AI Referrals
Understand visits from recognized AI sources and the limits of attribution.
AI Referrals reads saved Google Analytics source/medium evidence. It helps answer “How many measured sessions arrived from recognized AI sources?” rather than “How many times did an AI engine mention us?”
Read sessions and share together
AI-source sessions are the numerator. All sessions in the same source/medium report form the denominator for the share.
The public table shows AI sources only. It does not show every source used in the denominator. Do not sum the visible rows and treat that as total site traffic.
Check the date window
Choose a saved, supported window and inspect its coverage. If the requested range is unavailable, use the available data explicitly rather than presenting a different period as the requested result.
A provider failure and zero observed AI referrals are different states.
Keep visibility separate from visits
An AI answer can mention or cite your business without sending a visit. A measured referral can also arrive outside the particular prompt sample your audits cover.
Use Visibility to inspect answers and citations. Use AI Referrals to inspect attributed visits. The datasets can inform a shared investigation, but they do not automatically establish a one-to-one conversion path.
Use the result
When one source changes, inspect the date range and provider coverage first. Then review relevant pages, demand and visibility evidence before choosing an Action.
Treat untracked or unavailable referral information as a limitation. Do not turn it into a claim that nobody reached your website through AI.