Key points
- The Verge reports DCN says 19 of its members lost 1 to 25 percent of Google traffic between May and June 2025.
- DMG Media told the UK's CMA click-through rates fell by up to 89 percent between June and July 2025, citing AI Overviews.
- STAT's Tom Capper says to track brand mentions and recommendations as outcomes, not domain citations, and to sample prompts broadly.
- Capper suggests post-purchase surveys passed into analytics as a dimension, to compare self-reported discovery with credited channels.
Why it matters for marketers: Falling Google clicks and AI answers that recommend brands without linking mean traffic alone understates results; teams need mention, recommendation and survey data in reports.
Publishers say Google's AI search features have cut the traffic they once relied on. The Verge reports that Digital Content Next saw 19 of its members lose between 1 and 25 percent of Google traffic from May to June 2025, and DMG Media told UK regulators its click-through rates fell by up to 89 percent over June and July 2025, citing AI Overviews. Google's executives have said click volume is relatively stable and that the web is thriving.
The dispute matters beyond publishers. If AI answers summarize content and users click less, any team that reports success mainly through organic visits will see a shrinking picture, whether or not the brand is being discussed in those answers.
In a webinar covered by Search Engine Journal, Tom Capper of STAT argued that zero-click search is an attribution problem as much as a traffic problem. He recommends tracking brand entities, including sub-brands and competitors, because an answer can recommend a brand while citing other sites. Citations remain useful, he said, for finding which third-party sources, such as review sites, may shape recommendations.
He also advises building a realistic prompt sample, using the web interface rather than the API because outputs can differ, and being cautious with volatile sentiment scores. For attribution, he suggests post-purchase surveys fed into analytics, with answer options rotated so the first choice is not favored.
A practical next step is to audit current dashboards for gaps: where do they count visits or domain citations but miss mentions and recommendations? Document the prompts, competitors and collection method, then compare visibility trends against revenue and survey data. Also watch whether Google's publisher payment tests and source-link changes alter click patterns.
Sources
This brief is our own summary. Read the full reporting at the source.
