The average Netflix user spends roughly 18 minutes browsing the home screen before finally picking what to watch. They scroll past preview tiles, hover to watch short trailers, scroll back up to a show they almost selected, and then circle back to the very row they started in. For the modern consumer, the browse is a fundamental part of the overall experience.
Search has quietly evolved into something very similar. For years, the industry lacked the large-scale user telemetry data needed to prove this behavioral shift. However, recent developments in search engine results pages (SERPs) and AI-driven interfaces have changed how users interact with information. The introduction of generative AI features on search engines has fundamentally fractured the search experience into two distinct pathways: quick automated summaries and immersive browsing environments.
To understand how these changes impact user behavior, Eric Van Buskirk of Clickstream Solutions analyzed anonymized clickstream data provided by Surfer SEO. Drawing findings from approximately 846,000 U.S.-based Google search sessions collected in February and March 2026, this research represents the fifth major user-behavior study on Google’s AI features within a 12-month period.
This massive dataset builds upon previous, more localized UX studies. These include a 70-user UX study from May 2025 that utilized think-aloud protocols and screen recordings, a 250-session AI Mode study from October 2025 capturing real-time behavior inside AI Mode, and an April 2026 study of 185 high-stakes purchase queries. By trading qualitative depth for massive quantitative scale, this latest research reveals behavioral patterns that smaller studies simply could not detect.
The core finding of this study is a fascinating paradox: users behave in opposite ways depending on whether they are interacting with AI Overviews or AI Mode. While AI Mode operates like an autoplay feature where decisions are fast and consolidated, AI Overviews turn the traditional SERP into a digital version of the Netflix browse. Below, we break down the four critical behavioral shifts revealed by the clickstream data and explain what they mean for search engine optimization strategy.
1. Same User, Opposite Behaviors
One of the most striking insights from the clickstream data is that the very same search user will exhibit completely opposite behaviors depending on whether they land in AI Mode or are presented with an AI Overview on a standard SERP.
In AI Mode, the user interface acts as a closed-loop system. When a searcher uses AI Mode, they typically treat the generative response as a definitive, single-source answer. According to data from the April 2026 study of high-stakes purchases, in 88% of AI Mode tasks, users accepted the AI’s compiled shortlist exactly as presented. Furthermore, 74% of those users selected the very first item recommended by the AI, and a staggering 64% clicked on absolutely nothing at all, satisfying their informational need entirely within the chat interface. AI Mode is highly transactional, automated, and streamlined; the user reads the output, clicks a recommended link, or leaves satisfied without exploring further.
Conversely, when an AI Overview (AIO) appears on a standard search results page, user behavior shifts entirely. Instead of treating the AI’s summary as a final destination, users treat it as a springboard for a deeper, more analytical exploration of the SERP.
Cursor tracking metrics highlight this difference clearly:
- Cursor Spread: On SERPs featuring an AI Overview, cursor positions spread across the equivalent of 83% of the viewport. When no AI Overview was present, cursor spread was limited to just 66% of the screen.
- Cursor Stillness: Searchers kept their cursors completely still for 44% of their session time when an AIO was present, compared to only 29% of the time on standard, non-AI SERPs. This indicates that users are actively pausing to read, process, and evaluate the information on the screen rather than blindly scrolling past it.
This data suggests that AI Overviews turn the SERP into a highly visual, multi-option comparison environment. The user does not just click the first link. Instead, they read the generated summary, pause, weigh their options, look down the page at organic listings, return to the top, and carefully reconsider their path before making a final click.
For digital marketers and search strategists, the implication is clear: optimizing for AI Mode and optimizing for AI Overviews are two completely different objectives. Winning in AI Mode is a model-layer visibility problem; your brand must be deeply ingrained in the training data and fine-tuning layers to appear in a closed-loop shortlist. Winning in AI Overviews, however, is an on-page comparison problem. Your listing must stand out visually and contextually to win the click while the user is actively weighing options.
2. Half of All Scrolling Now Goes Backward
Historically, scrolling through search results was largely a downward, one-way journey. A user would scan down the page until they found a result that matched their intent, click it, and leave the SERP. If they returned, it was usually to click the next result down.
The clickstream data reveals that this linear journey is officially dead on pages featuring AI Overviews. Among search users who reverse their scroll direction on an AIO-enabled SERP, the median user now spends 47.5% of their total scrolling action going back up the page. Without an AI Overview present, that reverse-scrolling figure drops to just 27%.
When nearly half of all scrolling movement on a page is directed upward, it indicates that users are not merely scanning; they are actively re-reading and cross-referencing information. This aligns with findings from the May 2025 UX study, which identified “reassurance-seeking clicks” in 38% of AI Overview sessions. In those sessions, users would click a second or third link simply to validate what the first link or the AI summary had already told them.
The latest clickstream data shows that this verification behavior has moved directly onto the SERP itself. Instead of leaving the search engine to validate information on external websites, users are validating claims by scrolling up and down the Google results page, comparing the AI’s summary with the organic listings below it.
Think of it as the ultimate comparison loop. Your search snippet or brand listing is no longer getting a single, fleeting impression as the user scrolls past. It is likely receiving two, three, or more impressions as the user scrolls down to organic listings, gets confused or curious, scrolls back up to the AI Overview, and then scrolls back down to compare descriptions. This means the visual and textual appeal of your title tags and meta descriptions is more critical than ever, particularly during the user’s second and third passes.
3. Search Type No Longer Predicts Behavior
For more than two decades, the concept of search intent has served as the foundation of search engine optimization. SEO professionals have categorized queries into transactional, informational, navigational, local, and commercial investigations. By knowing the search intent, you could easily predict user dwell time on the SERP.
