Why it's even worth the effort
Traffic from AI engines grew 16x between 2024 and 2026, and now makes up about 0.32% of all website traffic - up from 0.24% in 2025 and just 0.02% in 2024. ChatGPT alone accounts for roughly 74.78% of AI referral traffic, Gemini for about 11.56%, and Perplexity for about 7.23% (source: seranking.com). That share is still small relative to total traffic, but it isn't theoretical: Adobe Analytics reported that in March 2026, shoppers arriving at US retail sites from AI converted 42% better than non-AI traffic - a record high for the channel (source: business.adobe.com).
This isn't disconnected from GEO/AEO work itself, either. If you're investing in getting AI engines to cite your business, a study finding that 44.2% of ChatGPT citations come from the first 30% of a document shows that content structure matters too (source: searchengineland.com) - but without real measurement of the traffic that structure actually produces, there's no way to tell if the investment is working or you're just guessing in the dark.
Why GA4 doesn't catch this by default
The core problem is referrer behavior. Perplexity is the one major AI tool that consistently passes a Referer header when someone clicks from an answer into your site. ChatGPT only does it some of the time (for example from the browser version, but not reliably from the app), and Claude rarely does it at all. The result: a visit that genuinely arrived because of a ChatGPT answer looks in GA4 exactly like a direct visit, with no signal of where it actually came from (source: swydo.com).
GA4's May 2026 update - what it fixes and what it doesn't
On May 13, 2026, Google officially confirmed it added a new "AI Assistant" channel inside GA4's Default Channel Group, with no configuration required from site owners. When GA4 detects a referrer matching a recognized AI tool, it tags the medium as "ai-assistant" and groups the visit under that channel. Google names ChatGPT, Gemini, and Claude as examples, without publishing a full list of recognized domains. The rollout reached most accounts by June 2026.
The catch: the new channel doesn't include Perplexity at all, and it still doesn't catch visits that arrive with no referrer in the first place - by some estimates, up to 70% of AI-driven visits fall into exactly that gap. In other words, even Google's own official fix still only shows part of the picture (source: searchenginejournal.com).
How to build a custom channel group that catches more
- Go to GA4 Admin, then Data display > Channel groups, and create a new custom channel group.
- Set a rule with a regex on the Source field that matches domains like
chatgpt.com,chat.openai.com,perplexity.ai,claude.ai,gemini.google.com, andcopilot.microsoft.com. - Place that rule above the existing Referral channel so it gets evaluated first, instead of being swallowed into generic referral traffic.
- Add the custom channel as a standing report inside Exploration or the Acquisition Overview, so it sits alongside every other channel rather than living in a separate report nobody checks.
It's worth being honest about the limit here: even a custom channel group won't fix visits that arrive with no referrer at all - private-browser sessions or in-app browsers, for example. It meaningfully shrinks the blind spot, but it doesn't eliminate it.
What to actually do with this data
Once the custom channel is running, it's worth connecting it to your CRM data - if your GHL setup records lead source, you can check whether leads arriving from AI close differently than leads from organic search or a paid campaign, instead of just looking at visit counts. That's also the real way to test whether investing in GEO/AEO content is producing actual traffic, instead of relying on a gut feeling.
Want to know if your content is actually visible to AI engines?
We build websites ready for both GEO/AEO and real measurement - including tracking setup that tells AI traffic apart from regular traffic, not just a good-looking design.
Get started