How to Track AI Search Traffic in GA4: AI Assistant Guide
Measure traffic from ChatGPT, Gemini, and other AI assistants in GA4 using the right channels, sources, conversions, and visibility metrics.
Key Takeaways
5- GA4 now groups recognized AI referrals in its default AI Assistant channel, so the Traffic acquisition report is the right place to begin.
- AI Assistant includes sources such as ChatGPT, Gemini, DeepSeek, Copilot, and Grok; clicks from Google AI Overviews and AI Mode remain in Organic Search.
- GA4 measures only visits that reach your site. Brand mentions, citations, and zero-click visibility inside AI answers require separate measurement.
- ChatGPT adds utm_source=chatgpt.com to referral links from search results, allowing more detailed source analysis in GA4.
- A useful report compares sessions, landing pages, engagement, and key events instead of treating raw AI traffic as the final outcome.
The simplest way to measure traffic from AI search is to open Reports → Acquisition → Traffic acquisition in GA4 and inspect the AI Assistant row under Session default channel group. Google Analytics uses this channel for recognized referrals from sources such as ChatGPT, Gemini, DeepSeek, Copilot, and Grok.
That report is an important starting point, but it is not a complete measure of AI visibility. Someone can see your brand, read a summary of your content, or notice your page in an AI answer without clicking through. No website session means no GA4 record. A reliable measurement framework therefore separates visibility, visits, and business outcomes.
What is the GA4 AI Assistant channel?
According to Google’s default channel group definition, AI Assistant groups sessions that arrive from recognized AI assistants. When a referrer matches Google’s source list, Analytics can set the medium to ai-assistant and the campaign to (ai-assistant).
This makes AI referral reporting much easier than it was when those visits were scattered across Referral and other channels. Two limits still matter:
- The channel can measure only clicks that reach your website.
- It excludes Google AI Overviews and AI Mode, which Google classifies as Organic Search.
What GA4 can and cannot measure
| GA4 can measure | GA4 alone cannot measure |
|---|---|
| Sessions referred by an AI assistant | A brand mention that receives no click |
| Session source and medium | How often a page is cited across different prompts |
| The landing page a visitor reached | Impressions inside an AI-generated answer |
| Engagement, key events, and revenue | The visitor’s private prompt |
| Device, country, and subsequent page activity | The exact origin of a visit whose referrer was stripped |
This distinction prevents a common reporting mistake: attributing a rise in Direct traffic to AI exposure without supporting evidence.
How to find AI traffic in GA4
1. Open the Traffic acquisition report
In your GA4 property, go to Reports → Acquisition → Traffic acquisition. This is a session-scoped report. Do not confuse it with User acquisition, which describes how a user first arrived rather than the source of each new session.
2. Filter the AI Assistant channel
Use Session default channel group, or Session primary channel group when it is available in your property, as the primary dimension. Open the AI Assistant row or apply a filter for that value.
3. Break the channel down by source
Add Session source / medium as a secondary dimension. This shows which assistants are sending visits, how Google names each source, and whether a new domain is still sitting outside the expected classification.
4. Add the landing page
Use Landing page + query string to find the pages earning AI referrals. Avoid judging them on sessions alone. Compare the following metrics:
| Metric | Question it answers |
|---|---|
| Sessions | How many visits did AI assistants initiate? |
| Engaged sessions and engagement rate | Did visitors meaningfully interact with the site? |
| Average engagement time per session | How much active time did they spend with the content? |
| Key events | Did they request a demo, sign up, contact you, or buy? |
| Session key event rate | What share of those sessions reached a meaningful goal? |
| Revenue | If ecommerce is configured, how much value did the channel create? |
Weekly or monthly comparisons are usually more useful than daily reporting because small AI referral volumes can produce noisy percentage changes.
How to track ChatGPT traffic separately
OpenAI states in its Publishers and Developers FAQ that ChatGPT automatically adds utm_source=chatgpt.com to referral links from search results. In GA4, inspect Session source, Session source / medium, and Landing page + query string to analyze those visits.
Do not add an “AI traffic” UTM to your own internal links. UTMs on internal navigation can split sessions and corrupt acquisition reporting. The parameter discussed here is added to an outbound referral controlled by ChatGPT.
If you expect ChatGPT traffic but see none, check three things:
- Did a click actually occur during the selected date range?
