Understand traffic referred by AI assistants

The question to answer

Someone can discover your product in an AI answer and then visit your website. Sometimes the app sends a recognisable referrer; sometimes it does not. The useful question is what you can observe reliably: which recorded AI sources brought visits, what those visitors did, and whether any payments can be matched to their journeys.

Before you start

A consent-aware tracking installation on the pages people visit.

A reporting period with actual referrals to investigate.

Goals or revenue attribution for outcomes beyond arrival.

1. Confirm ordinary tracking first

Open your website and check a consented visit in the dashboard. AI referral reporting uses the same browser tracking pipeline as other human traffic; it does not require sending conversations or prompts to Analytics.

2. Inspect Channel and Referrer

Look for the AI channel and recognised assistant names under Referrer. Use source icons and labels to inspect what was reported. There is no need to assume a separate AI card exists on every dashboard layout.

3. Filter one source

Select a reported source and inspect its pages, goals and journeys. Compare the same reporting period across sources, and keep the visitor counts visible so that small samples are obvious.

4. Add payment context when configured

Review attributed payments for the filtered source. Check that consent and checkout metadata permit a match. Where source information is absent, leave it unknown rather than inferring an assistant from a guess.

5. Separate the crawler question

Open the separate Crawlers dashboard if you want server-side AI bot requests. A crawler reading a page is not a person arriving from an answer, and neither report reveals the content of a user’s private conversation.

Worked example

These are different datasets with different collection paths. A rise in crawler requests and no human AI referrals is possible and does not by itself mean either report is broken.

Check your setup

You can find a known consented visit before investigating sources.

Channel and Referrer labels are based on reported data.

Bot requests are read in Crawlers, separately from visitors.

How to interpret the result

Some AI apps strip referrers or users later navigate directly.

The service cannot recover hidden prompts or count every product mention.

Referrer tags can be copied or supplied explicitly, so labels are observations rather than endorsements.

Your next action

Review one assistant’s observed visits alongside the pages they entered and the goals they completed, then improve those pages where the next step is unclear.