Key takeaways
- GA4’s AI Assistant channel covers traffic from named assistants; it excludes Google AI Overviews and AI Mode.
- Google reports clicks from its AI search features inside Search Console’s web search data, with dedicated generative-AI reporting still rolling out.
- Use session-scoped dimensions for visits and event-scoped dimensions for attributed key events; they answer different questions.
- Preserve raw referrer and landing-page evidence before creating a neat executive summary.
01
Start with the reporting boundary
A new channel label can create the comforting impression that the measurement problem has been solved. It has not. Google Analytics now documents an AI Assistant default channel for referrals from services such as ChatGPT, Gemini, DeepSeek, Copilot, and Grok. The same documentation explicitly excludes Google’s AI Overviews and AI Mode.
Those Google experiences are counted within Search Console’s web search reporting. Google also announced dedicated generative-AI performance reports in June 2026, initially for a subset of sites. The practical consequence is simple: one dashboard cannot yet represent the whole AI-discovery journey without joining several kinds of evidence.
02
Do not mix acquisition scope with attribution scope
Session source tells you what originated a visit. Event-scoped source and channel dimensions assign credit to a key event under the property’s selected attribution model. A person can arrive from an assistant, return through branded search, and convert after an email. “Where did this session begin?” and “Which interactions received conversion credit?” are both legitimate questions, but they are not interchangeable.
Swipe to compare every column
| Question | Evidence to use | What it cannot prove |
|---|---|---|
| Did an assistant refer this visit? | GA4 session AI Assistant channel and source | Whether an answer mentioned the brand without a click |
| Did Google AI Search expose this page? | Search Console web and available generative-AI reports | Which passage caused the citation or click |
| Did the visit contribute to a key event? | GA4 event-scoped attribution and path reports | Incremental causal impact |
| Did discovery create pipeline? | CRM source history, timestamps, and opportunity data | Unobserved influence without a declared method |
03
Build a reconciliation view before a performance score
Export landing pages, sessions, key events, and source detail for the AI Assistant channel. Compare that with Search Console page and query changes, then join qualified inquiries or opportunities using stable identifiers where consent and policy allow. Keep a row for unattributed and unknown traffic; forcing every visit into a confident bucket only hides the measurement gap.
Annotate channel-definition changes and the date a site gains access to any new Search Console report. A sudden line on the chart may represent instrumentation, classification, or product rollout rather than a genuine demand shift.
04
Report decisions, not a ceremonial AI traffic number
The useful monthly question is not “How much AI traffic did we get?” Ask which landing pages attracted qualified visits, which answer themes preceded useful actions, where reporting is incomplete, and what editorial or product decision follows. If assistant referrals are small but unusually qualified, improve the destination. If Search visibility rises without useful visits, inspect the promise and the next step rather than celebrating impressions.
Primary sources and further reading
Use the source material to validate details against your own context and current platform configuration.
- Google Analytics: Default channel group
- Google Analytics: Scopes of traffic-source dimensions
- Google Search Central: AI features and your website
- Google Search Central: Generative AI performance reports
This field note follows the XenGrowth editorial policy: primary sources where available, visible limitations, material review dates, and no invented first-hand experience.
Stay with the problem
