How to Track AI Traffic in Google Analytics GA4 2026

Google Analytics 4 now has a dedicated AI Assistant channel that tracks traffic from ChatGPT, Gemini, and Claude automatically. That sounds like the end of the measurement problem. It is not. The default channel undercounts your AI traffic by splitting a single source across three different channels — and it ignores Perplexity entirely. Here is how to fix it and start making real attribution decisions from the data.

What GA4's AI Assistant Channel Actually Does

On May 13, 2026, Google added "AI Assistant" to GA4's default channel groups. When GA4 recognizes a referrer from a known AI platform, it assigns the session medium ai-assistant, drops it into the AI Assistant channel, and stamps the campaign dimension (ai-assistant). No configuration required — it is active in every GA4 property today.

The initial recognized platforms were ChatGPT, Gemini, and Claude. But Google's list is not stable. By June 2026, Search Engine Journal documented the same undercounting problem this creates: recognized AI sources can still split across AI Assistant, Referral, and Unassigned. Perplexity, which drives some of the highest-intent AI referral traffic, is still a source you need to catch outside the native channel.

One more distinction that matters for reporting: traffic from Google AI Overviews and AI Mode stays in Organic Search, not AI Assistant. If you are trying to measure the full impact of AI on your site traffic, the default channel is a starting point — not the answer.

Why Your AI Traffic Number Is Almost Certainly Wrong

Here is the problem I keep seeing in client accounts. The same source — chatgpt.com — shows up in three different channels simultaneously:

chatgpt.com / ai-assistant lands in AI Assistant. This is the slice GA4 recognized and tagged correctly.
chatgpt.com / referral lands in Referral. These are sessions GA4 did not tag, plus anything that arrived before the channel rolled out to your property (the rollout dragged into June for some accounts).
chatgpt.com / (not set) lands in Unassigned. When the ChatGPT mobile app's in-app browser strips the referrer, GA4 keeps the source but loses the medium — and there is no channel rule to catch that combination.

So your real ChatGPT traffic is smeared across three channels, one of which — Unassigned — is a bucket most teams never open. Read the AI Assistant channel by itself, and you are undercounting every single time. Add the Perplexity gap, and the number gets worse.

The Fix: One Custom Channel Group That Catches Everything

The move that solves this is straightforward. Build a custom channel group that matches on source instead of source-plus-medium. The moment you do that, all three versions of chatgpt.com collapse into a single line — and you can include the platforms Google leaves out.

Here is the setup:

Go to Admin > Data display > Channel groups and create a new group.
Add a channel called "AI" (or whatever name fits your reporting).
Set the condition to Source matches regex with a pattern covering the AI domains you want to track.
Drag the AI channel above Referral and Organic so it claims matching sessions first.
Save, then apply the group as your primary dimension in any acquisition report.

A solid starting regex:

.*(chatgpt\.com|openai\.com|perplexity|gemini\.google\.com|copilot\.microsoft\.com|claude\.ai|deepseek\.com|grok\.com|you\.com).*

Two details that matter: use gemini.google.com as a specific host so you do not accidentally sweep in all Google organic traffic. And treat the pattern as perishable — review it quarterly, because AI platforms launch, rebrand, and change domains regularly.

The bonus: custom channel groups apply retroactively across your entire date range. That rescues all the old ChatGPT sessions that were sitting in Referral before the native channel existed.

What the Conversion Data Actually Tells You

Once your measurement is clean, the numbers get interesting. Similarweb's clickstream data shows ChatGPT referrals converting at approximately 7 percent — ahead of organic search and close to paid. Adobe's Q1 2026 research found AI assistant traffic converting 42 percent better than standard organic search overall.

The volume is still small relative to Google. But a channel that converts at 7 percent on low volume is not a rounding error — it is a high-intent signal telling you exactly which pages AI engines trust enough to cite.

To identify those pages: go to Reports > Engagement > Pages and screens, add a filter where Session default channel group exactly matches your custom AI channel, and sort by sessions. The pages that surface are the ones AI platforms are sending people to. That is your best evidence for what content the models find useful — and where to invest next.

How to Turn AI Traffic Data Into a Budget Decision

Tracking AI traffic is not the goal. The goal is knowing whether to shift budget toward AI visibility or keep spending elsewhere. Here is the decision framework I use:

Calculate per-session value. Take your custom AI channel's conversion data and revenue attribution. Divide revenue by sessions. Compare that number to your paid search and organic per-session value. If AI traffic is converting at 7 percent while your paid campaigns convert at 3 percent, you have a channel that is outperforming your spend — and it costs nothing per click.

