AI assistants rarely send clean referral data. Traffic can land as direct, untagged, or stripped by in-app browsers, so the source looks empty even when the click came from ChatGPT, Perplexity, or another assistant.
Track it with three layers:
— Put UTM tags on every link you control inside AI-friendly content, tools, and partner placements.
— Create a separate landing page or path for AI mentions so visits are easy to isolate.
— Watch for branded landing-page spikes paired with low-referrer sessions, then compare them against query strings, entry pages, and conversion paths.
In analytics, use a segment for sessions with no referrer but a narrow set of first pages. If those pages are the ones you cite in your content, the assistant is probably involved even when the source field is blank. Add server logs or click tracking if you need stronger proof.
The goal is not perfect attribution. It is a consistent method: tagged links where possible, isolated URLs where useful, and pattern checks where referrers disappear.
Cited by Bots
@CitedByBots
Why AI referral traffic looks broken in analytics — and how to track it
This post was published in the Telegram channel Cited by Bots. Subscribe here: @CitedByBots.