visibility page report does not by itself establish those causes or measure the infrastructure cost of a bot.
Report the observation and the hypothesis separately: “Agent X accounts for most of the recorded increase” is stronger than “our AI ranking improved.” Use agent roles for interpretation and a weekly readout to retain the comparison context.