Navigational searchers, who know exactly which website they want to visit, traditionally leave the SERP almost instantly. Local searchers stay on the page much longer because they are interacting with rich map packs, reviews, and operating hours. Informational searchers fall somewhere in the middle.
However, the new cursor tracking data reveals that when an AI Overview is present on the page, search intent completely loses its ability to predict how long a user will stay on the SERP.
To put this into perspective, look at user retention on a standard SERP without an AI Overview at the 21-second mark of a search session:
- Only 12% of navigational searchers are still on the results page.
- 32% of local searchers remain on the page.
This classic 20-point spread is the baseline upon which most modern search strategies are built. We expect navigational searchers to bounce off the search page instantly, while we expect local and informational searchers to linger.
With an AI Overview present, however, this 20-point gap completely compresses. At the 21-second mark, all five major intent types (informational, local, navigational, transactional, and video) cluster tightly together, with retention rates sitting between 41.9% and 48.5%. The spread shrinks to a mere 6.6 percentage points. Regardless of why a user searched, they stay on the SERP for roughly the exact same amount of time when an AI Overview is present.
This compression of dwell-time metrics across all intent types is one of the most unexpected findings of the study. While intent still dictates physical scroll depth—with local queries jumping to the deepest scroll levels and video queries falling behind—the sheer amount of time users spend on the page is flattened into a uniform band.
This represents a massive shift for SEO measurement and planning. While intent-based segmentation remains crucial for determining what type of content to create, it is no longer a reliable framework for predicting user engagement and click-through speeds on the search results page itself. An informational query and a quick navigational brand query now demand similar amounts of on-screen attention from the user, changing how we must capture that attention.
4. Brand Searches Lost Their Shortcut
For years, a navigational search for a specific brand was considered a guaranteed, high-speed shortcut. If a user typed “Lenovo,” “Nordstrom,” or the name of a local business into Google, they were looking for a direct doorway to that brand’s website. They didn’t want to browse; they wanted to navigate.
The clickstream data shows that the presence of an AI Overview completely destroys this shortcut. Cursor activity and visual scatter for navigational queries increased by 40% when an AI Overview was generated on the page. Even when users arrive on the SERP knowing exactly which brand they want to interact with, they are pausing to sweep and evaluate the entire page first.
To understand how dramatic this shift is, consider the baseline metrics for navigational searchers without an AI Overview:
- They scored a low 19.7 on the cursor scatter metric, which measures how much a user’s mouse bounces around the viewport. This indicates highly focused, non-distracted movement directly toward the target brand link.
- Only 12% of these searchers remained active on the search page past 21 seconds.
When an AI Overview is introduced to a brand-name search, the user profile changes completely:
- Cursor scatter for navigational queries jumps to 27.5, indicating a highly distracted searcher who is scanning different elements of the page.
- A massive 45.8% of these branded searchers are still active on the SERP at the 21-second mark.
This behavior mirrors the “authority check” phenomenon observed in qualitative testing. When presented with an AI Overview, users instinctively check who the AI is citing and what it is saying about the brand before they click through to the official website. If a user searches for a specific brand and the AI Overview highlights recent product complaints, alternative competitors, or pricing summaries, the user is pulled out of their direct navigational shortcut and forced into a comparison mindset.
Simply having high brand recall and ranking first for your own brand name is no longer enough to secure an immediate click. If Google serves an AI Overview for your branded query, the searcher will evaluate what the AI says about you—and who else it mentions—before deciding to click through to your homepage.
Strategic Takeaways for Search Marketers
This deep dive into clickstream data reveals that AI has fundamentally changed how users consume information on search engine results pages. To succeed in this new landscape, digital marketers and SEO professionals must update their optimization playbooks to reflect how users actually interact with these layouts.
Optimize Snippets for the “Comparison Phase”
Because users are frequently reverse-scrolling and spending nearly half of their scroll time going back up the page, your search snippets are receiving multiple visual impressions. Writing basic, keyword-stuffed title tags and meta descriptions is no longer sufficient. Your titles and descriptions must be crafted like ad copy, designed to win a comparison shootout during a user’s second or third pass over the page. Focus on incorporating clear differentiators, trust signals, and compelling calls to value that stand out when placed side-by-side with an AI summary.
Separate Your AI Mode and AI Overview Strategies
Understand that AI Mode and AI Overviews represent two entirely different user mindsets. AI Mode optimization requires deep digital PR, strong brand entity authority, and structural optimization to ensure your brand is cited in definitive, zero-click answer blocks. AI Overview optimization, on the other hand, requires maximizing your visual footprint on the standard SERP. Focus on structured data, review stars, high-quality images, and clear informational headings that catch the eye of a user who is actively scanning and comparing options.
Protect Your Brand Space on the SERP
With navigational queries losing their direct shortcut status, brand defense is more critical than ever. Monitor what Google’s AI Overviews display when users search for your brand name. If the AI is pulling negative reviews, highlighting competitors, or summarizing outdated information, you must actively optimize your brand’s digital footprint. Ensure that the authoritative sources Google uses to generate your brand’s AI Overview—such as PR distribution networks, structured product databases, and third-party review platforms—are clean, accurate, and highly optimized.
Rethink Organic Click-Through Rate Models
Traditional SEO reporting relies heavily on predictable click-through rate (CTR) curves based solely on ranking position. These models are completely disrupted by the dwell-time compression and reverse-scrolling patterns caused by AI Overviews. Expect longer dwell times on the SERP itself and prepare for a shift where click distribution is more evenly spread across various elements of the page, rather than heavily concentrated purely on the first organic link. Adjust your search performance expectations and report on holistic SERP visibility rather than just static keyword positions.