AI referral traffic is website traffic that can be attributed to a click from an AI service.
Analytics platforms may identify the source through referral information or tracking parameters attached to the link. OpenAI, for example, says that referral URLs from ChatGPT Search automatically include utm_source=chatgpt.com, making inbound traffic from ChatGPT Search easier to identify.
That does not mean every AI-related visit will be classified cleanly as AI referral traffic. It also does not mean a visit recorded as (direct) / (none) came from AI.
In one sentence: AI referral traffic is a traffic-attribution metric — it measures visits that can be tied to clicks from AI services, not your full visibility in AI answers.
That distinction matters.
A business can be mentioned or used as a source in an AI answer without the user clicking a link. In that case, the business may have visibility in the answer, but there is no referral visit.
The reverse is also true: a click from an AI service can create a measurable website visit, but the visit alone does not reveal the exact question the user asked, why they chose that link, or whether the AI system presented 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 specific AI source.
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.
A business should not assume an AI system will fill in features, prices, or capabilities that are not stated publicly.
How do you find it in Google Analytics?
In GA4, a practical starting point is the Traffic acquisition report. Review Session source / medium, or the Session source and Session medium dimensions separately.
Source identifies the source associated with the start of the session. Medium describes the traffic medium, 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 AI platform sends the same information or will always appear in the same way.
If you want a combined AI-traffic view in GA4, build a filter or Exploration around domains and sources that you have already verified in your own data.
Are Google AI Overviews and AI Mode AI referral traffic?
Not in the same sense.
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”?
Do not assume that it is.
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.
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?
The first 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 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 and not 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.