By Rahul Saini, Author at Search Counsel Co. Last updated [July] 2026.

Why AI traffic hides

GA4 expects visitors to arrive through a standard search referrer like google.com or a campaign you tagged. AI platforms fit neither pattern, so their traffic scatters:

• Copy-pasted links land in Direct. Most people copy a URL out of a ChatGPT answer and paste it into a new tab, which passes no referrer. GA4 files that under Direct, indistinguishable from someone typing your address.
• Clicked links land in generic Referral. When users do click a citation in Perplexity or Claude, the visit arrives tagged perplexity.ai or claude.ai and gets lumped in with all your other referrals.
• Google’s own AI counts as Organic. Clicks from AI Overviews and AI Mode report as organic search, so they can’t be cleanly separated at all.

The result is that only a minority of AI-driven visits are visible in a default GA4 report, with the majority misclassified, and most brands can’t attribute the channel at all. Whatever number you eventually see is an undercount, which matters because these visits tend to convert at several times the rate of ordinary organic traffic.

The two-part fix

No single setting solves this, so use two together:

• GA4’s native AI Assistant channel. Added in 2026, it automatically routes recognized platforms like ChatGPT, Gemini, and Claude into their own channel, covered in the GA4 for SEO guide. It’s a good start, but it misses some sources, only works going forward, and can’t recover sessions that arrive with no referrer.
• A custom channel group with regex. This is the safety net. It catches the platforms the native channel misses, groups them into one clean channel, and, unlike the native channel, applies your rule across the data you already have when you view reports with it selected. Run both.

Build the channel group

The setup takes about fifteen minutes and you only do it once:

1. Go to Admin, Data settings, Channel groups, and click Create new channel group. Name it something clear like “AI Traffic.”
2. Add a new channel, name it AI Search, and set the condition to Source matches regex.
3. Paste a pattern covering the major platforms, for example: chatgpt\.com|chat\.openai\.com|openai\.com|perplexity\.ai|claude\.ai|gemini\.google\.com|copilot\.microsoft\.com|deepseek\.com|grok\.com|meta\.ai|you\.com
4. Reorder the channel so AI Search sits above Referral. This is the step people miss. GA4 assigns each session to the first channel it matches, so if Referral comes first, your AI visits stay buried in it.
5. Save. One note: GA4’s regex is case-sensitive, and sources normally appear in lowercase, so add a case-insensitive flag if you spot mixed-case values.

Three methods by access

Pick the one that matches your permissions:

• Custom channel group (needs Admin access). Permanent, property-wide, and visible in every standard report. This is the right default for most teams.
• Free-form Exploration (no admin needed). In Explore, add Session source/medium as a dimension, apply the same regex as a filter, and add Landing page with Sessions and key events. Quick and good for ad-hoc analysis, though it’s private to your account.
• Looker Studio calculated field (no GA4 configuration). Use a CASE statement with a regex match on the source to label AI traffic inside your dashboard. It only works in that report, but it’s the fastest option when you can’t touch the property.

The report, and the honest caveats

Once it’s live, the single most useful view is landing page by AI source: it shows exactly which of your pages each AI platform cites, and how that traffic converts. That’s your citation footprint made tangible, the traffic-side companion to the AI visibility KPIs.

Now the honesty, so you don’t misread your own data:

• It’s a floor, not a full count. The no-referrer leak means real AI influence is larger than what you can see, so never conclude “AI sends us nothing” from a small number. This is exactly why you also measure AI visibility directly, per the KPIs.
• Google’s AI is invisible here. AI Overview and AI Mode clicks stay in organic, so pair this with the Search Console AI report read qualitatively.
• Don’t confuse crawlers with referrals. Bots scraping your site to train models are not humans clicking citations; they’re a separate analysis, and GA4’s bot filter handles the former, not the latter.
• Don’t track only ChatGPT. It dominates today, but Perplexity, Gemini, and Claude are growing, so keep the pattern broad.
• Use multi-touch, not last-click. AI often plays an early awareness role, which last-click attribution hands to a later channel. And remember much AI influence also surfaces as branded or direct traffic, covered in branded vs non-branded.

Keep it current: refresh the regex quarterly as platforms launch and rebrand, and scan your Referral list monthly for new domains containing “ai,” “chat,” or “gpt.” AI browsers add a wrinkle too, some pass a clean referrer while others mask their origin as Direct.

FAQ

The one thing to do next

Set up the custom channel group today, it’s fifteen minutes, then check the landing-page-by-AI-source report each month to see which pages AI platforms cite. That closes the measurement gap. Earning more of that traffic, the content and signals that get you cited in the first place, is the next job, covered in our AI visibility guide, and the numbers feed straight into your reporting dashboard.