How to Track ChatGPT and AI Traffic in GA4 and Search Console

To see visits from ChatGPT in Google Analytics 4, look at session source in the Acquisition reports: this traffic usually appears with the source "chatgpt.com" in the referral channel. The same goes for Perplexity, Gemini and Copilot. The cleanest way to track these visits over time is to create a custom channel in GA4 called "AI assistants". Visibility that does not turn into clicks is measured separately, with the generative AI report in Search Console, the AI Performance screen in Bing Webmaster Tools, and regular question tests.

Where does ChatGPT traffic show up in GA4?

In GA4, open Reports > Acquisition > Traffic acquisition, change the primary dimension to "Session source / medium" and type "chatgpt" into the table search. On most sites these visits are grouped in the "chatgpt.com / referral" row. Repeat the same search for "perplexity", "gemini" and "copilot".

Watch out for two traps. First, links from ChatGPT sometimes arrive with a source tag (utm_source) added to the end of the URL, in which case the traffic also appears in the campaign dimensions. Check which form shows up in your own report. Second, some links opened from mobile apps do not pass referrer information to the browser and get lost inside "direct" traffic. So read the number you see in GA4 as a lower bound for AI traffic.

How do you set up an AI channel in GA4?

In the default channel grouping, AI assistants are mixed in with other sites under "Referral". To create a separate channel:

1. In Admin, go to the Channel groups page under Data display.
2. Use Create new channel group to make a copy of the default group and give it a name.
3. Add a new channel and name it "AI assistants".
4. As the condition, choose the Source dimension, use the "matches regex" operator and enter this expression: chatgpt|openai|perplexity|gemini|copilot|claude
5. Move this channel above the Referral channel in the list. GA4 assigns traffic to the first channel whose definition matches; if you leave it below, the visits still fall into Referral.
6. Save.

According to Google's help page, custom channel groups can be applied to reports retroactively, so you will see your historical data with the new grouping too. Standard GA4 properties allow at most two custom channel groups alongside the default group, so name yours carefully. Update the regular expression to match the source names you actually see in your reports; as new assistants appear, you will need to add them.

How should you interpret this traffic?

Once the channel is in place, look at three questions: which pages do visitors from AI assistants land on, how much do they engage, and what do they contribute to conversions (forms, carts, searches)? The landing page list is an indirect but valuable clue to which of your content the assistants are citing as a source.

Compare against the organic search channel. A visitor from an AI assistant has often already asked their question, received part of the answer and arrives with a more specific need. So even when session numbers are low, engagement per page and conversion rate can behave differently. See which direction the difference goes in your own data; do not assume a general rule.

If the numbers are small, look monthly rather than weekly, and do not draw conclusions from swings of a few visits. What matters is the trend in the same channel over months. When you update content or publish a new page, note the date and look at the change over the following two or three months. That is the most practical way to tell which work had an effect.

How do you measure visibility that does not produce clicks?

Most AI answers do not generate a click: your brand is mentioned, but nobody comes to the site. To measure this, use the official reports the platforms provide.

• Google Search Console, generative AI performance report (Search): shows impressions for links to your site in AI Overviews and AI Mode, by page, country, device and date. According to Google's help page, it was opened to all sites worldwide as of August 31, 2026. The report is impression-based and does not give clicks. This data also continues to be counted within the Web search type in the general Performance report.
• Bing Webmaster Tools, AI Performance: opened in public preview in February 2026, this screen shows how many times your site was cited as a source in Microsoft Copilot and AI summaries in Bing, how many of your pages receive citations on an average day, and the query phrases the AI used when looking for content (grounding queries). This screen does not provide click data either.

How do you track brand visibility in ChatGPT and Perplexity?

These platforms do not yet offer brand owners a panel like Search Console. Two methods remain.

A fixed question set: define 20 to 40 questions your customers would ask (branded, category, comparison and problem-focused). Each month, ask the same questions under the same conditions and record in a table whether your brand is mentioned, whether it is cited as a source, and who the competitors are. Because answers vary, try each question more than once.

Server logs: OpenAI uses OAI-SearchBot for ChatGPT search results and ChatGPT-User for page visits made at a user's request. Perplexity's equivalents are PerplexityBot and Perplexity-User. Tracking which pages these user agents visit, and how often, in your server logs shows whether your content is reaching these systems. According to Perplexity's documentation, Perplexity-User generally does not follow robots.txt rules because it acts on a user's request.

Why is your brand not showing up in AI answers?

Before building a report, rule out these possible causes:

• Your robots.txt file may be blocking search crawlers such as OAI-SearchBot or PerplexityBot.
• Important pages may not be in Google's index, or snippet display may be restricted.
• Your site may be excluded through the generative AI control for Search in Search Console.
• Very little may have been written about your brand in third-party sources (industry publications, directories, communities).
• Your company information may be described differently in different places, or your content may be out of date.

A monthly report template

• GA4: sessions, landing pages and conversions from the AI assistants channel
• Search Console: impressions and top pages in the generative AI report
• Bing Webmaster Tools: citation count, cited pages, leading query phrases
• Question set: mention rate, citation rate, competitor list
• Server logs: pages visited by AI crawlers

For how citations and brand presence work in AI answers, see our article on AI visibility and AI citations. For measurement setup and improvement, see our AI search optimization service.