Google Analytics 4 has added AI Assistants as a new default channel in the Default Channel Group. The change was first confirmed in the Analytics Help Center’s “What’s New” documentation on May 13, 2026, and reports of it appearing live in properties followed through late May and into June, with more properties continuing to see the channel populate as the gradual rollout continues.
Google’s official description of the feature states: site owners can now identify how users are discovering their site through chatbots like ChatGPT, Gemini, and Claude via a new AI Assistant channel in their Default Channel Group reports.
What Changed
Before this update, traffic from AI chatbots like ChatGPT, Gemini, and Claude landed in the Referral bucket by default, mixed in with links from news sites, blogs, and every other external source. There was no clean way to separate “a person clicked through from ChatGPT” from “a person clicked through from a random website” without building a custom channel group with regex rules. Last year, Google’s own recommended workaround for this involved manually creating a custom channel group, writing regex patterns to match known AI referrer domains, and reordering that channel above Referral so it would be matched first. Most properties never set this up, meaning a growing traffic source stayed invisible.
Now, when GA4 detects a referrer matching a recognized AI assistant, it automatically classifies that session under a new AI Assistant channel, sitting alongside Organic Search, Paid Search, Direct, Referral, and the other default channels.
How GA4 Classifies AI Assistant Traffic
The update touches three traffic source dimensions at once, and all three happen automatically with no configuration required:
- Medium: assigned the value “ai-assistant.”
- Channel Group: the session is grouped under “AI Assistant” in Default Channel Group reports
- Campaign: assigned the reserved label “(ai-assistant)”, following the same convention Google uses for system values like “(direct)” or “(not set)”
Google’s official guidance states that AI Assistant is the channel by which users arrive at a site from sources like AI chatbots, and that it specifically excludes Google’s own AI Overviews and AI Mode, which continue to be counted under Organic Search.
Which AI Assistants Are Recognized
Google has named ChatGPT, Gemini, and Claude as confirmed examples of recognized AI assistant referrers. Google has not published a full, exhaustive list of every platform covered, so site owners won’t know with certainty whether other tools (such as Perplexity, Copilot, or others) are included until they see traffic from those sources land in the new channel.
What’s Captured and What Isn’t
The AI Assistant channel only works off the referrer header. That means:
Captured: Sessions where a user clicks a link inside an AI assistant’s response and arrives at the site with a referrer header that GA4 recognizes — for example, traffic referred from chatgpt.com, gemini.google.com, or claude.ai. These sessions get tagged with source/medium values reflecting the originating platform alongside the new “ai-assistant” medium and “(ai-assistant)” campaign label.
Not captured: Sessions without a referrer header still fall into Direct. This commonly happens with in-app browsers, mobile apps, and when users copy and paste a link rather than clicking through directly. Industry estimates suggest only 60% to 80% of actual AI-originated visits carry a clean referrer header, meaning a meaningful share of AI traffic will still be invisible inside Direct.
Also not captured here: Google’s own AI Overviews and AI Mode traffic. That stays inside Organic Search, and is instead tracked through the separate Generative AI performance reports in Search Console.
Where to Find It in Reports
The new channel appears in any report that uses the Default Channel Group dimension, including:
- Acquisition > Traffic acquisition
- Acquisition > User acquisition
- Advertising reports
- Custom explorations that include the Default Channel Group dimension
Within these reports, look for “AI Assistant” as a row in the channel breakdown — specifically under Reports > Acquisition > Traffic acquisition > Session default channel group. You can also segment by Source or Source/Medium to see exactly which platform (chatgpt.com versus gemini.google.com, for example) is sending the traffic.
Important: It’s Not Retroactive
This is the detail that trips up most reporting. The new channel only applies going forward from when the rollout reaches a given property. Historical AI traffic that was previously bucketed under Referral or Direct stays exactly where it was. It will not be reclassified.
This creates a real complication for month-over-month and year-over-year comparisons that span the rollout date. Organic Search and Referral may appear to drop simply because traffic is now being routed into the new AI Assistant channel, not because actual visits declined. Anyone reviewing trend reports across this period needs to read the numbers with that shift in mind.
Should You Still Use a Custom Channel Group?
Even with the native channel live, there are good reasons to keep or build a custom channel group for AI traffic:
- To track additional AI platforms that may not be on Google’s recognized list yet
- To create a broader “Generative AI” grouping that combines the native AI Assistant channel with other LLM-driven sources you track manually
- To bridge the retroactivity gap, since a custom channel group with the right regex conditions can still classify historical traffic the native channel can’t touch
If building a custom channel group, Google’s documentation outlines the general process: go to Admin > Data display > Channel groups, create a new group based on the default, add a channel with a “matches regex” condition covering AI referrers, and make sure it’s ordered above Referral so AI traffic gets matched first rather than falling through to the generic Referral bucket.
