What AI Overviews actually did to clicks

What is the most reliable number on AI Overview clicks?

Pew Research Center ran the cleanest study available. In March 2025 it tracked the actual browsing of 900 US adults on its KnowledgePanel, covering 68,879 searches, of which 12,593 returned an AI summary. The finding: users clicked a traditional search result on 8% of visits where an AI summary appeared, against 15% where it did not.

Roughly half the click rate. Only 1% clicked a link inside the summary itself.

Pew is independent, has no product to sell in this market, and published its method. That is the number to plan with.

Where does the 58% figure come from?

The 58% number has a single source and a narrow scope. The 58% figure comes from Ahrefs, published 4 February 2026, aggregating Google Search Console data across 300,000 keywords, half with AI Overviews and half informational keywords without, comparing December 2023 with December 2025.

What it measured is narrower than the headline: a roughly 58% reduction in click-through rate for the top organic position, on keywords where an AI Overview appears. Position-one click-through fell from 0.073 to 0.016 against a forecast 0.037.

It does not mean sites lost 58% of their traffic. It supersedes Ahrefs’ own earlier 34.5% figure from April 2025, and Ahrefs sells SEO software, which is worth stating when quoting it.

How do the Pew and Ahrefs numbers compare?

Two studies sit behind the numbers people quote, and they did not measure the same thing. Pew Research Center measured whether people clicked a traditional search result at all. Ahrefs measured click-through rate for the top organic position, on keywords where an AI Overview appears. Sample, date and finding for each are below.

What we read from that, and this part is our judgement rather than anything the sources state: the two numbers are not interchangeable, because a click-through rate on one position across a keyword sample is a narrower thing than a site’s traffic, and of the two publishers only one has no product to sell in this market.

Sources: Pew Research Center, March 2025; Ahrefs, published 4 February 2026.

What does this mean for a home-service business?

Less than it means for a publisher, and the reason is intent. The queries that lose clicks to a summary are the informational ones. “How does a heat pump work” can be answered in the summary. “Who installs heat pumps near me” cannot, because the answer is a business and the user has to reach it.

So the exposure is concentrated in top-of-funnel content, and the commercial queries that actually produce calls are the least affected.

The real shift is that appearing inside the answer starts to matter as much as appearing under it. That is a different measurement problem, which is why we sample it separately rather than folding it into a ranking report.

What we would not tell you

That we can restore the clicks. Nobody can, and the mechanism sits with Google. That any single percentage describes your site. These figures are aggregates across hundreds of thousands of keywords, and your own Search Console data for your own commercial terms is the only number that describes you.

“Appearing inside the answer starts to matter as much as appearing under it.”