The Hidden Side Of AI Overviews Click Quality: What Commerce Brands Need To Know

TL;DR — A randomized field experiment found that clicks lost to Google's AI Overviews aren't lower quality than remaining clicks—bounce rates, time on site, and return-to-search behavior were statistically identical, contradicting Google's "bounce click" defense and confirming that brands are losing genuinely valuable traffic.

The debate over AI Overviews has centered on a convenient narrative: sure, you're getting fewer clicks, but the ones you're losing weren't worth much anyway. Google has leaned into this framing, with VP of Search Liz Reid suggesting that AI summaries primarily absorb "bounce clicks"—users who would have clicked through, found nothing useful, and immediately returned to search.

New experimental data challenges that assumption directly. Researchers from Carnegie Mellon and the Indian School of Business updated their working paper in June 2026 with click quality metrics, and the results undercut Google's quality defense entirely. For commerce brands already watching organic traffic erode, this changes the strategic calculus.

The Experiment: What It Actually Tested

Unlike prior observational studies that compared different queries or time periods, this research used a custom browser extension to randomly assign 1,065 real users to see or not see AI Overviews during their normal searches. Data collection ran from January 7 to February 10, 2026, capturing actual browsing behavior rather than theoretical models.

The methodology matters because it isolates the AI Overview variable. Same users, same query types, different experiences—then measure what happens after the click.

• Back-button behavior: About 4 in 10 same-tab clicks led back to the results page in both conditions
• Quick bounces: Roughly 18% of visits ended within 10 seconds without interaction, regardless of whether an Overview was shown
• Time on site: Statistically indistinguishable between groups
• Sample scope: AI Overviews triggered on approximately 41% of queries in the study

The paper's authors wrote that the result is "at odds with the view that AIOs primarily eliminate low-engagement website visits."

What The 39.8% Click Loss Actually Represents

The headline figure from the original paper—a 39.8% decrease in organic clicks when AI Overviews appear—now carries heavier implications. These aren't junk clicks being filtered out; they're legitimate potential customers who never reach your site.

The quality analysis shows that the extra clicks websites receive when Overviews are removed behave identically to clicks with Overviews present. Users engage similarly, spend comparable time, and return to search at the same rates.

For commerce brands, this means the traffic you're not getting would have converted at the same rate as traffic you do get. The math on lost revenue becomes straightforward multiplication rather than speculative discounting.

Informational Queries Take The Biggest Hit

The updated paper breaks down impact by query type, revealing where commerce brands feel the most pressure. AI Overviews triggered on 53% of informational queries, compared with just 15% of navigational and 6% of transactional ones.

This concentration matters for brand strategy. Informational queries—"best running shoes for flat feet," "how to choose a coffee grinder"—represent the research phase where purchase intent is forming. Losing visibility during this phase means losing influence over the consideration set.

• Informational queries saw measurable click losses when Overviews appeared
• Navigational and transactional queries showed no statistically significant change (though sample sizes were smaller)
• Top three organic positions gained the most when Overviews were removed
• Position one nearly doubled its click share without an Overview present

The implications for AI visibility strategy are clear: if you're not being cited in the Overview itself, you're increasingly invisible during the research phase when brand preferences form.

Why Google's Quality Argument Needed Testing

Google's defense of AI Overviews has consistently emphasized quality over quantity. The argument goes: summaries give users immediate answers, eliminating the need to click through to sites that wouldn't satisfy their intent anyway. Users who do click now are more qualified, more intentional.

This framing served two purposes. It positioned traffic loss as a feature rather than a bug, and it suggested publishers complaining about lost clicks were defending low-value inventory.

The experimental data tests this claim empirically rather than accepting it at face value. According to Search Engine Journal's coverage of the study, the researchers found no measurable difference in engagement metrics—directly contradicting the quality-for-quantity trade-off narrative.

The study authors note that Google hasn't released supporting data for the bounce-click reduction claim. Without that transparency, the experiment provides the most rigorous available test of the assertion.

The Rotation Effect: Behavior Changes Both Directions

The researchers added a treatment-switch phase after the initial two weeks, rotating which users saw AI Overviews. The behavioral changes moved in predictable directions:

• Users newly served AI Overviews saw their external clicks per search decrease
• Users who stopped seeing them saw their external clicks per search increase
• Zero-click rates mirrored each other in both directions

This rotation design strengthens the causal claim. It's not that certain users are inherently low-clickers or high-clickers—the AI Overview presence itself changes behavior consistently.

For brands, this confirms that AI Overviews don't just correlate with reduced traffic; they actively cause it. The distinction matters when building a case for strategic investment in generative engine optimization.

What This Means For Commerce Strategy

The quality-parity finding shifts how commerce brands should think about AI Overview adaptation. If lost clicks aren't lower quality, then:

• Citation value increases: Getting mentioned in AI Overviews isn't just about visibility—it may be the only way to capture research-phase traffic that would have clicked through previously
• Content strategy refocuses: Creating content that AI systems cite becomes as important as traditional SEO for top organic positions
• Measurement frameworks need updating: Tracking brand mentions in AI responses supplements traditional click-through metrics

The agentic shopping transformation accelerates these dynamics. As AI intermediates more of the purchase journey, brands that treat AI visibility as a channel rather than a curiosity will capture the upstream influence that traditional organic used to provide.

The Peer Review Caveat

The paper remains a working draft on the Social Science Research Network and hasn't completed peer review. That's worth noting—academic scrutiny may surface methodological questions or call for replications.

That said, the methodology (randomized assignment via browser extension) is more rigorous than prior observational approaches, and the findings align with what multiple other studies have shown about traffic impact. The quality analysis adds a new dimension that previous research hadn't addressed.

For commerce brands making strategy decisions, waiting for perfect data means waiting too long. The directional evidence is strong enough to inform action.

Key Takeaways

• Lost clicks aren't lower quality: Experimental data shows identical engagement metrics between clicks with and without AI Overviews, contradicting Google's bounce-click narrative.
• Informational queries matter most: With 53% of research-phase queries triggering AI Overviews, brands lose influence during the consideration phase where preferences form.
• Citation is the new click: Getting mentioned in AI Overviews may be the primary way to capture traffic that would have clicked through organically.
• Measure AI visibility directly: Supplement traditional traffic metrics with tracking of brand mentions and citations in AI-generated responses.
• Don't wait for perfect data: The directional evidence is strong enough to justify strategic investment in generative engine optimization now.

The quality debate was always a distraction. The real question is whether your brand shows up when AI answers the questions your customers are asking—and this study confirms that the stakes are higher than Google's narrative suggested.