How to measure AI referral traffic

AI referral traffic means a browser visit that follows a link from an AI service, such as ChatGPT, Perplexity, Claude or Copilot. It is different from an AI bot fetching a page or an AI answer mentioning your brand. Website analytics measures visits and actions; it does not count every answer in which your site appears.

Three different signals

Keep these measures separate. Adding them together as “AI traffic” mixes people, automated requests and answer exposure.

1. Find the sources that actually arrive

In Seean, open your website’s Overview and inspect the source breakdown. Look for hostnames actually recorded in your data. Depending on how a service opens links, these may include chatgpt.com, perplexity.ai, claude.ai or copilot.microsoft.com. This is an example list, not a guarantee that every service sends a referrer.

Select a source row to filter the dashboard. Inspect which landing pages receive visits and which actions follow. A service can omit the referrer, or a person can copy a URL into a new tab. Those visits cannot reliably be assigned to an AI service from the referrer alone.

2. Measure useful actions

Choose a small number of outcomes before evaluating a source: a completed signup, a requested demo or a verified purchase. Send a custom event only when the action succeeds. For example, after a signup has completed and the Seean tracker is available:

window.seean?.track("signup");

Do not send the event merely because a person opened the signup page. Build a funnel using the landing page and the completed event to see the difference between an interested visit and an outcome. See custom events.

3. Tag links you control

For a link in a resource you publish, use stable UTM values that describe the actual distribution. For example, a newsletter announcing an AI referral guide could use:

https://example.com/guides/ai-referrals/?utm_source=newsletter&utm_medium=email&utm_campaign=ai_referral_guide

Tagging a newsletter as chatgpt would manufacture an AI source, not measure one. You generally cannot control the links generated in another service’s answers. Treat observed referrers and your own campaign tags as different sources of evidence.

4. Connect revenue carefully

If a visit leads to a sale, your server can send the verified order and attribution details to Seean. Use the order integration to report the fields your checkout supplies. Missing fields stay unattributed; the dashboard does not infer an AI source or reconstruct the full journey.

Compare AI-referred visits, successful events and orders over a consistent date range. With only a few visits, inspect the actual journeys and wait for more evidence before changing your acquisition budget.

5. Review crawling separately

Use your hosting or CDN logs to check crawler requests to public pages, robots.txt and your sitemap. A JavaScript analytics tracker is not a complete crawler log: many crawlers do not execute it. A user-agent name can also be copied, so use the provider’s published verification guidance before treating a request as authentic.

OpenAI documents OAI-SearchBot and its other agents. Perplexity publishes crawler purposes and IP ranges. Google explains that AI search features use the same underlying SEO foundations.

A weekly review that stays useful

Record the date range, observed source hostnames, landing pages, successful events and attributable orders. Compare the previous period using the same definitions. Keep a separate citation sample with the exact question, service, date and cited URL. A sampled answer describes that observation, not your universal position in AI search.

Explore Seean’s analytics · Install the tracker · Try the demo