Organic clicks fell 38% when Google AI Overviews (AIOs) were present in search results. Zero-click rates rose from 54% to 72%. Users who saw the same queries without AIOs were no more satisfied than those who saw them with. Those are the findings from a randomized field experiment by Saharsh Agarwal and Ananya Sen, published on SSRN on April 3, 2026, and the first causal evidence of AIO traffic impact. Search Engine Journal surfaced the paper for the SEO industry on April 30.
The setup: researchers randomly assigned users to see standard Google Search (with AIOs) or a version without them, then measured outbound clicks, zero-click rates, and satisfaction scores across both groups. AIOs triggered on about 42% of tested queries. The click and zero-click findings are consistent with prior observational work. The satisfaction finding is not.
Why Randomized Assignment Changes the Argument
Every prior analysis of AIO traffic effects was observational: clicks dropped when AIOs appeared, but observational data cannot rule out simultaneous changes in query mix, SERP layout, or user behavior. Agarwal and Sen's randomized design removes most of those confounders. When the same user sees the same query with and without an AIO and clicks drop, that is not a correlation. The Searchless Journal's summary of the paper frames the distinction plainly: prior studies found association; this one found cause.
The paper is a working paper, not yet peer-reviewed. That is worth naming. But the methodology is still the strongest basis available for making causal claims about AIO traffic impact in internal business cases or client conversations.
Being Cited in an AIO Is a Partial Offset
Seer Interactive's 15-month study of 3,119 queries across 42 organizations found that brands cited inside AIOs achieved 35% higher organic CTR than competitors not cited. The two findings are compatible: AIOs cut total outbound traffic to the web; citation redistributes some of what remains toward cited sources. Practitioners who end up ahead under this dynamic are those who win citation, not those who merely hold rank.
Neither study tells you how to win citation reliably. That is the harder problem.
The 38% click reduction and the absence of any satisfaction gain together suggest AIOs serve Google's session retention goals rather than user information needs. If that reading holds under peer review, publishers will have a substantially stronger argument for regulatory and contractual challenges to the feature. The SEO program implication is worth sitting with: if this traffic reallocation has no user experience justification, the productive response is not to find a better schema combination.
If you are building an internal business case for AIO citation investment this week, the SSRN paper is now your strongest cite for causal impact. The SEJ writeup covers the context for stakeholders who need the story without the methodology section.
Produced with AI assistance.