How to track AI traffic in GA4—and what GA4 still misses

Define the question before opening GA4

Acquisition, first-touch discovery, and key-event attribution use different scopes and can return different answers.

Use Traffic acquisition when the question is how sessions arrived during the selected period. Use User acquisition for the source that first acquired a user under GA4’s identity and attribution rules. Use advertising attribution reports when the question concerns credit assigned to key events.

For a practical AI traffic report, begin with sessions. Write down the date range, comparison period, reporting identity, internal-traffic exclusions, key events, and whether the denominator is sessions or users before exporting a number.

Open and configure the Traffic acquisition report

The maintained AI Assistants channel provides the clearest current starting point.

Open Reports, Acquisition, Traffic acquisition. Use Session default channel group as the primary dimension and locate AI Assistants. Add Session source as a secondary dimension so the provider values behind the channel remain visible.

Add or compare Sessions, Engaged sessions, Engagement rate, Key events, Session key event rate, and Total revenue where those metrics are correctly configured. Save a comparison or exploration only after the base report matches the property’s expected totals.

• Session default channel group
• Session source / medium
• Landing page + query string
• Sessions and engaged sessions
• Key events and rate
• Revenue or qualified outcome

Add landing pages and meaningful outcomes

A provider total becomes actionable when it shows where the visit began and what happened next.

Create an exploration or customized report using Session default channel group, Session source, and Landing page + query string. Filter the channel to AI Assistants, then add the events that represent product evaluation or value rather than treating every pageview as equal.

Review the report by page type and source. A documentation page may have fewer conversions but create highly qualified trial starts; a pricing page may convert quickly; a legacy article may reveal a content problem. Keep counts visible when provider samples are small.

Preserve the current channel boundaries

Not every AI-assisted discovery surface belongs in AI Assistants.

Google documents recognized sources such as ChatGPT, Gemini, DeepSeek, Copilot, and Grok as examples of AI Assistants. Google AI Overviews and AI Mode are included in Organic Search. Keep that distinction visible when comparing channels.

Direct remains a source-less category. A custom channel can organize additional observed sources, but it cannot identify an origin that the browser or app never sent. Do not merge Direct into AI Assistants or relabel all Google organic visits as AI-generated traffic.

Validate classification with controlled visits

A report is more trustworthy when the team knows how its own properties receive the handoff.

Follow a public link from each relevant assistant in a controlled browser and record the landing URL, observed source and medium, default channel, device, client type, and timestamp. Repeat across web and app surfaces only where testing is practical and permitted.

Use DebugView or Realtime for collection checks, then confirm the processed session in standard reporting after data is available. Preserve screenshots or exports in the analytics change record rather than hard-coding a temporary source list into a permanent policy.

• Provider and client surface
• Destination URL
• Session source / medium
• Default channel
• Landing page
• Test date and result

Troubleshoot a missing or unexpected source

Rule out collection and handoff problems before concluding that traffic disappeared.

Check consent changes, tag deployment, cross-domain configuration, redirects, URL shorteners, self-referrals, unwanted referrals, campaign parameters, browser mix, and app-to-browser transitions. Compare the change with landing pages and raw request evidence where your policy allows it.

If GA4 recognizes a provider differently from a custom taxonomy, document both rule versions and avoid double counting. If a session has no source evidence, keep it Direct while investigating adjacent signals such as citations, brand search, and surveys separately.

Publish the report with a coverage statement

The conclusion should describe what GA4 observed and what it could not recover.

Present observed AI Assistant sessions, landing pages, meaningful outcomes, conversion counts, and value. Compare with Organic Search and other channels under the same definitions. Add the date, property, scope, exclusions, and any custom rule version.

Include one sentence explaining that source-less and cross-device journeys are not recoverable from GA4 alone. This makes the result useful without suggesting that the reported total represents every visit influenced by an AI answer.

Observed AI traffic = sessions classified as AI Assistants under the stated GA4 or validated custom rule
Coverage note = source-less, copied-link, and cross-device journeys are not assigned without additional evidence