How to Track AI Traffic in GA4 (ChatGPT, Perplexity & More)

Something shifted in 2024. Marketers started noticing referral traffic from sources they didn't recognize — chat.openai.com, perplexity.ai, claude.ai. Then they noticed something more unsettling: a chunk of traffic with no referrer at all, lumped into "direct," that didn't behave like direct traffic should. Sessions that converted well, spent real time on page, and arrived at oddly specific blog posts nobody had linked to.

That's AI traffic. And GA4 doesn't label it automatically. If you're not actively looking for it, you're flying blind on one of the fastest-growing referral channels of the last two years.

Why GA4 Doesn't Label AI Traffic by Default

Traditional referral tracking works on a simple assumption: when a user clicks a link, the browser sends a Referer HTTP header to the destination site. GA4 reads that header and attributes the session accordingly. Most of the time, this works fine.

AI assistants break this in two ways. First, some — like the ChatGPT web interface — do pass a referrer, so you'll see sessions from chat.openai.comin your referral report. That part is measurable today. Second, many AI environments strip referrers entirely — ChatGPT mobile apps, Perplexity's in-app browser, Claude in certain configurations. Those sessions land in "direct" and look indistinguishable from someone typing your URL.

This is the fundamental challenge: there's no single clean report in GA4 that shows you "all AI traffic." You have to build it from a combination of sources.

Step 1 — Find the AI Referrers You Can Already See

Start with what's measurable right now. In GA4, go to Reports > Acquisition > Traffic Acquisition. Set your primary dimension to Session source and scan for these known AI referral domains:

• chat.openai.com — ChatGPT web (desktop)
• perplexity.ai — Perplexity AI
• claude.ai — Anthropic's Claude web interface
• gemini.google.com — Google Gemini
• copilot.microsoft.com — Microsoft Copilot
• grok.com — xAI's Grok

In the Explore section, create a free-form exploration with sessionSource as your dimension and add a filter matching these domains as a regex: openai|perplexity|claude|gemini|copilot|grok. Pull in engagement rate and average session duration — AI referral traffic often shows high engagement because users arrive with specific intent shaped by the AI's response.

Step 2 — Build a Custom Channel Group for AI Traffic

GA4's default channel grouping has no "AI" channel. Fix this by going to Admin > Data display > Channel groups > Create new channel group. Add a channel called "AI Referral" with conditions targeting the known AI domains using the Session source contains condition.

Important: channel groups only apply going forward — they won't backfill historical data. Create yours today even if you're not ready to act on it. The sooner you start the clock, the more trend data you'll have in six months when AI traffic becomes a standard reporting metric.

Step 3 — Investigate the "Direct" Blind Spot

A significant portion of AI traffic arrives without a referrer — estimates suggest 40–70% depending on the tool. It's invisible inside your existing direct traffic bucket. You can't definitively identify which direct sessions came from AI, but behavioral signals help narrow it down:

• Landing directly on deep interior pages (not the homepage) — the landingPage dimension is your friend here
• First-visit sessions (new users) with no prior organic or paid history
• Short-to-medium sessions on informational content that answers a specific question
• Device and geo patterns inconsistent with your typical direct audience

None of these signals are definitive on their own, but a cluster of them — especially the deep landing page combined with new user status — is a strong indicator of AI referral with a stripped referrer.

Step 4 — Monitor Trends Across All Your Properties

Identifying AI traffic once is useful. Monitoring it continuously is where the real insight lives. AI traffic patterns move fast — a mention in a popular ChatGPT response can spike sessions overnight, and changes in how AI tools cite sources can crater traffic from a specific source within weeks.

If you manage multiple GA4 properties, watching this across all of them is operationally tedious inside native GA4. Perch Analyticssurfaces AI traffic as a dedicated metric alongside standard GA4 data for every property in your account — no custom segments or explorations to build, no tab-switching. It's free for up to three properties.

What AI Traffic Data Actually Tells You

A few grounding notes before you go deep on this data. AI citations don't work like backlinks — a high-ranking page in Search sends consistent traffic, but an AI citation depends on what a user asked and how the AI's retrieval systems work in that moment. The traffic can be lumpy and non-linear.

High engagement doesn't always mean high commercial intent either. AI referral sessions often show excellent engagement metrics because users arrive primed by the AI's framing — but that engagement may be informational rather than transactional. Check conversion rates alongside engagement to understand actual business value.

Finally, treat your measured AI traffic as a floor, not a ceiling. Given how many sessions arrive as "direct," the real number is almost certainly higher than what GA4 shows.

Get Started Today

The setup described here — a custom channel group, a focused Exploration, and behavioral analysis of direct traffic — takes under an hour. The channel group won't apply retroactively, so the only cost of waiting is less trend data later.

If you manage multiple GA4 properties and want AI traffic consolidated alongside all your other key metrics in one place, Perch Analytics is worth trying. Sign in with Google and your data is there in under a minute — no configuration required.