A new clickstream study of approximately 846,000 U.S. Google sessions, analyzed by Eric Van Buskirk of Clickstream Solutions and published by Kevin Indig, surfaced a finding that should reframe how every brand thinks about visibility: when an AI Overview is on the page, even users who searched for your brand by name stop and evaluate what surrounds it before they click. The brand-name shortcut, the fastest path through Google search for twenty years, no longer works the way it used to. Brand recall got the user to the page. What the AI says decides whether they click through.
The finding that matters
For two decades, a branded search was the most predictable behavior in all of search. Someone types "Lenovo," they know where they are going, they click the Lenovo result, they leave. In the study, navigational searchers without an AI Overview were the most focused group measured: a cursor-scatter score of 19.7, the lowest of any intent type, with only 12% still active on the page at 21 seconds. They arrived with a destination and left almost immediately.
With an AI Overview present, that profile inverts. Cursor scatter for the same navigational searches jumps to 27.5, a 40% increase in cursor activity. And 45.8% are still active at 21 seconds, nearly four times the share without an AIO. The person who typed your brand name is now sweeping the page, reading the AI's summary, checking what it says about you and about the alternatives next to you, before deciding whether to proceed.
The qualitative version of this showed up a year earlier. A 70-user UX study in May 2025 found that brand and authority is the first gate users apply when reading a result page: they check who is cited before they check what is said. What the new clickstream data adds is that the gate now fires even when there is nothing to gate against. A user who typed a specific brand name is still running the authority check on the AI Overview content first. The decision moved from recall to evaluation.
Why this is a citability problem, not a SERP problem
It is tempting to read this as a SERP-layout issue and respond with title-tag and meta-description rewrites. That is the conventional move, and it addresses the wrong layer. The behavior the study documents is not about your snippet. It is about what the AI Overview says when your brand name is the query: how it describes you, what it lists alongside you, whether the summary frames you as the obvious answer or as one option among several.
That is representation, and representation is exactly what the Citability Score measures. When someone searches your brand and an AI Overview generates a summary, the model is drawing on the same structured signal that determines citation everywhere else: entity coherence, the consistency between what you declare and what third parties confirm, the clarity of your positioning, the strength of your external reputation. A brand that is structurally well-represented gets an AIO that reinforces the click. A brand that is poorly represented gets an AIO that invites the user to look around, and the study shows the user now does exactly that.
Brand recall gets the user to the page. How the AI represents you decides whether they stay with you or start comparing. That second step is citability, and it now fires on your own brand name.
This collapses a distinction brands used to rely on. Branded search was the protected channel, the traffic you owned, the query where you did not have to compete. The study shows that protection is gone. When an AI Overview sits above your brand result, you are competing for your own brand searcher against whatever the AI chooses to surface in that summary. If a competitor is better represented in the AI's understanding of your category, your branded search becomes their discovery opportunity.
The deeper shift: intent stopped predicting behavior
The brand-search finding is one face of a larger pattern in the same study. For twenty years, search intent has been the foundational segmentation model in the discipline: you could predict how long a user would stay on a results page by knowing what kind of search they ran. Navigational users left fast, local users lingered, informational users sat in between. That predictable spread was around 20 points of difference in time-on-page.
With an AI Overview on the page, the spread compresses to barely 6 points. All five intent types, informational, local, navigational, transactional, and video, cluster between 41.9% and 48.5% time-on-page at 21 seconds. Intent stops predicting how long users stay once an AIO appears. The thing that used to organize the entire optimization playbook loses its predictive power.
What replaces it is not another segmentation variable. It is representation. When intent no longer tells you how a user will behave on the page, the variable that does matter is whether your brand is well-represented in the AI layer that now mediates every query type. The study frames this as a reading-mode shift, the SERP becoming a comparison environment where users scroll backward, re-read, and validate before clicking. Roughly half of all scrolling on an AIO page now goes backward, against 27% without one. People are not scanning. They are re-evaluating. And what they re-evaluate is how each option, including yours, is presented to them by the AI.
Two surfaces, two different jobs
The study is precise about one more thing worth carrying forward. AI Mode and AI Overviews produce opposite behaviors, and they require different work. In AI Mode, users accept the answer: prior research in the same body of work found that in 88% of AI Mode tasks users took the shortlist as-is and 64% clicked nothing at all. AI Mode is a closed loop, and showing up in it is a visibility problem at the model layer. AI Overviews, by contrast, turn the page into the comparison environment this study documents, and showing up well there is a representation problem on the surface itself.
Both are citability problems in the broad sense, but they are not the same job. AI Mode rewards being in the model's shortlist at all. AI Overviews reward being the option that survives the comparison once the user starts scrolling backward. A brand that measures only one of these is measuring half of how it is discovered.
What to check this cycle
Three practical moves follow from the study. First, run your own brand name as a query in the major AI surfaces and read what the AI Overview says about you, not just whether you appear: the summary above your own result is now part of your branded search experience, and the data says users read it before clicking you. Second, look at how you are represented alongside competitors in your category's AI Overviews, because the comparison the user now performs on the page is the comparison you need to be winning before they perform it. Third, stop relying on intent as your primary behavioral predictor on AIO pages, and start treating representation quality, how clearly and favorably the AI understands and frames your brand, as the variable that determines whether the click happens.
The brand-name shortcut is gone. The user who searched for you specifically is now auditing you before they commit. The audit runs on how the AI represents you, and that representation is measurable. The brands that read their own AI Overview the way their customers now do will see the gap before it costs them the click.