Editor's note: This blog was updated September 2026 to cover the AI Assistant channel in the Google Analytics default channel group, explain where AI traffic can land outside it, and refresh the custom channel guidance.
Generative AI tools like Gemini and ChatGPT are becoming a new and significant source of website traffic. While they’re not entirely replacing traditional search, they’re rapidly expanding how users find information, offering personalized experiences, and easily digestible content and conversational tones. To stay on top of industry trends and make data-informed decisions as users gravitate toward AI-driven search, you’ll need to segment traffic from these tools.
In the context of attribution measurement, Google Analytics 4 (GA4) now sorts visits from recognized AI assistants into their own AI Assistant channel, so they no longer get lumped in with visits from traditional blogs or websites in the Referral, Organic, or Unassigned channel. Pairing that native channel with a custom AI logic covers the sources it misses. With both in place, you can:
Getting this right matters because AI traffic numbers increasingly shape decisions about which content to invest in, and a channel that shows only part of the picture can undersell what AI discovery is doing for your site. Below, I’ll walk through how the native AI Assistant channel works, where AI-driven visits still land outside it, and how to build a custom channel that fills the gaps.
Google Analytics now recognizes traffic from AI assistants on its own. When a session's referrer matches Google's list of AI assistants, Analytics sets the medium to "ai-assistant," sets the campaign to "(ai-assistant)," and reports the session in the AI Assistant channel of the default channel group.
Google's documentation names ChatGPT, Gemini, DeepSeek, Copilot, and Grok as examples, and I’ve also seen Claude and Perplexity sessions classified this way. To see which assistants send you traffic, open the Traffic acquisition report and add Session source / medium as a secondary dimension, and search for “ai-assistant.” This will not include all AI traffic, as various AI sources may have a medium of (not set), organic, etc.
Even with the native channel in place, AI traffic can look smaller in Google Analytics than it really is. Clicks from Google’s AI Overviews and AI Mode count as Organic Search rather than AI Assistant, so some AI-driven discovery lands in your search numbers. We’ve also seen AI traffic arrive with a source of chatgpt but a medium of organic or (not set), which sends those sessions to Organic Search or Unassigned instead of AI Assistant. The native channel also depends on the referrer, so a visit from an assistant that doesn’t pass one can show up as Direct traffic, and we can’t fully tell how much of Direct really started in an AI tool. A custom channel group can recover the sessions that carry an AI source, even when the medium doesn’t match.
That’s when it becomes valuable to set up a custom channel.
How to set up a custom AI channel
If you want to catch assistants Google doesn’t recognize yet or group AI traffic differently than the default does, you can create a custom channel within Google Analytics.
Like the native channel, custom channel groups apply to your reports retroactively. The tradeoff is capacity and reach: standard properties allow two custom channel groups and 360 properties allow five, and custom channel groups don't appear in the BigQuery export, so plan where this definition will live before you build it.
First, navigate to Admin > Data Display > Channel Groups. Select 'Create new channel group', or click into an existing custom channel group to add this channel.
Next, select "Add new channel." Name your channel something like "AI Traffic" or "AI Chatbots."
Then, define AI traffic sources. One approach is to start with a condition where Default channel group exactly matches AI Assistant, then add an OR condition that uses "matches regex" on Source for anything the native channel misses. Some popular AI traffic sources include Gemini, ChatGPT, Microsoft Copilot, Claude, DeepSeek, and Perplexity. Use "matches regex" to ensure you're capturing all potential AI sources.
Below is a regex example that Google provides:
As more AI agents are popping up, here is a more inclusive list for defining AI traffic:
| PRO TIP Be sure to check this against your actual AI traffic sources to ensure the regex is capturing all traffic that should be sorted as AI. Broad patterns like .*ai can match any referring domain that ends in "ai." |
GA4 sorts through channel groupings in top-down order, so be sure to place the AI channel before Referral.
Updates to channel groupings are retroactive in the GA4 UI, so this new channel can be tested in GA4 Explore. This is done by pulling in the session source/medium and session custom channel grouping. Then, you can review the source against the updated channel rule to ensure it falls under AI traffic.
Keep in mind that now that more users are leveraging AI tools for search, you may notice a decrease in overall traffic YoY (specifically for content/publisher sites). However, segmenting out traffic from AI sources will help you tell a more accurate data story moving forward, and it pairs well with measuring your AI search visibility.
Have questions about AI and search? Check out our AI solutions or reach out to connect with our knowledgeable team.