Nine months of tracking: over 22% of AI Overview traffic is miscounted in analytics

Google still gives no clear signal in Search Console for traffic coming from AI Overview (the AI summary that appears in search results), so most brands have no idea how much of their organic traffic comes from this source. The author tracked visits via AI Overview for a brand in the transport sector from September 2025 to June 2026, logging a total of 51,200 events across 1,661 cited text snippets over that period.

The method is simple: when a user clicks a citation inside Google AI Overview, Google sometimes appends a #:~:text= marker to the destination URL. The author built a custom metric in GA4 that identifies sessions carrying that marker. The results showed that a single top snippet on its own drove 2,276 events, while the average across all 1,661 snippets was just 31 — meaning a small number of pages do most of the work.

It also emerged that snippets have their own "life cycle": some peak during a given period and then fade — usually because of seasonal shifts in demand or because Google starts favouring other content. Others only begin to climb several months after publication. Clear, structured content — prices, deadlines, comparison tables — is cited noticeably more often than general descriptive text.

The most important finding: on average, 22.4% of all events were wrongly attributed by Google Analytics to the "Direct" channel instead of "Organic Search" — 11,468 out of more than 51,000 events. The figure varied month to month: the peak misattribution came in May (29.3%) and the low in April (16.8%). That means brands may be underestimating the true scale of their organic traffic.

Overall, across the observation period an average of 7.53% of organic sessions arrived via AI Overview, though the figure is volatile: in February–March 2026 it climbed to 16–17%, and more recently it has fallen to 2–4%. The author concludes that structured, precise content wins, that keeping content fresh matters more, and that brands should account for the systemic gap in their organic-traffic reporting.