AI Referral Traffic

AI referral traffic is website traffic generated when someone clicks a link in an AI assistant's response.

For example, a person might ask ChatGPT to recommend local SEO software, receive a link to a relevant website and click through to learn more.

Similar visits can originate from Perplexity, Claude, Gemini, Microsoft Copilot and other AI services.

For businesses, the challenge is measuring those visits accurately.

Google Analytics 4 (GA4) now has a dedicated AI Assistant channel, but it cannot identify every visit influenced by an AI-generated answer.

Some visits arrive without usable referral information and may appear as direct traffic.

AI referral traffic is measurable, but the numbers in your analytics reports shouldn't be treated as a complete count of every customer who discovered you through AI.

How Does AI Referral Traffic Work?

AI assistants can link to websites when responding to questions, explaining topics or recommending resources.

For example, a dental patient might ask an assistant about a treatment and click a link to a local clinic.

If the visit carries identifiable referral information, an analytics platform may associate it with the AI service.

If the visitor opens the website without that information, the original source may be unclear.

A customer might also read an AI answer, remember a business name and later search for the business on Google.

That customer was influenced by AI, but the eventual website session may be attributed to organic search rather than the assistant.

This is the central limitation of AI traffic reporting: referral traffic and AI-influenced discovery aren't the same measurement.

How Does GA4 Track AI Referral Traffic?

In May 2026, Google introduced an AI Assistant channel in the default channel grouping for Google Analytics 4.

GA4 can automatically classify traffic from recognised AI assistant referrers.

This makes it easier to distinguish some AI-driven visits from general referral traffic without creating a custom channel first.

However, Google's publicly documented examples do not establish that every assistant or referral domain is recognised.

Traffic from an AI platform may still appear in another channel depending on the source information GA4 receives.

How to Find AI Assistant Traffic in GA4

Open GA4 and navigate to:

Reports → Acquisition → Traffic acquisition

Use Session default channel group as the primary dimension and look for AI Assistant.

You can also inspect session source and medium to identify the platforms sending visitors.

For example, relevant sources may include domains associated with ChatGPT, Perplexity, Claude or Microsoft Copilot.

Check the actual source values in your property rather than assuming every platform uses one consistent domain.

Why Does Some AI Traffic Appear as Direct?

Website analytics relies partly on information supplied during navigation.

That information isn't always available.

Missing Referrer Information

Some application-to-browser transitions, privacy configurations and other navigation paths do not preserve a usable referrer.

When GA4 has no identifiable source or campaign information, it may classify a session as Direct.

However, a Direct session isn't necessarily an AI referral.

It could come from a bookmark, manually entered URL, untagged link or another unidentifiable source.

Inconsistent Campaign Parameters

Some AI services may append tracking parameters to outbound links.

Others may not.

These differences affect how analytics platforms classify visits.

Customers Who Return Later

A person might discover a business through an AI answer but return through a branded Google search the following day.

The later visit does not necessarily preserve the original AI discovery source.

This is one reason AI-assisted customer journeys are difficult to reconstruct from web analytics alone.

An increase in Direct traffic is not proof that AI referrals have increased. It is a reason to investigate, not a figure that can be reassigned to AI.

How to Create a Custom AI Channel Group in GA4

If the default AI Assistant channel doesn't capture sources you want to monitor, GA4 supports custom channel groups.

Google's own documentation provides an example for tracking AI assistant traffic.

To create one:

1. Open Admin → Data display → Channel groups.
2. Create a custom channel group based on an existing group.
3. Add a channel named AI Assistants.
4. Create rules using relevant session source or medium values.
5. Include the AI assistant domains supported by your observed traffic.
6. Position the new channel above general Referral and other overlapping channels.
7. Save the group and test it in an acquisition report.

Ordering is important.

GA4 evaluates custom channel group rules in sequence and assigns traffic to the first matching channel.

If general Referral matches first, the same sessions may never reach your new AI channel.

Google's documentation includes an example regular expression for matching multiple AI assistants. Use it as a starting point, then check the source values and results against your own property.

Can Custom Channel Groups Analyse Historical Traffic?

Yes.

Google confirms that custom channel groups can be applied retroactively to existing report and exploration data, where the underlying source information is available.

This means you can create a group today and use it to analyse earlier qualifying sessions.

However, a channel rule cannot reconstruct missing referrers or identify the true source of visits originally recorded without usable attribution information.

There is also an important distinction between custom channel groups used in reports and the primary channel group, whose changes affect future reporting differently.

Can You Track AI Overview Traffic Separately?

