Research library

AI summaries cut Google link clicks roughly in half

Strong evidence

Official documentation, sworn testimony, or a large controlled test that others have replicated.

Researcher
Pew Research Center (Athena Chapekis and colleagues)
Published by
Pew Research Center
Date
22 July 2025
Method
Passive browser-tracking panel. Members of KnowledgePanel Digital installed an app that logged their online browsing; every Google search and click during March 2025 was recorded, then AI-summary presence was coded from results collected 7-17 April 2025.
Sample
900 US adults; 68,879 unique Google searches, of which 12,593 produced an AI summary
Original
www.pewresearch.org →

What they did

Pew Research Center recruited 900 US adults from an online probability panel who let an app record their browsing. It logged every Google search and click during March 2025, then coded which result pages carried an AI summary.

What they found

Around 18% of searches produced an AI summary. When one showed, people clicked a normal search result on 8% of visits, against 15% when no summary appeared. Only 1% of visits included a click on a source cited inside the summary itself. Sessions ended on 26% of pages with a summary versus 16% without.

How much weight to give it

This is one of the few studies watching real logged behaviour rather than modelled clickstream estimates. The panel is probability-based and the query count is large, so the click gap is credible.

What it doesn't prove

It covers a single month and US users only. Google disputed the findings. AI Overviews were still young in March 2025, so the numbers will have moved since.

AI search

Spotted an error, or did we grade this unfairly? Tell us and we'll amend the entry: use the contact form.