AI Overviews Reduce Clicks by 34.5%

Summary:

This is an Ahrefs study looking at whether Google's AI Overviews (the AI-written summaries that sometimes appear at the top of search results) are stealing clicks away from websites.

Ahrefs is an SEO software company, so they have a stake in this: their whole business depends on organic search still being worth optimizing for, and they sell tools that track exactly this kind of SERP feature.

The goal was to test Google's own claim that AI Overviews actually get people to click links more, not less.

To check this, Ahrefs pulled 300,000 keywords from their own keyword database: 150,000 that trigger an AI Overview and 150,000 similar informational keywords that don't.

They looked at real click and impression data from Google Search Console for the top-ranking page on each keyword, comparing March 2024 (before AI Overviews rolled out widely in the US) to March 2025 (after).

They found that click-through rates dropped for both groups over that year, which makes sense since search habits shift naturally.

But the keywords with AI Overviews dropped much more sharply.

• For plain informational keywords, average top-spot CTR fell from 5.6% to 3.1%.
• For AI Overview keywords, it fell from 7.3% to 2.6%.

Using the natural decline seen in the non-AI-Overview group, Ahrefs calculated what CTR 'should' have been for AI Overview keywords if AI Overviews never existed: about 4.0%.

The actual number, 2.6%, is about 34.5% lower than that.

In short, AI Overviews seem to be cutting clicks to the top-ranking website by roughly a third, which contradicts Google's public statements that links inside AI Overviews get more clicks than regular listings.

A few things to flag: this is a correlational, before-and-after comparison rather than a true controlled experiment, so other factors changing between March 2024 and March 2025 could muddy the result, even though the design (comparing an AI

Overview group against a similar non-AI-Overview control group) is a reasonable way to isolate the effect.

The methodology and math are shown clearly, which is good, but Ahrefs is not a neutral party here — as an SEO tool vendor, findings that make SEO clicks look more fragile and worth monitoring closely support their own product story.

The article also notes this study was later re-run for 2026, so the numbers here reflect a specific snapshot in time.