Explainers built on public data
Grab The Data

AI Overviews and Publisher Traffic: What the Click Studies Show

When Google shows an AI summary, fewer people click through to websites. What the Pew Research Center and Ahrefs measured, how big the effect is, and what Google says in response.

Google’s AI Overviews put a written answer above the search results. For anyone searching, that can save a click. For the websites the answer is built from, it can cost one. Two of the most cited measurements of that cost come from the Pew Research Center, which tracked real people’s browsing, and Ahrefs, which analysed click-through rates across hundreds of thousands of keywords. They use different methods and reach the same conclusion.

How AI engines decide which sites to cite is a separate question, covered in our explainer How AI Search Engines Choose Which Websites to Cite.

What Pew measured

In March 2025, Pew recorded the browsing of about 900 US adults who agreed to share it, covering 68,879 Google searches. It then compared what people did on results pages with an AI summary against pages without one.

People click through less when an AI summary appears

Share of Google visits, March 2025

  • No AI summary
  • With an AI summary
  1. Clicked a traditional search result15%8%
  2. Ended the browsing session there16%26%

Source: Pew Research Center (about 900 US adults, 68,879 searches). Google disputes the study's methodology.

Show as table
BehaviourNo AI summaryWith AI summary
Clicked a traditional search result15%8%
Ended the browsing session16%26%

Two findings stand out. Clicks on ordinary results roughly halved, from 15% of visits to 8%. And people were more likely to stop browsing altogether after seeing a summary. The links inside the AI summaries themselves were clicked in only about 1% of visits.

What Ahrefs measured

Ahrefs took a different approach: it compared the average click-through rate of the top-ranking page for keywords that show an AI Overview, before and after AI Overviews appeared. It has run the analysis twice.

The top result loses more clicks than it used to

Lower average click-through rate for the #1 result when an AI Overview appears

  1. First study (March 2024 vs. March 2025)−34.5%
  2. Update (Dec. 2023 vs. Dec. 2025)−58%

Source: Ahrefs. The update analysed about 300,000 keywords. Bars show the size of the reduction; a correlation, not a controlled experiment.

Show as table
StudyLower CTR for the top result
First study (March 2024 vs. March 2025)34.5%
Update (December 2023 vs. December 2025)58%

The effect grew between the two studies, and it doesn’t stop at the first result. In the update, Ahrefs measured a drop of about 51% for the page in position two, shrinking to about 19% by position ten.

What Google says

Google rejects these conclusions. It called Pew’s methodology flawed and its set of searches unrepresentative, and it says it has not seen significant drops in overall traffic to the web. Google also argues that the clicks AI Overviews do send are “higher quality”, with visitors more likely to stay on a page.

Both sides have a point worth keeping in mind. Pew’s sample is a few hundred people over one month. Ahrefs measures correlation across keywords, not a controlled experiment, and click-through rates also shift for other reasons. But two independent methods pointing the same way, with the effect growing over time, is hard to dismiss.

What it means for publishers

  • Informational queries are hit hardest. Simple factual questions are exactly what AI summaries answer well, so pages built only to answer them lose the most.
  • Being cited isn’t the same as being visited. Citation links inside summaries receive very few clicks.
  • Depth still earns the click. Original data, analysis, tools and first-hand experience give people a reason to visit that a summary can’t replace.

The searches people still click through from are the ones where a summary isn’t enough. That’s where the opportunity for publishers now sits.

Sources