Monitoring AI Traffic: 15 Preset AI Source Sites in GA4

Somewhere in the middle of the year we started noticing visits that looked normal in GA4 but behaved oddly: short sessions, one page, then gone. Tracing the referrers, they came from AI platforms — ChatGPT, Claude, Perplexity, Gemini and a growing list of AI tools. AI traffic had quietly arrived, but GA4 does not group these sources by default. If you want to answer the question 'is our AI traffic growing?', the default reports will not show it.

The problem: AI sources are invisible in default GA4

AI traffic is real, and it behaves differently from search traffic — shorter sessions, deeper follow-ups, fewer immediate conversions. But out of the box, GA4 scatters those sessions across hundreds of individual domains or dumps them into 'unassigned'. Watching them day by day is impossible. We needed one view that answered three questions at once: how much AI traffic we get, which platforms send it, and which pages it lands on.

Some teams accept this blind spot because AI traffic is still a small slice. We did not — you cannot decide whether to invest in GEO if you cannot measure the channel it is supposed to improve.

How we solved it: 15 preset AI source sites

We maintain a preset list — an ai_source_sites table with 15 domains covering the mainstream AI platforms and AI tools. Every GA4 query that needs AI traffic applies an inListFilter that aggregates those source domains into a single multi-source view. From that view we output the daily AI traffic trend, source distribution, landing pages and regional drill-down.

Three decisions made this approach work:

• Keep the list maintainable. New AI platforms appear constantly; adding one is a one-row change, not a query rewrite.
• Aggregate across sources. Reading a single referrer row understates the whole channel — AI traffic is a sum, not a line item.
• Trends beat daily absolutes. One day of AI visits tells you nothing; the slope of the weekly trend is what tells you whether to invest in GEO.

In practice this is a standing weekly report, not an ad-hoc query. The same view feeds our regular reporting and our GEO decisions, so the numbers stay comparable week over week.

Reusable checklist: monitoring AI traffic in 4 steps

1. Maintain a list of AI source domains — start with the well-known platforms, then add tools as they appear.
2. Run GA4 queries that aggregate the whole list with an inListFilter.
3. Review the daily trend, source distribution, landing pages and regional drill-down.
4. Review and update the list every month.

The same preset-source approach powers the AI traffic monitoring in our GEO competitive monitoring module.