filtering-bot-traffic

Add an event property filter with key $virt_is_bot, operator exact, and value false to any TrendsQuery, FunnelsQuery, or insight. Visitor, session, and pageview counts then reflect human traffic only without changing stored data.

Break a pageview trend down by the $virt_traffic_type property to split traffic into Regular, AI Agent, Bot, and Automation buckets. To see which specific bots visit, filter $virt_is_bot to true and break down by $virt_bot_name or $virt_bot_operator.

Both use the same user-agent classification. Virtual properties like $virt_is_bot read the user agent automatically and work in the insight builder and filters, while isLikelyBot(ua) takes an explicit user agent argument for hand-written SQL queries.

Most crawlers and AI agents never execute JavaScript, so posthog-js never captures a $pageview for them. You need server-side capture that forwards access logs as $http_log events with $raw_user_agent, via the Vercel logs source, an edge worker, or the capture API.

Yes, classification is computed at query time from the event's $raw_user_agent or $user_agent, so it works on any historical event that captured a user agent. Events without a user agent classify as Automation with category no_user_agent.

Detection is a user-agent heuristic, so treat it as best-effort rather than ground truth. Some bots spoof real browser user agents and some legitimate tools use bot-like ones, and the detected-bot list evolves over time.