How to track traffic from ChatGPT (and other AI assistants) in Google Analytics 4

Quick summary

• GA4 has a default "AI Assistant" channel for visits from assistants such as ChatGPT, Gemini, DeepSeek, Copilot or Grok. It's documented in Google Analytics Help.
• That channel excludes AI Overviews and AI Mode, which count as Organic Search. You can't separate them in Analytics; for those, use Search Console's generative AI performance report.
• Google doesn't publish the full list of assistants in the channel, so it's worth adding a source filter (chatgpt.com, perplexity.ai, claude.ai…).
• Measure what matters: which pages get that traffic and whether it converts, not just how many sessions there are.

What GA4 already does for you

For a long time, visits from ChatGPT were lumped into "Referral" and you had to dig them out by hand. Now GA4's default channel group includes a dedicated channel:

AI Assistant is the channel by which users arrive at your site from sources like ChatGPT, Gemini, Deepseek, Copilot, or Grok. It excludes Google's AI Overviews and AI Mode.

According to the same documentation, when the referrer matches a list of AI assistants, GA4 sets the medium to ai-assistant and the campaign to (ai-assistant), and the session lands in that channel.

Where to find it

1. Reports → Acquisition → Traffic acquisition.
2. The default primary dimension is Session default channel group. Look for the "AI Assistant" row.
3. To see which assistant each session came from, add Session source as a secondary dimension.
4. To see where they land, switch to Landing page or build an exploration (below).

If you compare long periods and see chatgpt.com visits under "Referral" in older months, don't rely on the channel for that comparison: filter by source, which is stable.

Complete it with a source filter

Google names five assistants as examples but doesn't publish the full list. Perplexity and Claude, for instance, aren't named. To avoid depending on it, build an exploration with a source filter:

1. Explore → Free form.
2. Dimensions: Session source, Landing page + query string, Date.
3. Metrics: Sessions, Engaged sessions, Key events (or your conversions).
4. Filter: Session source matches regex:

chatgpt\.com|chat\.openai\.com|perplexity|gemini\.google\.com|copilot\.microsoft\.com|claude\.ai|chat\.deepseek\.com|grok\.com|meta\.ai|chat\.mistral\.ai

Check your source report now and then for new assistants and add them to the expression. If an assistant adds a utm_source parameter to its links, GA4 uses that value as the source; look at how it appears in your data before trusting the regex.

A custom channel group (optional)

If you want each assistant as its own channel, you can create a custom channel group in your property's admin settings (channel groups section), with one rule per assistant based on source. It's handy if you report to clients or leadership and want "ChatGPT" and "Perplexity" as separate lines.

Questions worth answering with that data

Counting sessions is the easy part. The useful parts:

• Which pages get AI traffic? They're often different from your organic landing pages. If a comparison or pricing page concentrates ChatGPT visits, that's a hint about what's being cited.
• Does it convert? Compare key events per session for AI Assistant against Organic Search. Google says, about AI Overviews, that those clicks tend to be "higher quality"; that's its own claim, so check it against your data for the other assistants.
• Does it change after you publish? If you published a page for a specific question, see whether AI-referred visits start landing on that URL.

What GA4 can't see

Worth spelling out, because AI traffic is almost always undercounted in Analytics:

1. AI Overviews and AI Mode. They arrive with a Google referrer and GA4 counts them as Organic Search, by design. To see your impressions in those features, use Search Console's generative AI performance report, available to all sites since 31 August 2026. It shows impressions by page, country, device and date.
2. Apps that don't pass a referrer. Some desktop or mobile apps open links without one. Those visits fall into "Direct".
3. People who don't click. Many users read the recommendation, remember your name and search for you later or type your URL. AI caused the visit, but GA4 shows it as branded search or direct.
4. Cookie consent. If GA4 only measures people who accept cookies, everything above shrinks proportionally.

That's why traffic is an outcome metric, not a visibility metric. To know whether you're being recommended, you have to ask the questions and count mentions and citations; we explain how in how to measure your brand's visibility in ChatGPT.

A ten-minute weekly routine

1. Open Traffic acquisition and note sessions and key events for AI Assistant against last week.
2. Open your exploration with the source filter and check the top landing pages from AI assistants.
3. Look for new sources in the referral list and add them to the regex if they're assistants.
4. Cross-check with Search Console's generative AI report for the Google side.
5. Write down anything you published that week, so you can connect changes to actions later.

How Citable does it

Citable measures AI-referred visits in two ways, and you can use either or both:

• By connecting your GA4. It reads sessions by day, landing page and source, and classifies them with a list of assistant domains (ChatGPT, Perplexity, Gemini, Copilot, Claude, DeepSeek, Grok and others) and utm_source values. Only aggregated counts are stored, with no personal data.
• With its own cookieless script, which counts visits arriving from AI assistants even if you don't use GA4.

Both apply the same honest limit: clicks from AI Overviews and AI Mode can't be separated from regular search, and we don't mix them in. What we do is put AI traffic next to mention and citation measurement for each engine, so you see visibility and visits in one place. Plans are on the pricing page.

FAQ

Where do I see ChatGPT traffic in GA4?

In Reports, Acquisition, Traffic acquisition, with Session default channel group as the dimension: the "AI Assistant" row. Add Session source as a secondary dimension to see which assistant sent each visit.

Why don't AI Overview visits show in the AI Assistant channel?

Because Google excludes them on purpose: clicks from AI Overviews and AI Mode arrive with a Google referrer and GA4 counts them as Organic Search. To measure your presence in those features, use Search Console's generative AI report.

Does the AI Assistant channel include Perplexity and Claude?

Google doesn't publish the full list; its documentation names ChatGPT, Gemini, DeepSeek, Copilot and Grok as examples. To be safe, build an exploration with a source filter that includes perplexity and claude.ai.

Why do I get so few AI visits if ChatGPT recommends me?

Because many recommendations never become a click: the user remembers your name and searches for you later. Some apps also don't pass a referrer, so those visits land in Direct. Measure visibility by sampling questions and treat traffic as a complementary signal.

Conclusion

GA4 gives you the AI Assistant channel with no setup. Complete it with a source filter, look at it by landing page and conversion, and don't ask it for what it can't give: AI Overviews sit in organic and many recommendations never arrive as a click. Traffic tells you whether visibility converts; to know whether it exists, you have to measure it in the answers.