Google revealed at I/O 2026 that AI Overviews now has over 2.5 billion monthly active users. What he didn’t reveal is how these users actually behave once a preview appears. New data from GWIthe consumer research company whose surveys cover 3 billion people worldwide, is filling that gap – and the results challenge some of the assumptions on which SEO practitioners have built their strategy.
GWI’s most actionable number is one the industry hasn’t been talking about. Among users who interact with AI-powered search on a daily basis, 50% click on one of the sources cited. Among users who only engage once a week or a few times a month, that number drops to 28%. Among those who use it less often, that figure drops to 14%.
This discrepancy is not a minor variation. That’s a 3.5x difference in click behavior between your most frequent and least frequent visitors – and it has direct implications for where investment in content pays off.
The users most likely to click are also the ones who review most actively
Chris BeerSenior Data Analyst at GWI, proposed an additional layer that makes frequency data more useful. When asked whether younger and older users experience AI insights differently, Beer noted a trend that runs counter to the assumption that younger users are simply more comfortable with AI-generated responses.
“Younger users are more likely to say that AI previews increased their confidence in search results, but they are also more likely to say it also decreased their confidence,” Beer said. “The takeaway is that younger users appear to be more actively evaluating the role of AI in search, whether positively or negatively, while older users are more likely to remain neutral or indifferent.”
Active assessment – not passive acceptance – is what the high click-through rate among daily users reflects. These are not users who trust AI insights so much that they stop there. These are users who have developed the constant habit of using the Overview as a starting point and the cited source as a destination. For content strategists, this behavior pattern is the signal worth optimizing.
The broader research shift GWI is seeing
Beer’s response to a question about how GWI will track the next wave of changes in AI research served as a reminder that AI insights are not the only moving variable. Social research has grown considerably Over the past five years, 35% of Americans now use social platforms to find information online, up from 30% in 2020.
This five percentage point change is not dramatic in itself. But this happens alongside Deploying the AI presentationTHE AI mode extensionand the Gemini integrated search interface Google announced at I/O – all simultaneously. Practitioners who strategize around a single variable, whether traditional rankings, AI citation presence, or social discovery, map a route through a city that rebuilds its entire network.
Two Steps Practitioners Can Take This Week
GWI frequency data points to two specific actions, neither of which requires waiting for Google’s next announcement.
The first is to identify which of your pages are already cited in AI Previews and run a simple test: Do these pages offer significantly more depth, specificity, or expert perspective than the Preview Summary that cites them? If the answer is no – if clicking on an AI preview lands the user on content that essentially repeats what the preview already said – then the 50% of daily users who click on quotes won’t find anything worthwhile, and your bounce rate on AI-referenced traffic will reflect this. The solution is to add a layer of original content on every cited page that AI can’t reproduce from your existing text: a specific data point from your own measurement, a direct quote from an expert, a named case study, or a step-by-step process that goes beyond the conceptual level.
The second step is to stop process AI quote presence and social research as distinct areas of work. GWI data on social search growth means that a piece of content that appeared in an AI preview, shared on LinkedIn, and discovered through social search does not constitute three separate distribution events. This is content reaching an audience through three channels that users use in parallel. Content that works in all three tends to share a common characteristic: it answers a specific question with a specific answer, not a general topic with a general comment.
Daily audiences are most likely to visit your site
GWI frequency data makes the issues of this specificity visible. Daily users click at 50%. Casual users click at 14%. The difference is not the AI. It’s the audience. Everyday users are those who have already decided that AI-augmented search is worth it. They’re also the most likely to follow a citation to your site, evaluate what they find there, and form a lasting judgment about whether your content is worth checking out.
More resources:
Featured image: Prostock-studio/Shutterstock
