Skip to content
Praion
ANSWER · 5 MINUTES

What is AI referral traffic and how do you measure it?

Published

QUICK ANSWER

AI referral traffic is website traffic you can attribute to a click from an AI service. OpenAI, for example, adds utm_source=chatgpt.com to referral URLs from ChatGPT Search so that traffic can be identified in analytics. It does not measure all AI visibility: a mention without a click creates no referral visit, and a session recorded as (direct) / (none) does not prove the visit came from AI.

That distinction is critical for accurate measurement.

A business can be mentioned or used as a source in an AI answer without the user clicking a link. In that case there is AI visibility, but no referral visit.

The reverse is also true: a click from an AI service can create a measurable website visit, but that visit alone does not reveal the exact question the user asked, why they chose the link or whether the AI system framed the business as the “best” option.

What does AI referral traffic actually measure?

Consider a question such as:

“I run a small business and need a CRM that connects with my ecommerce store without being difficult to use. Which options should I consider?”

If an AI service shows a link to a relevant page and the user clicks it, analytics may record the resulting visit together with source information.

That is the visit you can attribute to the AI source, provided the attribution data survives the click.

It does not prove the AI “recommended” your business as the best option. The system may simply have surfaced the page as a relevant source or link.

For the page to be useful, its public information needs to be accurate: what the product or service offers, who it is for, which capabilities are available, which limitations apply, and how someone can learn more.

Do not assume an AI system will correctly fill in features, prices or capabilities that are not stated publicly.

How do you find it in Google Analytics?

In GA4, start with the Traffic acquisition report. Review Session source / medium, or the Session source and Session medium dimensions separately.

Source identifies the source associated with the session, while Medium categorises the type of traffic, with examples such as organic, referral, cpc or email.

One important GA4 detail: direct traffic is typically recorded with source (direct) and medium (none), shown as (direct) / (none). direct is not a medium.

For ChatGPT, OpenAI provides one confirmed tracking signal: referral URLs from ChatGPT Search include utm_source=chatgpt.com. That helps identify ChatGPT as the source in analytics.

Other AI services may also appear as identifiable sources when referral information survives the click. Do not assume every platform sends the same attribution data or appears in the same way.

If you want a combined AI-traffic view in GA4, build a filter or Exploration using only domains and sources you have already verified in your own data.

Are Google AI Overviews and AI Mode AI referral traffic?

Not in the same sense as traffic from an external AI service.

AI Overviews and AI Mode are features within Google Search, not external AI services sending referral traffic in the same way as a service such as ChatGPT.

In Search Console's main Performance report, data from AI Overviews and AI Mode is included in overall Web Search performance. A click on an external link in AI Mode counts as a click under Search Console's methodology.

As of August 31, 2026, Google says the separate Generative AI performance report for Search has been rolled out worldwide. The dedicated report currently isolates impressions from AI Overviews and AI Mode and can be analyzed by page, country, device, and date.

It does not currently provide a separate click metric for those generative AI features. Clicks remain included in the overall Web Search data in the main Performance report.

For Google, it is useful to separate two measurement layers:

  • Search Console: visibility and overall performance in Google Search, including generative AI features,
  • Google Analytics: what users do after they arrive on your website.

Clicks from AI Overviews or AI Mode should therefore not be grouped mechanically with referral traffic from an external AI service.

Is AI referral traffic “high intent”?

Not necessarily.

Intent depends on the question, landing page, research stage, and specific AI experience. A visitor may be close to a decision, or they may simply be researching a topic.

Google has said that, in its own data, people who click from search results pages with AI Overviews are more likely to spend more time on the sites they visit. That is an observation about Google Search. It is not proof of stronger purchase intent and should not be generalized to ChatGPT, Perplexity, or other AI sources.

Evaluate each source using your own data rather than a general assumption.

For every verified source, look at:

  • the number of sessions it drives,
  • the landing pages those visitors reach,
  • how many sessions are meaningfully engaged,
  • which key events are completed,
  • and, where measurement is reliable, what revenue or other business value is associated with those visits.

Our guide to measuring AEO ROI explains how to connect those actions to overall performance without judging a source only by visit volume.

What can referral traffic not tell you?

A core limitation is that source attribution is never perfect.

Referral information can be lost because of redirects, privacy settings, blocking tools, or other technical factors. A visit may then be recorded as (direct) / (none).

That does not make it “hidden AI traffic.” It simply means Analytics does not have a clear source for the session.

AI referral traffic also does not tell you:

  • how often the business was mentioned without a click,
  • the exact question that preceded each visit,
  • whether the mention was positive, neutral, or simply informational,
  • whether the user chose the page because the AI framed it as a recommendation,
  • how often the business appears overall in AI answers.

AI referral traffic is therefore a traffic-attribution metric, not a complete AI visibility metric or an overall “AI score.”

A fuller view requires a combination of Search Console, analytics, and — where a repeatable process exists — consistent checks of selected questions across AI systems.

If you want us to assess your measurement setup, we start with the sources being recorded today and what visitors do after the click.

No AI referral traffic metric guarantees a particular ranking, appearance, mention, or recommendation in Google, ChatGPT, Gemini, Perplexity, or any other AI system.

YOUR OWN QUESTION

Yours not answered here?

Send us your question. We answer personally by email — and if enough people ask the same thing, it becomes the next answer.

Want to measure your AI traffic the right way?

We start with the sources being recorded today and what visitors do after the click — not an arbitrary "AI score."