Google built an analytics channel that names your competitors and hides itself.
In the GA4 Default Channel Group, the AI Assistants channel captures visits from sources like ChatGPT, Gemini, DeepSeek, Copilot, and Grok. Google's own documentation then draws the line in one sentence: the channel "excludes Google's AI Overviews and AI Mode" (Google Analytics Help). Traffic from Google's own AI surfaces gets no row of its own. It merges into google / organic and cannot be separated from standard search clicks (GrowthX).
The incumbent way
Most teams read that new row as the AI number. Google added the native AI Assistants channel to the GA4 Default Channel Group on May 13, 2026, and marketing leaders have been dropping the line straight into board decks as the answer to how much traffic AI search sends them (GrowthX). The row is clean, it is labeled, and it sits next to Organic and Paid like every other channel, so it gets treated like every other channel.
The incumbent move is to accept the label at face value and report the AI Assistants line as total AI impact. That reading is not merely incomplete, a gap we have already mapped. It leans in one direction. Every assistant on the list is a Google competitor. The one assistant missing from the list is Google. The same rule that gives ChatGPT and Gemini a labeled channel routes Google's own AI Overviews and AI Mode into Organic Search, which GA4 defines to include them by design (Google Analytics Help). You are not looking at an AI traffic report. You are looking at a competitor traffic report with Google's own contribution painted out.
The consequence ladder
Adopt that row as your AI number this quarter, and the first cost is a distorted mix. Your deck shows AI-driven visits arriving almost entirely from ChatGPT, Gemini, and the other named assistants, because those are the only ones the channel can see. Google's AI surfaces, the ones sitting on the largest search audience on the planet, contribute a number you report as zero. Not small. Absent.
Carry that mix two quarters, and the second cost is budget that follows the distortion. If the AI Assistants line is the scoreboard, spend flows toward optimizing for the assistants on it and away from the surface the report cannot measure. You over-invest against rivals that look oversized and under-invest against the incumbent that looks invisible, and you do it with a chart that appears authoritative because Google printed the label. The same week Google was telling the market its AI features send billions of clicks, its flagship analytics product was filing those clicks where no channel report would ever surface them.
Carry it a full year, and the third cost is a governance problem. The vendor whose AI products you are trying to size controls both the traffic and the ruler. Google decides which sources get a labeled channel and which get folded into organic, and it placed its own AI on the invisible side of that line. A budget defended with "our AI channel shows X" rests on a definition set by the company selling the outcome, with its own contribution excluded by that company's own rule. That is the kind of dependency finance and procurement eventually ask about, and the default GA4 label is not an answer that survives the question.
What the evidence shows
The exclusion is Google's written rule, not an inference. GA4 assigns the medium ai-assistant only when a session's referrer matches Google's recognized list, and that list, by Google's own definition, leaves out AI Overviews and AI Mode while Organic Search is defined to include them (Google Analytics Help). GrowthX, an independent measurement practice, confirms the May 13, 2026 rollout and the core consequence: Google's AI traffic "merges into google / organic and cannot be separated from standard search clicks" (GrowthX).
The same source adds the edges that make the row narrower still. The change is forward only and does not reclassify historical data, so any comparison across the launch measures a definition, not a trend. Sessions that arrive without a referrer GA4 recognizes fall to Direct rather than AI Assistants. And Perplexity, a widely used answer engine, is absent from the recognized list entirely (GrowthX). Every one of those gaps points the same way. The labeled number is smaller than reality, and the piece it hides most completely is Google's own.
The binary
That is not an AI traffic channel. That is a competitor scoreboard with the home team erased.
Path one: keep reporting the AI Assistants row as total AI impact. Let a board decision ride on a figure that credits every rival assistant, omits Google's own AI by rule, drops referrer-less sessions to Direct, and skips Perplexity, then explain in the next budget cycle why the AI investment tracked against a number that structurally could not see the largest AI surface in the market.
Path two: treat the row as one labeled subset, not the total. Build a supplementary layer now. Add a custom channel group that captures the assistants Google omits, enforce UTM discipline on any campaign that can carry one, and capture lead source in the CRM so pipeline, not just sessions, tells you where buyers came from. Then rewrite the board slide to say what the number is: rival-assistant traffic with a known Google-shaped hole, sized against a corrected estimate rather than a default label.
The verdict
The teams that win the AI traffic conversation in 2026 will not be the ones who found the number in GA4. They will be the ones who noticed which name was missing from the list, rebuilt the measure around it, and refused to let a vendor's default label decide where the budget went.
Magnet's Attribution and Measurement work exists for exactly this gap: reconciling AI-driven traffic against one validated data layer instead of whichever platform label looks most official this quarter. If your AI reporting currently traces back to a single GA4 row, talk to Magnet about a measurement audit before it moves another dollar.
Sources
- Google Analytics Help, Default channel group definitions
- GrowthX, How to track AI referral traffic in GA4
