How to track ChatGPT and AI traffic in GA4

How do you track ChatGPT and AI traffic in GA4?

In GA4, go to Admin > Data display > Channel groups and create a custom channel group. Add a channel called "AI assistants", give it a regex condition on session source that matches ChatGPT, Perplexity, Gemini and Copilot domains, then move it above Referral. Custom channel groups also apply to your historical data.

The setup takes about half an hour. The valuable part comes after: seeing which pages these visitors land on, whether they add to cart and how much revenue they bring, in the same table as every other channel.

Where does AI traffic show up in GA4 by default?

In the default channel group, a click from ChatGPT or Perplexity usually lands in Referral. So it's already in your reports, just mixed in with visits from your courier's tracking page or a forum thread. Unless you check sources one by one, you won't notice whether it's growing.

According to OpenAI's publisher FAQ, ChatGPT automatically adds utm_source=chatgpt.com to referral URLs from its search results. That's why ChatGPT visits usually appear clearly in GA4 with chatgpt.com as the source. The same page notes that your site must not block OAI-SearchBot if you want this traffic in the first place.

Some visits arrive without a trace. If someone copies a link out of an app and pastes it into a browser, or the referrer isn't passed, the visit can show up as Direct. Read whatever number you measure as a floor for AI-driven traffic, not the full picture.

How the sources compare

Source	Typical view in GA4	Can you separate it?
ChatGPT	chatgpt.com, with utm_source	Yes, by source name
Perplexity	perplexity.ai referral	Yes, if the referrer is passed
Gemini	gemini.google.com referral	Yes, if the referrer is passed
Copilot	copilot.microsoft.com referral	Yes, if the referrer is passed
Google AI Overviews and AI Mode	Inside organic Google search	No, not as a separate source
Copy and paste from apps	Direct	No

How do you set up a custom channel group?

Google's help page on custom channel groups uses exactly this scenario as its example. Here are the steps:

1. In GA4, open Admin, then Channel groups under Data display. You need the Editor role or higher.
2. Create a new group by copying the default channel group, so every other channel stays the same.
3. Click Add new channel and give it a clear name such as "AI assistants".
4. Choose session source as the condition, set the match type to "matches regex" and paste your pattern.
5. Move the channel above Referral in the list. GA4 assigns traffic to the first channel whose definition it matches; get the order wrong and visits still fall into Referral.
6. Save the group. Per the help page, custom channel groups apply to reports retroactively, so you see previous months with the new breakdown too.

Standard GA4 properties can have up to 2 custom channel groups, each with up to 50 channels. This is worth one of them, but keep the second for another need.

Which regex pattern should you use?

An example starting pattern:

chatgpt\.com|openai\.com|perplexity\.ai|gemini\.google\.com|copilot\.microsoft\.com|claude\.ai

The sample pattern in Google's help page is broader. Broad patterns carry a risk: they can catch unrelated sources that happen to contain "ai" or "gpt". After setup, open the Source/Medium report and check which sources land in the new channel. Put a reminder in the calendar to update the pattern when a new assistant takes off; the help page also recommends updating the regex as the list of assistants changes.

How do you measure AI traffic's impact on sales?

Once the channel exists, the real question is whether these visitors make you money. That depends on your ecommerce events working. If purchase, add_to_cart and begin_checkout are missing, the AI channel's conversion rate will look wrong too. If you aren't sure, run the checks in our GA4 ecommerce tracking guide first.

Then open a free-form exploration and set it up like this:

• Rows: the new channel group, then landing page
• Metrics: sessions, engagement rate, add to carts, purchases, revenue
• Filter: the AI assistants channel only

A pattern we see often: visits from AI assistants are few, but the people landing on comparison and guide pages already know what they want. Look at revenue per session and which pages get picked, not raw volume.

How do you measure Google AI Mode and AI Overviews traffic?

Set expectations here. Google Search Central's page on AI features says sites appearing in AI Overviews and AI Mode are included in overall search traffic in Search Console and reported in the Performance report under the "Web" search type. In GA4, those clicks arrive as organic Google traffic; they don't get their own source name the way ChatGPT does.

The practical answer is to read both tools together: trends in impressions and clicks in Search Console, conversions and revenue from organic traffic in GA4. We explain what it takes to get cited in AI answers in our GEO guide for ChatGPT and Perplexity.

Common mistakes when tracking AI traffic

• Leaving the channel below Referral and concluding no AI traffic exists.
• Writing an overly broad regex that counts email or other site referrals as AI traffic.
• Editing the default channel group instead of working on a copy.
• Looking only at sessions and never opening add-to-cart and revenue data.
• Blocking search crawlers in robots.txt and then wondering why ChatGPT sends nothing.

How often should you review the AI traffic report?

Monthly is enough for most ecommerce brands. Weekly numbers are small and noisy; one blog post starting to get cited in ChatGPT can look like a huge jump in a single week. Once a month, ask three questions: is the AI channel's share of total sessions growing, how does its revenue per session compare with organic search, and which five pages get the most entries from it? Keeping those three answers on one Looker Studio page lets you cover the topic in a few minutes at a leadership meeting. Example: if a store with 40,000 monthly sessions sees the AI channel go from 300 to 900 sessions over six months, the share is still small but the trend is a strong enough signal to adjust the content plan.

What should you do with the data?

Measurement isn't the goal on its own. Once you see which pages AI visitors land on, you have a concrete list of pages to strengthen and similar pages to create. The same table shows the categories where you don't appear at all. We cover how we combine that with content and technical work on our ecommerce SEO and GEO service page.

At Performetic we handle tracking setup and AI visibility in the same team. If you want to check whether your current GA4 setup separates this traffic correctly, request a free growth audit.

Key takeaways

• By default, AI assistant traffic in GA4 mostly sits in Referral, mixed in with other referrals.
• ChatGPT adds utm_source=chatgpt.com to its referrals, so those visits can be separated by source.
• Create a custom channel group with a regex condition and move the new channel above Referral.
• Custom channel groups apply to historical data; standard properties get 2 groups.
• AI Overviews and AI Mode clicks arrive as organic Google traffic in GA4; read them alongside Search Console.
• Judge AI traffic by revenue per session and landing pages, not volume.