Alex Galinos, a contributing analyst for Search Engine Land, tracked 51,200 AI Overview referral events for one transportation-industry brand over nine months. He found that 22.4% of those events were logged in Google Analytics 4 as Direct traffic rather than Organic Search. That is 11,468 events reassigned to the wrong channel, and it means any AI Overview impact measured from analytics alone is being systematically undercounted. This publication reported on Aug. 14 that a separate study put AI referrers at just 1.1% of publisher visits. The two findings point the same way: referral and channel data cannot see this traffic cleanly.

The tracking method relies on a URL quirk: Google occasionally tacks a #:~

= fragment onto the link a user clicks from inside an AI Overview citation. Galinos set up a matching GA4 custom dimension: any session that lands carrying the fragment gets flagged and grouped by content theme, working around Search Console’s continued lack of a dedicated AI Overview line.

The fragment is not exclusive to AI Overviews. People Also Ask boxes and Featured Snippet results carry the same identifier, so some of the tracked events could originate there instead. Galinos checked Ahrefs and found the brand appeared in fewer than 30 Featured Snippets, so he treats AI Overviews as the dominant source given the event volume, though he calls the signal imperfect. The share metric also compares event-scoped data against session-scoped totals, a mismatch he describes as directionally accurate rather than exact.

That 22.4% figure is the dataset-wide average: more than 50,000 tracked AI Overview events across nine months, from September 2025 to June 2026. The rate swung sharply by month. AI Overview clicks landing in Direct hit 29.3% in May 2026, the highest recorded, and fell to 16.8% in April 2026, the lowest.

A second metric in the same dataset asks a different question: what share of all organic sessions originated from AI Overviews, rather than what share of AI Overview traffic was misattributed. That figure came out to 7.53% across the nine-month window, a different population from the 22.4% misattribution rate and not comparable to it directly. The share moved substantially over time. It peaked at 16% to 17% in February and March 2026, then fell to 2% to 4% more recently, a swing specific to this one brand’s SERP exposure rather than a confirmed industry trend.

The dataset also showed steep citation concentration. The single best-performing snippet drove 2,276 of the 51,200 events, against an average of 31 across all 1,661 tracked snippets. Structured formats such as transfer times, pricing pages, and HTML comparison tables outperformed general editorial content in citation frequency.

This is a single practitioner’s own tracked analytics for one brand, published as a contributor analysis rather than an independent Search Engine Land study or a vendor report with disclosed methodology. It has not been independently replicated. It covers one transportation-sector company, not a market-wide sample.

Teams sizing AI Overview exposure from Search Console or default GA4 channel groupings should build the same #:~

= dimension before trusting that number, and should treat whatever share they find as a snapshot of one moment rather than a baseline to plan a quarter around.

Search Engine Land published this contributor analysis by Alex Galinos on Aug. 18, 2026.