Step-by-Step: Setting Up a Custom “AI Assistants” Channel Group
This walkthrough covers the manual setup, useful for catching platforms not on Google’s native list, or for reclassifying historical data.
Step 1: Open Channel Groups Go to Admin (gear icon, bottom left) > under the Data display column, click Channel groups.
Step 2: Create a new group Click Create new channel group. You’ll be given the option to start from scratch or copy the existing Default Channel Group. Copying the default is recommended, since it preserves all your existing channel definitions and lets you simply insert a new rule above the rest.
Step 3: Add the AI Assistants channel Click Add new channel and name it something clear, such as “AI Assistants” or “Generative AI.”
Step 4: Define the matching condition Set the condition type to “Session source / medium” or “Session source” matches regex, and enter a regex pattern covering known AI referrer domains. An example pattern:
chatgpt\.com|openai\.com|gemini\.google\.com|
claude\.ai|perplexity\.ai|copilot\.microsoft\.com|
bard\.google\.com
This pattern can be expanded over time as new AI platforms emerge or as you identify additional referrers showing up in your raw Source/Medium reports.
Step 5: Reorder the channel Drag the new “AI Assistants” channel so it sits above Referral (and ideally above Organic Search, depending on how you want edge cases handled). Channel groups are evaluated top to bottom, and a session is assigned to the first channel whose condition it matches. If “AI Assistants” sits below Referral, sessions matching both conditions will be caught by Referral first and never reach your AI rule.
Step 6: Save and apply Save the channel group. To use it in reports, open any Acquisition report (for example, Traffic acquisition), and change the primary dimension from “Session default channel group” to your new custom channel group’s name.
Step 7: Validate with Source/Medium After a few days of data collection, cross-check your new channel against the raw Session source/medium report. Look for any AI-related domains still falling into Referral or Direct that your regex might be missing, then update the pattern accordingly.
Practical Example
Suppose a SaaS site sees the following sessions in its raw Source/Medium report over a week:
- chatgpt.com / referral — 340 sessions
- gemini.google.com / referral — 95 sessions
- perplexity.ai / referral — 60 sessions
- (direct) / (none) — 2,100 sessions
- google / organic — 8,400 sessions
With only the native AI Assistant channel active, the chatgpt.com and gemini.google.com sessions (435 total) move automatically into the AI Assistant channel, since both are on Google’s confirmed list. The perplexity.ai sessions (60) remain in Referral, since Perplexity isn’t a confirmed name on Google’s published examples.
By adding a custom channel group with the regex pattern above, ordered correctly, those 60 Perplexity sessions also get pulled into a unified “AI Assistants” view, alongside the natively-classified ChatGPT and Gemini traffic, giving a more complete 495-session total for AI-driven acquisition that week. Some portion of the 2,100 Direct sessions may also be AI-originated traffic that arrived without a referrer header, but this can’t be isolated through channel grouping alone; that gap requires separate methods such as UTM tagging on any links shared by AI tools, where possible.
What This Means for Marketers
This update doesn’t solve AI traffic attribution outright, but it removes the manual setup barrier that meant most properties simply had no visibility into how much traffic AI assistants were sending. For the first time, every GA4 user has a free, comparable baseline for measuring AI-driven acquisition without writing a single line of regex.
The practical next steps:
- Check Acquisition reports for the new AI Assistant row once the rollout reaches your property
- Compare AI Assistant traffic against Organic Search (which still includes AI Overviews and AI Mode) to get a fuller picture of total AI-influenced visibility
- Use Source/Medium to identify which AI platforms are actually sending visitors
- Treat any trend comparisons that cross the rollout date with caution, since the shift from Referral to AI Assistant can distort period-over-period numbers
- Decide whether a custom channel group is still worth maintaining for broader coverage or historical continuity
The Bottom Line
For years, AI chatbot traffic was effectively invisible inside Google Analytics, lumped together with every other referral source. The new AI Assistant default channel changes that by giving every property a built-in way to see how much traffic ChatGPT, Gemini, and Claude are sending, no configuration required.
It’s not a complete picture. Referrer-stripped traffic still lands in Direct, the channel isn’t retroactive, and Google hasn’t disclosed its full list of recognized platforms. But combined with the Generative AI performance reports now live in Search Console, marketers finally have two complementary, native tools for measuring how generative AI is reshaping where their traffic comes from.