Identify what AI engines cite. The landing page data from your custom channel shows you exactly which content earns AI referrals. In my experience, pages that get cited share a pattern: they answer a specific question with sourced evidence, not generic overviews. If your best-converting AI landing pages are all deep-dive guides with named data sources, that tells you where to put your next content dollar.

Set a baseline now. AI traffic is still sub-1 percent of total sessions for most sites, but the growth rate matters more than the current number. With your custom channel retroactively capturing historical data, you can track the trajectory month over month. I have been tracking AI brand mentions separately, and the correlation between citation frequency and referral traffic is becoming clearer by the quarter.

The Measurement Gap Nobody Is Solving Yet

Even a perfectly configured custom channel has blind spots you need to account for:

Mobile app traffic disappears into Direct. When someone asks ChatGPT a question in the mobile app and taps a link in the response, the in-app browser frequently strips the HTTP referrer. GA4 has no source to read, so the session lands in Direct. There is no channel rule — custom or default — that can rescue a session with no referrer. For some sites, this means the majority of ChatGPT mobile traffic is invisible in GA4.

AI Overviews clicks count as Organic Search. If Google's AI Overviews surfaces your content and a user clicks through, that session is attributed to Organic Search — not AI Assistant. You are getting AI-driven traffic that does not show up in any AI channel. The only way to separate it today is through Search Console's AI Mode reporting, which is still limited.

Brand mentions without clicks are unmeasurable in GA4. An AI engine can recommend your company in dozens of responses daily without generating a single referral click. GA4 measures visits, not mentions. For the full picture of your AI visibility, you need to supplement GA4 with direct AI citation monitoring — probing what the models actually say about you, not just what they send your way.

Server logs fill some of the gaps. If you have access to raw server logs, you can identify AI bot user agents (ChatGPT-User, PerplexityBot, ClaudeBot, Applebot) hitting your pages. This does not tell you about human referral traffic, but it tells you which pages AI engines are actively retrieving for their answers — a leading indicator of future referral traffic.

The Setup Checklist

If you want to get this right this week, here is the sequence:

Step	Action	Time
1	Open GA4, go to Reports > Acquisition > Traffic acquisition, and check if the AI Assistant channel appears	2 min
2	Add Session source/medium as a secondary dimension and look for AI sources in Referral or Unassigned	5 min
3	Go to Admin > Channel groups and create a custom group with a source-based regex covering all major AI platforms	15 min
4	Drag the custom AI channel above Referral and Organic, save, and verify it captures historical data	5 min
5	Filter Pages and screens by your custom AI channel to identify which pages AI engines cite	10 min
6	Calculate per-session value for AI traffic vs. your other channels	15 min
7	Set a monthly review to update the regex and track growth	Recurring

The total setup is under an hour. The insight it unlocks — knowing exactly what AI engines send your way, which pages they cite, and what that traffic is worth — is the foundation for every AI visibility decision you make from here.

FAQ

Does the GA4 AI Assistant channel track Perplexity traffic?

No. As of July 2026, Perplexity traffic still needs to be caught with custom source rules instead of relying on the native AI Assistant channel alone. Perplexity traffic lands in Referral by default. The only way to capture it in an AI-specific channel is to build a custom channel group with a source-based regex that includes perplexity in the pattern.

Can I see AI traffic data from before May 2026?

The default AI Assistant channel only tags sessions from its rollout date forward. However, a custom channel group applies retroactively across your entire GA4 date range. Sessions from ChatGPT or other AI sources that were previously categorized as Referral or Direct will appear under your custom AI channel for any historical period where GA4 captured the source.

How do I separate ChatGPT traffic from Google AI Overviews traffic?

They are in different channels by design. ChatGPT referrals appear in the AI Assistant channel (or your custom AI channel). Google AI Overviews clicks are attributed to Organic Search because they originate within Google's search interface. To isolate AI Overviews impact specifically, use Search Console's AI Mode reporting alongside GA4.

What percentage of total traffic comes from AI assistants right now?

For most sites, AI assistant referral traffic is still under 1 percent of total sessions. But the conversion quality matters more than the volume. Similarweb data shows ChatGPT referrals converting around 7 percent, and Adobe Q1 2026 research found AI traffic converting 42 percent above organic search. A small channel with outsized conversion rates deserves measurement investment, not dismissal.

Is the GA4 AI Assistant channel the same as tracking AI bot crawlers?

No. The AI Assistant channel tracks human visitors who arrive at your site after clicking a link in an AI response. AI bot crawlers — like GPTBot, ClaudeBot, and PerplexityBot — are automated agents retrieving your content to build their knowledge base. Bot traffic typically does not appear in GA4 at all; it shows up in server access logs. Both signals matter, but they answer different questions: the channel tells you who AI engines send your way, and the logs tell you which pages AI engines are actively reading.