Google's AI Overviews and AI Mode require different reporting considerations.

A click from an AI Overview in Google Search generally belongs to Google organic search traffic, not the third-party AI Assistant referral channel.

In June 2026, Google introduced dedicated Generative AI performance reports in Search Console.

By August 31, 2026, Google said these reports had rolled out worldwide.

They provide visibility data for generative AI features, including impressions and the pages shown.

However, the dedicated generative AI reports do not provide a separate click metric.

Google Search Console's overall Search performance reporting continues to include eligible clicks and impressions from AI Overviews and AI Mode.

This means you should distinguish between:

• Third-party AI referrals: Identifiable website visits from assistants such as ChatGPT or Perplexity.
• Google AI feature visibility: Appearances in Google's generative AI search experiences.
• Google organic search traffic: Website clicks from Google Search, including eligible AI-feature interactions.

These measurements describe different aspects of discovery and shouldn't be combined into one unexplained total.

How Should Businesses Report AI Referral Traffic?

Start with the identifiable sessions in GA4.

Review which assistants send visits, which landing pages receive those visits and whether visitors complete meaningful actions.

For a local business, those actions might include appointment enquiries, form submissions or telephone link clicks.

Compare the findings with LLM visibility checks, which examine whether and how the business appears in AI-generated responses.

However, visibility and referral traffic aren't interchangeable.

A business can be mentioned in an AI answer without receiving a website visit.

Likewise, an AI referral session doesn't automatically represent a qualified enquiry.

Where practical, ask customers how they found the business. Responses mentioning AI assistants can provide additional context, although self-reported attribution has limitations.

Ranksphere's White-Label Reporting helps agencies present local SEO performance in client-ready reports.

Identifiable AI referrals can be included as supporting context alongside rankings, Business Profile interactions and enquiries when the relevant data is available.

Common AI Referral Traffic Mistakes

• Assuming GA4 identifies every AI visit. Some sources may be unrecognised or arrive without usable referral data.
• Treating all Direct traffic as AI traffic. There is no reliable basis for that assumption.
• Ignoring custom channel rule order. Overlapping channels can capture sessions before the AI rule is evaluated.
• Assuming custom channel groups cannot analyse historical data. GA4 supports retroactive application in reports.
• Confusing Google AI Overviews with third-party AI referrals. They are measured differently.
• Treating impressions as clicks. Search Console's dedicated generative AI visibility reports do not provide separate click counts.
• Comparing conversion benchmarks without checking methodology. Industries and attribution methods differ considerably.

AI Referral Traffic Best Practices

Check the default AI Assistant channel before building custom tracking.

Review session source values to identify additional AI platforms.

Use a custom channel group where it adds useful reporting detail, and verify that its rules are correctly ordered.

Monitor referral sessions, landing pages and meaningful customer actions.

Keep Google AI feature impressions separate from identifiable third-party AI visits.

Compare trends consistently and explain the attribution limitations to clients.

Example of AI Referral Tracking for a Veterinary Practice

Consider a hypothetical veterinary practice whose agency reports very little traffic from AI assistants.

The practice manager has nevertheless heard patients mention discovering the clinic through ChatGPT.

What the Agency Finds

The team checks GA4's AI Assistant channel and notices that some sessions from AI-related sources are reported under Referral.

It also discovers that an existing custom channel group places the general Referral rule above the AI-specific rule.

This prevents some identifiable visits from appearing in the intended custom channel.

What the Agency Changes

The agency updates the rule order and tests the results.

Because custom channel groups can be applied retroactively in GA4 reports, the team can review how the corrected classification affects historical sessions that contain usable source data.

It also adds a non-mandatory question to the enquiry form asking how patients first heard about the practice.

Example

“The revised reporting captures more identifiable AI referrals.

Patient responses also suggest that AI assistants may influence some customer journeys that analytics cannot directly attribute.

However, the agency does not assign an estimated share of Direct traffic to AI without supporting evidence.

Instead, it reports measurable AI referral sessions and separately notes the limitations of attribution.

AI referral traffic is worth measuring, but accurate reporting means distinguishing what analytics can identify from what it can only suggest.”

See Also

• LLM Visibility — Measuring appearances and references in AI-generated responses.
• Direct Traffic — Understanding visits without an identifiable acquisition source.
• Google Analytics 4 — Measuring website sessions and traffic channels.
• AI Overviews — Google's AI-generated search experiences.
• UTM Parameters — Tracking information attached to campaign URLs.
• Attribution — Understanding which sources contribute to customer journeys.