Full Answer

Attribution in traditional analytics works because the referring site sends a header that identifies itself. AI search platforms break this contract in three different ways. Google AI Mode applies noreferrer to every outbound link, so clicks from AI Overviews arrive in GA4 as organic or direct traffic with no indication they came from an AI-generated answer (Seresa, 2026). ChatGPT’s mobile app inconsistently passes referrer data, meaning a portion of ChatGPT-referred visits also disappear into the Direct bucket. And Perplexity, while it does pass referrers reliably, embeds citations as inline numbered links within generated text — a pattern that standard UTM-based attribution was never designed to track.

The cumulative effect is a 20–40% blind spot. Visits arrive, pages are viewed, purchases happen — but your analytics cannot connect them to the AI engine that drove them. This makes it impossible to measure AI search as a channel, calculate its ROI, or justify content investment in answer engine optimisation.

Closing the gap requires a layered approach: server-side log analysis to recover stripped referrers, user-agent matching to identify AI crawler and platform traffic, and custom GA4 channel groupings that separate AI-attributed visits from genuine direct traffic. Without these additions, AI search will continue growing as a revenue source that your reporting cannot see.