Google’s AI Overviews Are Killing Clicks Says New Pew Study 

Earlier this year, Google rolled out AI Overviews, a feature that shows AI-generated summaries at the top of many search results in the U.S.

It was meant to save us time. But now, new research shows it might be hurting the websites that give us those answers in the first place.

According to a new study by the Pew Research Center, when people see an AI Overview, they are far less likely to click on the links below.

Toggle

Free SEO Audit: Uncover Hidden SEO Opportunities Before Your Competitors Do

Gain early access to a tailored SEO audit that reveals untapped SEO opportunities and gaps in your website.

Analyze My Site

In fact, only 8% of visits with AI Overviews led to any clicks at all. When no AI summary was shown, 15% of visits led to clicks.

That is almost double. Even more surprising is that only 1% of people clicked on the links inside the AI summary itself.

It makes me wonder what is the use of ranking high on Google if no one’s clicking anymore? Is Google helping users more… or slowly cutting out the websites that power the web?

How Have Click-Through Rates Changed Since the Rollout of AI Summaries?

In 2023, Google began testing a feature now known as AI Overviews (or what users are experiencing as Google AI Mode).

This AI-powered summary appears right at the top of search result pages and provides a condensed answer, citing sources but eliminating the need to scroll down.

By March 2025, this feature was rolled out widely across the U.S., reaching millions of users. And the implications?

Well, they are not just UX tweaks but they are rewiring user behavior.

The Pew Research Center’s latest report examined the online activity of 900 U.S. adults who opted to share their browsing data and the findings were eye-opening.

Users were nearly half as likely to click on links when an AI summary appeared.

Let that sink in.

• When users saw an AI summary, only 8% clicked on a traditional search result.
• When there was no summary, that number jumped to 15%.
• Even more shocking is that only 1% of users clicked on a link within the AI summary itself.

So what exactly is happening here?

Why Are Fewer People Clicking on Google Search Results Now?

Honestly, I get it. I use Google every day and when I see a neat, polished AI summary that answers my question right away, I often do not scroll down either.

But then I think… What about the creators of the original content? What happens to publishers, small businesses, educational bloggers and niche SaaS websites who rely on organic search traffic to survive?

This is no longer a theoretical concern but it is happening.

Publishers have already begun sounding the alarm. Several news organizations like MailOnline, Wired and others have seen measurable declines in traffic since AI Overviews began dominating above-the-fold real estate.

Here is an example:

“When we rank #1, we used to see CTRs of 13% (desktop) and 20% (mobile). But with an AI summary above, that drops to 5% or even less.” – MailOnline SEO Director

So here is the question: Are we training users not to click anymore?

Is Google’s AI Summary Box Causing Users to Stop Searching?

What makes this data even more interesting is what users do after they see the AI box.

According to the Pew study:

• On pages with an AI summary, 26% of users ended their search session entirely.
• On traditional results pages, that number was only 16%.

That means over a quarter of users are stopping their browsing journey right after the AI box. They are not clicking links. They are not comparing sources but they are trusting the AI and bouncing.

And as a marketer or publisher, that is terrifying.

It is not just that AI is winning the click. It is that day by day it is becoming the final destination for users.

What’s Actually Inside These AI Overviews?

Let’s understand deep from the SEO perspective

The usual AI summary is about 67 words long, though they can vary from 7 to 369 words. They are mostly generated in response to: full-sentence queries, questions starting with “what”, “how”, “why”, etc and long-tail queries with 10+ words

In fact, just:

• 8% of short (1–2 word) searches triggered AI summaries
• While 53% of 10+ word queries did

So the more natural, conversational, or specific your query, the more likely you will see an AI answer.

That changes how SEO professionals and content creators have to approach content optimization.

You are not just writing for users or algorithms now but you are writing for AI extractive summaries.

Who Gets Cited Most Often in AI Overviews—And Who Gets Left Out?

Another interesting layer in the study was about source attribution. You do hope that diverse voices get represented in these AI summaries but…

The most cited sources were:

• Wikipedia
• YouTube
• Reddit

Combined, these made up:

• 15% of citations in AI Overviews

• 17% of citations in standard search results

It is not just that your traffic is being siphoned but also that Google’s AI seems to trust a limited set of familiar names which make it harder for smaller or niche content creators to break through.

Want to hear something even more surprising?

• .gov websites were cited in 6% of AI summaries vs. just 2% in traditional results.
• Only 5% of AI summaries cited news websites and the same as regular results.

From Search Engine to Answer Engine: What This Means for You

Google’s AI Overviews are impacting website traffic, with the Pew study revealing a significant drop in clicks for pages where summaries appear.

Publishers and content creators, who rely on organic search for revenue and visibility, are facing a challenging new landscape.

Moving forward, content creators must adapt SEO strategies which involve optimizing for AI by providing clear, concise answers to long-tail and conversational queries, enhancing E-E-A-T signals, utilizing structured data and diversifying traffic sources beyond Google Search.

Google Responds: Is the Pew Research Methodology Flawed?

Following the viral impact of the Pew study, Google has pushed back. According to recent updates, multiple concerns have emerged around the validity of the Pew Research findings. 

Google’s official statement challenges the report on several grounds:

• AI usage is growing and people are asking more questions.
• AI features in Search are increasing visibility for content creators—not reducing it.
• The study used a flawed methodology and a non-representative queryset.

“People are gravitating to AI-powered experiences, and AI features in Search enable people to ask even more questions, creating new opportunities for people to connect with websites.” – Google Spokesperson

Criticism of Pew’s Sample Size and Reliability

Experts like Duane Forrester (formerly of Bing) also critiqued the sample size of just 66,000 queries across 900 users out of the nearly 500 billion queries per month on Google. He noted: 

“Out of 500 billion queries, 66,000 is just 0.0000134%—that is not representative in any meaningful way.”

Additionally, Pew’s statistical reliability was ranked low across various age groups, with margins of error as high as ±13.7% for users aged 18–29 and ±10.2% for those 65+. 

Comparison Timing and Dynamic Nature of AI Overviews

Google has also raised one more issue that Pew compared March’s user search data to April’s researcher-replicated queries. 

That is problematic in many ways because AI summaries are dynamic and changing frequently based on updates and personalization. 

Search trends can influence what queries trigger AI Overviews from week to week. 

The links and summaries displayed may differ from browser to browser even for the same query. 

Google concludes that this makes it invalid to compare user sessions in March to scraped results in April, as the AI output itself may have evolved during that time.

A Case of Be Careful What You Wish For?

Publishers often ask for more diverse link representation and now, Google says that AI Overviews are exactly that: dynamic and varied.

• Links shown are no longer static like traditional SERPs.
• The top three links may change for the same query depending on timing, browsers and personalization.

So could inconsistent traffic be due to AI-driven diversification, not AI stealing the click?

Key Takeaways 

Whether the Pew study holds up under scrutiny or not, the conversation it has sparked is critical.

1. The way users interact with Google is changing.
2. AI Overviews may reduce clicks, but they may also encourage more complex queries.
3. SEO professionals must adapt, not only by ranking but by earning AI citations.
4. And above all, marketers should diversify and don’t rely on Google alone to bring in your audience.