How to Track ChatGPT, Perplexity & Gemini Referral Traffic in GA4

Why GA4 Doesn't Track AI Referrals by Default

GA4's default channel grouping was built for search engines, social platforms, and paid ads, not AI answer engines. When someone clicks a citation link inside ChatGPT, Perplexity, or Gemini, the click often arrives with a referrer domain GA4 doesn't recognize, or with no referrer at all if the app opens links in an in-app browser that strips this data. The result is that AI-driven visits get folded into Direct or a generic Referral bucket alongside dozens of unrelated sources, hiding a channel that is becoming genuinely important for top-of-funnel discovery.

Step 1: Check Your Referral Traffic Report First

Start in Reports > Acquisition > Traffic acquisition and switch the primary dimension to Session source. Search the table for known AI domains: chatgpt.com, chat.openai.com, perplexity.ai, gemini.google.com, and copilot.microsoft.com are the most common right now. Whatever shows up here is a confirmed floor, not a ceiling. It tells you AI platforms are already sending traffic, even before you build anything custom.

Step 2: Build a Custom Channel Group for AI Referrals

Go to Admin > Data display > Channel groups and create a new custom channel. Add a condition where Session source contains any of the AI domains you found in Step 1, and name the channel something like "AI Chatbots." Once it's live, this channel groups all matching sessions together going forward, so you can track it in standard reports like Traffic acquisition and Engagement without rebuilding the same filter every time.

Step 3: Create a Free-Form Exploration for AI Traffic

For a more detailed view, build a Free form exploration under Explore. Add Session source/medium as a dimension, filter it down to just your AI domains, and add Landing page as a secondary dimension. This shows you exactly which pages are being cited by which AI tool, which is far more useful than a single traffic number. Pages that show up repeatedly are the ones your AI-visibility efforts are actually working on.

Step 4: Watch for Traffic That's Hiding as Direct

Not all AI referral traffic can be isolated this way. Mobile apps and some in-chat browsers drop the referrer entirely, so those sessions land in GA4 as (direct)/(none) with no way to separate them from someone typing your URL directly. You won't get a perfect number here, but watching for unexplained spikes in direct traffic that line up with a new piece of content going live is a reasonable proxy signal.

Step 5: Add UTM Parameters Where You Can

Anywhere you control the outbound link, add UTM parameters so GA4 doesn't have to guess. If you list your own pages in an llms.txt file or submit content directly to a platform, tag those URLs with utm_source and utm_medium values like ai-referral. It won't cover traffic initiated by the AI tool itself, but it removes ambiguity for the channels you can control.

What to Do Once You Can See This Traffic

Once this traffic is visible, treat it like any other channel: track it over time, not as a single snapshot. Compare volume before and after you publish GEO-focused content or update your llms.txt file, and look at which landing pages recur. Most importantly, connect it to what happens after the click, not just the click itself. A trickle of AI referral traffic that converts into leads is worth more than a spike in visits that bounce immediately.

Frequently Asked Questions

Why does ChatGPT traffic show up as Direct in GA4?

Many AI apps, especially mobile versions, open links in an in-app browser that doesn't pass a referrer header. Without that header, GA4 has no source information to work with and defaults to Direct. This is a platform limitation, not something you can fix from the GA4 side.

What's the fastest way to see AI referral traffic in GA4 today?

Go to Traffic acquisition, set the primary dimension to Session source, and manually search for domains like perplexity.ai, chatgpt.com, or gemini.google.com. This takes about two minutes and works immediately, without needing to build a custom channel group first.

Should I create a custom channel group or just filter reports manually?

Manual filtering is fine for a one-time check. A custom channel group is worth setting up if you plan to monitor this monthly, since it groups the traffic automatically going forward and shows up in standard reports without repeating the same filter each time.

Do UTM parameters work for AI-driven traffic?

Only on links you control. If you tag a link inside your own llms.txt file or a piece of content you submit directly to a platform, GA4 will attribute it correctly. UTM tags can't be added to a link the AI model generates on its own, so this only closes part of the gap.

How often should I check this data?

Monthly is enough for most sites. AI referral volume is still small for the vast majority of businesses, so daily monitoring rarely reveals anything actionable. Check monthly, and pay closer attention right after publishing new GEO or AEO-focused content.