The "Zero-Click" Reality: Groundbreaking Study Quantifies the Impact of Google’s AI Overviews on the Open Web

In the ongoing battle between search engine giants and the digital publishing ecosystem, a new study has provided what experts are calling the "gold standard" of evidence regarding the impact of AI-generated search results. A randomized field experiment, published on the Social Science Research Network (SSRN) and authored by Saharsh Agarwal (Indian School of Business) and Ananya Sen (Carnegie Mellon University), has delivered the first definitive causal proof that Google’s AI Overviews (AIO) act as a significant barrier between users and publisher websites.

The study, which meticulously tracked the behavior of 1,065 desktop Chrome users, reveals that the presence of AI-generated summaries at the top of search results pages leads to a sharp decline in traffic to organic links, while simultaneously failing to deliver the "higher quality" user experience that Google has long touted.

Main Facts: A Controlled Look at Search Behavior

The experiment, conducted in early 2026, moved beyond the observational data that has dominated industry discourse for the past two years. By utilizing a custom-built browser extension, researchers were able to create a controlled environment where users were randomly assigned to different search experiences.

The findings are stark: When an AI Overview was present, outbound organic clicks fell by 39.8%, while the rate of "zero-click" searches—sessions where the user ends their search without visiting an external website—surged by 34.5%. Crucially, clicks on sponsored search results remained unaffected, suggesting that AI Overviews serve specifically to cannibalize organic traffic rather than shifting user behavior across the entire page.

Unlike previous studies that relied on comparing traffic patterns before and after feature rollouts, this experiment isolated the variable of the AI Overview itself. By hiding the feature for a treatment group and allowing it for a control group, the researchers were able to prove that the loss of traffic is a direct result of the AI summary’s presence.

Chronology: The Road to the 2026 Findings

The debate over AI-driven search has escalated rapidly over the last 18 months, moving from industry whispers to formal regulatory investigations.

• June–September 2025: Antitrust pressure mounts as groups like the Independent Publishers Alliance and Foxglove file formal complaints with the European Commission. Penske Media Corporation initiates a major federal antitrust lawsuit in the U.S., arguing that Google’s "forced choice" destroys the fair exchange of the open web.
• December 9, 2025: The European Commission officially opens a formal antitrust probe into Google’s AI content practices, focusing on the lack of compensation and the absence of viable opt-out mechanisms for publishers.
• January 7, 2026: Researchers begin recruiting participants via Prolific, launching the two-week behavioral observation window.
• February 10, 2026: Data collection concludes. Researchers begin the rigorous process of filtering for data integrity, excluding bot-like behavior and internal Google quality-control raters.
• April 3, 2026: The initial paper is published on SSRN, providing the first randomized causal evidence in the AI search debate.
• June 17, 2026: The paper receives its final revision, solidifying the findings and strengthening the methodological defense against potential critiques.

Supporting Data: Dissecting the "Quality" Myth

One of the most persistent defenses offered by Google is the claim that AI Overviews improve the "quality" of traffic by filtering out low-intent users. The Agarwal-Sen study put this claim to the test using three specific engagement metrics: bounce rates (sessions under 10 seconds), the use of the browser’s "back" button, and total time spent on the destination site.

The results offered no support for Google’s assertion. There was no statistically meaningful difference in engagement quality between clicks generated in the control group (with AIO) and the treatment group (without AIO). If the "higher quality" theory held water, one would expect users who reached websites via traditional links in the absence of AI summaries to be less engaged. Instead, they performed identically to those who reached websites despite the AI Overview.

Furthermore, user satisfaction surveys mirrored these findings. Participants reported no difference in ease of use, information quality, or general satisfaction when the AI Overviews were hidden. This directly contradicts the narrative that AI-generated summaries provide a superior user experience.

The study also highlighted the "position zero" effect. Over 87% of AI Overviews appear at the very top of the page. The study found that the click-loss effect was overwhelmingly concentrated in this top slot, showing an 88% relative increase in outbound clicks when the AI summary was removed.

Official Responses and the Industry Narrative

Google’s leadership has historically taken a defiant stance against claims that its AI features harm the ecosystem. Liz Reid, Google’s VP of Product for Search, has previously characterized AIO-driven traffic as "higher-quality," while other executives have argued that "the web is thriving" despite these shifts.

However, the Agarwal-Sen paper’s findings sit in sharp tension with these public statements. Industry research from firms like Ahrefs has documented a worsening trend, with traffic losses for top-ranking pages jumping from 34.5% in 2025 to 58% by early 2026. While Google has previously dismissed such studies as having "flawed methodologies," the randomized, experimental nature of the Agarwal-Sen study makes it significantly harder to ignore. By controlling for variables that observational studies cannot, this research provides the empirical backbone that regulators have been seeking.

Implications: The Future of the Open Web

For marketers, publishers, and SEO professionals, this study serves as a critical reality check. The assumption that AI Overviews represent a "new normal" that can be navigated through simple optimization is being challenged by the data.

1. The Death of the "Quality" Argument

The finding that click quality does not improve with AI-filtered traffic removes a key pillar of the defense used by tech platforms. If publishers are losing nearly 40% of their organic traffic without seeing a compensatory increase in the "value" of the remaining visits, the economic argument for mandatory compensation or licensing fees becomes significantly stronger.

2. Regulatory Leverage

With the European Commission and various U.S. federal courts currently weighing the future of AI search, this paper provides a "smoking gun" for regulators. It confirms that the AI Overview is not merely a feature, but a structural intervention that diverts traffic away from the creators of the information it summarizes. The study provides a clear, defensible basis for regulators to mandate opt-out mechanisms or profit-sharing models.

3. Strategy for Publishers

Publishers are now left to grapple with a shrinking pool of traffic. The study suggests that content strategies based on the hope of "being cited" by AI may be insufficient. As Google’s AI trigger rate—currently at 41% in the study—continues to rise, the aggregate impact on the open web could prove existential for business models that rely on ad-supported traffic.

4. Methodological Precedent

Finally, the use of the "Webmunk" tool and randomized field experimentation sets a new bar for how researchers should measure the impact of algorithmic changes. Future debates will no longer be satisfied with simple "before and after" traffic logs. As the legal battles in the U.S. and Europe continue to unfold, this study will likely serve as the primary exhibit in the argument that the current trajectory of search-driven AI is fundamentally incompatible with the sustainability of the open web.

In conclusion, the Agarwal-Sen paper transforms the conversation from one of speculation to one of hard, causal fact. Google is no longer just "evolving" the search experience; it is demonstrably re-routing the flow of the internet, and the costs of that re-routing are now, for the first time, clearly visible on the balance sheets of the web’s original architects.