- Does the landing-page redirect preserve the query parameters?
- Did consent settings, tag configuration, or a redirect chain prevent GA4 from recording the session?
Do you need a custom AI channel group?
For most teams, GA4’s default AI Assistant channel is the best starting point. A custom channel group is useful when a new assistant is not yet recognized or when your organization needs a stable internal reporting definition.
Google’s custom channel group guide explains how to copy a channel group under Admin → Data display → Channel groups, add a source regex rule, and place the AI channel above Referral. Order matters because a session is assigned to the first channel whose rule it matches.
A practical starting expression is:
chatgpt\.com|perplexity\.ai|claude\.ai|gemini\.google\.com|copilot\.microsoft\.comTreat this as a starting point, not an authoritative or permanent list. Inspect your actual Session source / medium values, verify each domain, and update the rule as the market changes. GA4 custom channel groups can be applied retroactively in reports, so you can use the revised definition to examine historical data as well.
How to measure traffic from Google AI Overviews and AI Mode
Google explicitly excludes AI Overviews and AI Mode from the AI Assistant channel and includes their clicks in Organic Search. As a result, GA4 cannot isolate those visits as neatly as it can isolate ChatGPT referrals.
Use a combined approach:
- Monitor Google Organic sessions and landing pages in GA4.
- Compare clicks, impressions, queries, and countries for those pages in Search Console.
- Track whether the same pages appear as cited sources in AI answers.
- Do not assume that two changes are causal simply because visibility and traffic moved at the same time.
Search Console and GA4 measure different parts of the journey. Search Console describes performance in Google Search; GA4 describes what happened after the user reached your site. Reading them together produces a more useful weekly view.
Set up conversions before evaluating AI traffic
Traffic volume is not the business outcome. Mark actions such as a qualified form submission, demo request, trial registration, purchase, or another meaningful behavior as key events in GA4.
If visitors move from your marketing site to a separate application or checkout domain, verify cross-domain measurement. Otherwise the signup or purchase may appear as a new session and lose its connection to the original AI source. Test the journey with DebugView and confirm the processed result in your acquisition reports after data has been collected.
Measure AI traffic and AI visibility together
A useful AI search framework has four layers:
| Layer | Core question | Data source |
|---|---|---|
| Access | Can AI crawlers reach the content? | Server logs and technical audits |
| Visibility | In which prompts is the brand mentioned or cited? | AI visibility monitoring |
| Visit | Which assistants send sessions to the site? | GA4 AI Assistant and source/medium |
| Outcome | Do those visits create signups, demos, or revenue? | GA4 key events and CRM data |
Brantial’s AI visibility feature tracks brand mentions and sources across answer engines, while Agent Analytics brings search and AI visibility signals into one workspace. For the access layer, see our guide to AI crawler traffic and crawl-log analysis.
A weekly AI traffic checklist
- Compare AI Assistant sessions with the previous period.
- Review source/medium for new or unexpected referrers.
- Check the landing pages receiving AI traffic and their key events.
- Compare Google Organic and Search Console trends for the same pages.
- Review changes in AI mentions and citations alongside traffic data.
- Update your custom channel regex when verified new sources appear.
Frequently asked questions
Does GA4 have a separate channel for AI traffic?
Yes. The current GA4 default channel group includes AI Assistant, which classifies clicks from AI assistants recognized by Google.
Why can ChatGPT traffic sometimes appear as Direct?
Referrer information can be lost because of a browser, app, privacy setting, or redirect chain. GA4 cannot reliably reconstruct the original source in that case, so it is inaccurate to treat every Direct session as ChatGPT traffic.
Why can AI visibility increase while traffic stays flat?
The user may get a complete answer without visiting the cited site. Mentions and citations can rise without producing the same increase in sessions, which is why visibility and traffic need separate metrics.
What is the difference between AI Assistant and Organic Search?
AI Assistant covers recognized referrals from services such as ChatGPT. Google AI Overviews and AI Mode clicks remain in Organic Search under Google’s current channel definitions.
Is GA4 enough to measure AI search performance?
No. GA4 is essential for post-click behavior and conversions, but it does not measure zero-click mentions, prompt coverage, or citations. Combine it with Search Console, server logs, and AI visibility data for a complete view.