How to Inspect Identifiable AI Referrals in Website Analytics

Website analytics can identify a visit from an AI service when the visit includes a useful source signal, such as a referring domain or a campaign tag in the landing-page URL. “Identifiable” describes the source reported for a visit. It does not reveal the visitor’s identity or every step that led them to the site.

A referrer is information a browser may send with a page request to show where the visitor came from. A campaign tag, such as a UTM parameter, is text added to a URL to label a visit. In Google Analytics 4 (GA4), check session source and landing-page data for these signals. Treat visits without a clear source as unknown. They may be plausible AI referrals, but the data does not prove that AI discovery led to them.

Find visits with a visible AI source

Start in GA4’s Traffic acquisition report and inspect Session source / medium. The report shows which source was associated with each visit, while user-acquisition reports show how users first found the site. Google explains the distinction in its acquisition report documentation.

Look for source values in your GA4 data that could identify an AI service. The spelling and medium depend on the information GA4 receives, so do not assume every site will show the same source labels. A row with an AI service’s domain is evidence that the visit arrived with that source information. It does not establish how the visitor first encountered the site.

Next, connect each source row to the page people entered. In the Landing page report, add Session source / medium as a secondary dimension, meaning an extra report field. Or use Landing page + query string with the source dimension in Traffic acquisition. The query string is the part of a URL after the question mark; it can show campaign parameters that were present when the visitor arrived. Google documents these options in its Landing page report instructions.

For a focused inspection:

1. Choose a date range, then open Traffic acquisition.
2. Review Session source / medium for values that could identify an AI service.
3. Compare those rows with their landing pages and query strings.
4. Check the same requested URLs and referrer information in website or CDN logs, if those logs retain them.

GA4 uses campaign parameters and the document referrer to assign a traffic source. A UTM tag can label a visit even when the browser provides no useful referrer. The GA4 documentation on campaigns and traffic sources describes these inputs.

Treat direct and unknown traffic as unattributed

GA4 labels a visit (direct) / (none) when it lacks a clear referral source. A visitor who types a URL or uses a bookmark can appear this way. A visit also appears this way when its source information is lost, for example, when a campaign tag is missing or a redirect removes it. Google lists these and other causes in its explanation of direct and none traffic.

Some direct visits may follow AI discovery. For example, someone may read a recommendation, leave without clicking, and later return by typing the address. The landing request does not record that earlier exposure. Many other paths also produce the same direct label, so assigning some or all direct traffic to AI would turn unknown visits into an unsupported estimate.

Use separate labels in reports: observed AI referral for a session with a source signal that identifies an AI service, plausible AI referral for a visit with suggestive context but no identifying signal, and unattributed visit when the available data supports no specific explanation. Keep the latter two out of measured AI-referral totals.

Check where attribution breaks

A browser can send an HTTP Referer header to tell a destination where a request came from. The header may leave out part of the referring address, or the browser may not send it. The HTTP Semantics standard describes the header and its limits. A visitor can therefore click through from an AI service while the website receives no clear AI referrer.

A redirect or URL-cleaning rule can remove parameters before analytics records the visit. Check the original link and follow its redirects. Then compare the final landing URL with GA4’s query-string data. A tag also records the label on the URL, not the visitor’s full discovery history. Someone can share or copy a tagged link, so the tag alone does not prove that the named service introduced them to the site.

Depending on its configuration and retention period, a server or CDN log may record the requested URL and Referer header. Compare those fields with GA4 when a source seems to be missing. A logged request is not necessarily a human visit, however, and a crawler request does not count as a referral from an AI answer.

Key Takeaways

• Count visible source signals: Use GA4’s session source and landing-page data to identify visits carrying an AI-related referrer or campaign tag.
• Keep unknown visits unknown: (direct) / (none) means GA4 lacks a clear source; it does not measure AI discovery.
• Trace lost attribution: Check redirects, final landing URLs, query strings, and available request logs when a source seems to disappear.
• Separate a label from a cause: A referrer or UTM tag reports information attached to the visit, not the complete path that led the person there.

What I'm building

Delegate tasks. Get software.

Give Vroni a GitHub issue, bug report, spec, or rough idea. It reads the repo, plans the change, writes code, runs checks, and works toward a review-ready pull request.

Take a look at vroni.com