← Back to all posts
GA4 ChatGPT traffic: how to find and report it

GA4 ChatGPT traffic: how to find and report it

LA
Jedi Solomon
··5 min read

GA4 ChatGPT traffic is easiest to find in Reports, Acquisition, Traffic acquisition: switch the dimension to Session source / medium and search for chatgpt. Most of it lands there as a referral, and GA4's default channel groups bury it inside the general Referral bucket. A custom channel group fixes that.

What AI referral traffic in GA4 looks like

Each assistant sends a different referrer, and some send none. ChatGPT usually adds utm_source=chatgpt.com to the links in its answers, so those sessions show chatgpt.com as the source. Older sessions may show chat.openai.com.

AssistantSource values to look forWhat to watch
ChatGPTchatgpt.com, chat.openai.comMedium may read (not set)
Perplexityperplexity.aiArrives as an ordinary referral
Geminigemini.google.comSome visits carry no referrer and land in Direct
Copilotcopilot.microsoft.comRead the full source so it is not mixed up with Bing organic
Claudeclaude.aiAdd it now so it is counted as it grows

Perplexity referral traffic is easy to spot because a source search for perplexity finds it straight away. Gemini is the engine most likely to hide.

Setting up GA4 ChatGPT traffic as its own channel

The default groups put these visits under Referral, next to every partner link and forum post. A custom channel group pulls them out.

  1. Go to Admin, Data display, Channel groups, and choose Create new channel group.
  2. Add a channel and name it AI assistants.
  3. Set the condition to Session source, matches regex, and paste: chatgpt|openai|perplexity|gemini|copilot|claude
  4. Drag the new channel above Referral. GA4 reads the list top down, and the first match wins.
  5. Save the group.

Once it is saved, GA4 ChatGPT traffic and every other assistant show up as one channel you can chart against Organic Search. The regex matches on fragments: openai catches chat.openai.com and chatgpt catches chatgpt.com, so older and newer ChatGPT sessions land in the same channel. The group applies to past data in your reports, so you get history as well as the future.

To look at one engine at a time, use an Exploration: a free form report with Session source as the row and a filter on the same regex, which leaves your standard reports untouched.

Read it by landing page and outcome

Open Reports, Engagement, Landing page, and add your new channel group as the secondary dimension. The pages that appear are the pages assistants send people to. Look at which ones come up and ask whether they are the pages you would want an assistant to recommend. For a store, expect guides and product pages. For a service firm, expect service and pricing pages.

Then compare engagement rate and purchases or form submissions against Organic Search. First Page Sage reported that AI-referred visits were 4.73% of US retail site visits in Q3 2026, up from 0.43% in Q4 2024, and converted about 1.6x other traffic. Adobe measured AI traffic to US retailers converting 42% better than non-AI traffic in Q1 2026. Your site will have its own ratio, and at small volumes a single month proves little.

What GA4 cannot see

  • Visits that arrive with no referrer fall into Direct. Gemini is a frequent source of these.
  • A buyer who reads your name in an answer and then types your address or searches your brand shows up as Direct or Organic Search.
  • A buyer who reads your name and never visits leaves no session at all. SparkToro and Similarweb found that 68% of US Google searches end without a click, so zero-click behavior is normal.

Treat the AI assistants channel as a floor: the real number is higher. Two habits cover those gaps. Add a "How did you hear about us" field to your forms or order confirmation, with "ChatGPT or another AI assistant" as an option. And run a fixed panel every month: 25 buyer prompts, each tried in ChatGPT, Gemini and Perplexity, with a note on whether you were named or linked. The AI visibility audit page explains how we set that panel.

A worked example

JCS Fashions, a 4,878-product Shopify apparel store in Milpitas, California, saw sessions from AI assistants rise from 20 in January to 569 in October 2026. Google clicks over the same year went from 205 a month in January to 626 in September. Both lines are readable because they sit in separate channels. The full numbers are in the JCS Fashions customer story.

A monthly routine to start this week

  1. Build the AI assistants channel group today.
  2. Save a landing page report with that group as the secondary dimension.
  3. Add the self-reported source option to your lead form or checkout.
  4. Write down your 25 prompts and record where you are named today as your baseline. Keep the prompts the same each month so the results can be compared.
  5. On the first of each month, log AI sessions, purchases or leads from that channel, the top landing pages and the panel result.

A monthly GA4 ChatGPT traffic number counts the clicks. For the answers that never produce a click, LynkAI's free AI visibility scan shows where you stand in ChatGPT, Gemini and Perplexity today.