GA4 AI Traffic: How to Track ChatGPT and AI Visits and See the Whole Picture

Open Reports, then Acquisition, then Traffic acquisition. Change the table to show Session source / medium. Every AI tool that sent visits will show up as its own row.

By default the report groups visits by Session default channel grouping, which lumps them into Direct, Organic Search and similar buckets. You can switch the primary dimension from the dropdown (Google Analytics Help). The source view is where AI visits become visible.

Do not expect big numbers. AI referrals were 1.08% of website visits across 3.3 billion sessions in Conductor's 2026 benchmarks. What matters is quality. Compare engagement and key events with organic, because volume tells you little.

This quick look only finds visits that arrived with a source. The next sections show how to group them, and what it still misses.

Look for chatgpt.com, perplexity.ai, claude.ai, gemini.google.com and copilot.microsoft.com. The exact names vary, so scan your own source list instead of trusting a fixed one.

Start with ChatGPT. It sent 87.4% of all AI referral visits in the Conductor benchmarks, which covered 13,770 domains in 10 industries. If you only have time to track one source, track that one.

Also look for odd spellings and subdomains, such as a chat or search host on the same brand. New AI tools appear often, so review the list once a month.

Build one custom channel called AI search. Then every AI visit lands in one row, in every report.

1. Go to Admin, then Data display, then Channel groups. You need the Editor role or higher.
2. Click Create new channel group. Name it, then add a channel called AI search.
3. Set the condition to Source matches regex. Paste a pattern like the one below.
4. Drag AI search to the top of the list. Save the group.

Order matters. Traffic goes to the first channel whose rule it matches, so AI search has to sit above any default channel that would catch those visits first. The new group also works on past data, because custom channel groups apply to reports retroactively (Google Analytics Help).

A standard property can hold two custom channel groups. Keep one for this.

GA4 files a visit under Direct when it has no source to read. If the browser hides the referrer and the link has no tag, the visit has no source.

ChatGPT helps here. It adds utm_source=chatgpt.com to the links it sends, according to the OpenAI Help Center. That tag survives when the referrer is hidden, so those visits still show chatgpt.com as the source.

Other AI tools may not tag links. And many buyers never click. Similarweb found that 56% of AI-influenced visits arrived through a branded search, against 40% of standard visits (June 23, 2026). The buyer read the answer, then searched your name.

You cannot fix that inside GA4. A channel group only regroups what GA4 recorded. Three habits close the gap:

• Ask at signup. Add "How did you hear about us?" as free text. Buyers write "ChatGPT".
• Watch branded search. In Search Console, track queries that contain your brand after you win citations.
• Check landing pages. AI answers often send visitors to a deep page, so see which pages they land on.

Our cross-channel attribution guide has the UTM naming rules, and our AI search revenue attribution guide shows how to join these signals to revenue.

GA4 cannot show AI bots, and it cannot show answers that earned no click. Both matter.

First, bots. Google says traffic from known bots and spiders is automatically excluded in GA4, and you cannot turn that off (Google Analytics Help). On top of that, Vercel found that none of the major AI crawlers render JavaScript. GA4 runs on JavaScript, so its tag never fires for them. In one month Vercel counted 569 million GPTBot fetches and 370 million from Claude's crawler (December 17, 2024). None of that shows up in GA4.

Second, no-click answers. Ahrefs sees users from AI search click far less than users from classic organic search.

Fewer clicks mean more answers read in silence. GA4 only measures what arrives at your site. Server logs show the bots. Citation tracking shows the answers. Our guide to AI crawler logs explains how to read them.

Overlay four views on the same pages: bot fetches from server logs, human and AI-assistant sessions from GA4, a sitemap audit, and revenue from your CRM and billing.

ThriveStack citedby does this in one place. It connects to GA4 and reads human and AI-assistant traffic. It overlays the AI bot crawls from your server logs on the same pages. It audits your sitemap to show which pages are crawled, indexed and fetched by AI bots. And it ties the sessions to revenue.

The overlay tells you what to do next. Compare the three columns for each page:

• Bots but no buyers. AI reads the page, yet few buyers arrive. Improve the answer and the click path.
• Buyers but no bots. Visits come from elsewhere, and AI has not fetched the page. Check robots.txt, the sitemap and indexing.
• Bots and buyers, no revenue. The page gets attention but no deals. Fix the offer and the CRM join.

Know your bots before you read the logs. OpenAI uses OAI-SearchBot for ChatGPT search, GPTBot for training and ChatGPT-User when someone asks a question (OpenAI). Perplexity uses PerplexityBot and Perplexity-User (Perplexity). Anthropic uses ClaudeBot, Claude-User and Claude-SearchBot (Anthropic).

Work in three blocks: see AI in GA4, add the bots, then join to revenue.

Days 1 to 7. Build the AI search channel group. Save a Traffic acquisition view. Add the signup question.

Days 8 to 20. Pull the AI bot requests from your server logs. Run a sitemap audit. List the pages bots skip.

Days 21 to 30. Save the first source in your CRM. Match deals to invoices. Review the overlay page by page.

Read our AI visibility playbook for the content side, and our AI visibility metrics guide for what to report. The AI visibility tracking definition covers the citation layer.