Full Answer
GA4's AI Assistant channel works by pattern-matching the HTTP referrer against a list of known AI platforms — ChatGPT, Gemini, DeepSeek, Copilot, Grok, and others. When a visitor clicks through from an AI-generated answer, the referrer header identifies the source and GA4 slots the session into the new channel. That mechanism solves one problem well: it separates AI-cited traffic from organic search so you can see which content AI engines are recommending.
But agentic commerce operates differently. An AI shopping agent that places a WooCommerce order through MCP sends an API request, not a browser click. There's no referrer header because there's no browser navigation. A browser-based agent like Operator uses a standard Chrome user-agent string, so GA4 can't distinguish it from a human session either. Both types land in your direct or organic buckets, invisible to the AI Assistant channel.
The gap matters because agent purchases convert at rates that dwarf human traffic. If those conversions sit in your direct channel, attribution models over-credit direct while the actual source — an AI agent acting on a user's instruction — goes untracked. Closing that gap requires tagging agent sessions at the server, before GA4 ever sees the event. For more on why GA4's AI traffic regex falls short, see Your GA4 AI Traffic Regex Stopped Working.