Google AI Overviews Study: Lost Clicks Were Just as Valuable as Those With Summaries

A revised field experiment published on SSRN finds no measurable difference in click quality between searches with and without Google's AI Overviews — directly challenging Google's explanation for reduced website traffic.

The findings matter for millions of website owners and digital marketers who have watched organic traffic decline since Google expanded its AI-generated summaries across search results. If the lost clicks were genuinely low-quality visits, the traffic drop would be easier to accept. This study suggests the reality is more complicated — and more costly.

---

What the Experiment Found

Researchers Saharsh Agarwal and Ananya Sen conducted a randomized field experiment and updated their working paper to include new analyses of click quality, treatment switches, and query types. The revised study reported a 39.8% decrease in organic clicks when AI Overviews are shown — up slightly from the 38% reduction reported when Search Engine Journal first covered the paper in April.

To measure click quality, the authors examined three behavioural signals:

• How often users returned to the search results page after clicking a link
• How often a visit ended within 10 seconds without any interaction
• Total time spent on the destination site

None of the three metrics showed a statistically significant difference between the group that saw AI Overviews and the group that did not. Roughly 4 in 10 same-tab clicks led back to the results page in both conditions. Approximately 18% of visits ended within 10 seconds in both groups. Time on site was statistically indistinguishable across the board.

The authors conclude the result is "at odds with the view that AIOs primarily eliminate low-engagement website visits." That is a direct rebuttal to a claim made by Google VP of Search Liz Reid, who has stated that AI Overviews reduce "bounce clicks" — without releasing supporting data.

Why this matters: The absence of any quality difference across all three behavioural signals is not a marginal finding. It is a consistent pattern across independent measures — making it harder to dismiss as statistical noise.

---

Where the Traffic Loss Is Concentrated

Informational Queries Bear the Brunt

The experiment did not affect all search categories equally. When the researchers broke down results by query type, they found that informational searches carried nearly the entire effect.

AI Overviews were triggered or intended to trigger on 53% of informational queries — compared to just 15% of navigational queries and 6% of transactional ones. Removing AI Overviews from informational searches produced a measurable increase in outbound clicks per search. Navigational and transactional queries showed no statistically significant change, though the authors note those sample sizes were smaller.

For website owners who rely heavily on informational content to grow organic search traffic to their website, this finding signals a structural shift in how Google is redistributing attention — one that demands a strategic response rather than a passive wait-and-see approach.

The Position Advantage Has Never Been More Critical

A position-level breakdown adds further detail. The top three ranked results gained the most clicks when a top-of-page AI Overview was removed. The first-ranked result nearly doubled its click share under those conditions — a finding that underscores how much traffic the summaries absorb before users ever reach an organic listing.

This makes competitive ranking strategy more consequential than at any previous point in organic search. If your content is not consistently appearing in the top three positions for affected queries, the compounding effect of AI Overviews will disproportionately erode your visibility. Understanding how to increase traffic to your website through search is now inseparable from understanding how AI Overviews interact with position data.

The Rotation Data Reinforces the Pattern

After the initial two-week observation period, the researchers rotated group assignments. Participants who had been receiving AI Overviews saw their external clicks per search decrease. Those who stopped receiving them saw clicks rise. Zero-click rates mirrored each other in both directions — reinforcing the consistency of the pattern across time and across groups.

---

Why This Conflicts With Google's Position

Google's Argument and What the Data Shows

Google has framed AI Overviews as a feature that filters out low-value clicks while delivering higher-quality visits to websites. The argument positions the traffic decline as a net positive for publishers — fewer visits but better ones. This experiment tested that claim from the opposite direction.

If AI Overviews were primarily absorbing low-engagement visits, then the additional clicks generated by removing the summaries should look worse by quality measures. They do not. Bounce rates, short-visit rates, and time on site all remained consistent regardless of whether an AI Overview was present.

The paper cites two Search Engine Journal articles in its discussion of Google's claims — a disclosure the authors include in the revised version. For context on how Google's own tools present search performance data, it is worth reviewing the top Google tools available to help grow your business — particularly Google Search Console, which remains the most direct source of query-level traffic data available to publishers.

The Scale of Future Impact

Looking ahead, the researchers warn that aggregate traffic losses could grow significantly. As Google expands AI Overviews to appear on more queries over time, the share of searches affected will increase — and so will the cumulative reduction in outbound clicks. Because AI Overviews were already triggered on approximately 41% of queries in this experiment, the current impact is already broad. Future expansion would push that figure higher.

The paper remains a working draft on SSRN and has not yet completed peer review. Independent verification will be an important next step before the findings can be considered definitive. For further reading on the broader research landscape around AI in search, the SSRN working paper repository provides open access to the full study and related academic work in this area.

---

What Digital Marketers and Website Owners Should Do Now

For those managing organic search strategies, the study offers three practical considerations worth acting on immediately.

First, traffic declines tied to AI Overviews should not be dismissed as the loss of low-value visits. The data suggests those clicks were comparable in quality to any other organic traffic — meaning the revenue and engagement implications are real, not marginal.

Second, informational content is the highest-risk category. A strategic review of how those pages are structured is warranted — and specifically whether they are positioned to earn citation within AI-generated summaries rather than simply competing alongside them.

Third, ranking in the top three positions for affected queries has become significantly more important than it was even twelve months ago. Position one nearly doubled its click share when an Overview was removed. That is not a marginal ranking benefit — it is a structural advantage that compounds over time as AI Overviews expand across more query types.