Google’s AI Search still hides click data, leaving SEOs unable to measure traffic

Google search traffic used to be messy, but measurable. You ranked, you got impressions, you got clicks, and you could usually tell what changed. AI search breaks that loop. This week Google added more ways for people to click out from AI Mode and AI Overviews, but it still did not add the reporting publishers and marketers actually need to see whether those clicks are happening. ### What changed this week? Google announced five updates on May 6 for AI Mode and AI Overviews: follow-on article suggestions, subscription highlights, inline links, richer visual link treatments, and more previews that show where a click will go. The obvious goal is to make AI answers feel less like dead ends and more like jumping-off points to the web. That is real product movement. But none of it came with a new reporting layer for site owners. (blog.google) ### So what’s still missing? Granular click data. Google’s documentation says AI Mode now counts toward Search Console totals, which sounds helpful until you hit the catch — those totals are blended. Site owners still cannot cleanly separate classic search clicks from AI Mode clicks, compare one AI link surface against another, or tell whether a citation inside an answer outperformed a link cluster below it. (blog.google) Basically, the traffic may be there, but the measurement is still foggy. ### Why do SEOs care so much? Because attribution is the whole job. If a team cannot show which queries, pages, and search features drove visits, it cannot defend budget, prioritize content, or explain why traffic moved. Old SEO had plenty of ambiguity, but at least the core inputs were visible. AI search turns that into a black box — more exposure in theory, less proof in practice. (developers.google.com) ### Are clicks actually falling? Yes — and this is the part making everyone nervous. Pew tracked 68,000 Google searches from more than 900 U.S. adults in March 2025 and found users clicked a result 8% of the time when an AI summary appeared, versus 15% when it did not. Just 1% clicked a source link inside the AI summary itself. That does not mean every site loses the same amount, but it does mean the click pie gets smaller when Google answers more of the question on-page. (developers.google.com) ### What are publishers seeing? Publisher-side numbers point the same way. Digital Content Next said 19 member publishers saw a median 10% year-over-year drop in Google Search referrals over eight weeks in May and June 2025. DMG Media told the UK competition regulator that click-through rates for some Daily Mail queries fell from 25.2% to 2.8% on desktop when AI Overviews appeared — an 89% drop. Those are not edge-case blips. (pewresearch.org) They are the kind of numbers that change newsroom and marketing budgets. ### Doesn’t Google say the remaining clicks are better? Yes, that has been the company’s line for a while — fewer clicks, but more qualified ones. The problem is that Google has not published the evidence marketers need to test that claim for themselves. Without separated reporting, nobody outside Google can really verify whether AI clicks convert better, bounce less, or justify the lost volume. That is why this argument keeps hanging in midair. (digitalcontentnext.org) ### How does strategy change if measurement stays broken? You stop treating SEO like a game of page-count volume. Google’s own AI docs say the same core SEO practices still apply, but AI features surface a wider and more diverse set of supporting pages. That pushes teams toward denser, more authoritative assets that are easier for AI systems to cite, summarize, and trust. The KPI mix shifts too — from raw sessions alone toward citation share, branded search lift, assisted conversions, and whether your content keeps showing up in the answer layer at all. (searchenginejournal.com) ### What’s the bottom line? Google is trying to prove AI search still sends people outward by adding more visible links. But until it exposes clean click data, publishers and SEO teams are being asked to optimize for a channel they cannot properly measure. More links help. Hidden attribution still wins the argument. (blog.google) (developers.google.com)