TLDR: Google's AI Overviews deliver instant answers at the top of your search results—but this convenience quietly starves the open web. When these AI summaries appear, only 8% of users click through to source websites, compared to 15% without them (Pew Research Center, 2025). Publishers report traffic drops as high as 47% on affected queries (Authoritas via Press Gazette, 2025), threatening the independent creators, newsrooms, and niche experts who produce the knowledge these AI systems depend on.
You search. Google answers. You move on with your day.
It's a deceptively elegant exchange, made possible by AI Overviews—Google's Gemini-powered summaries that expanded across search results in 2025. The feature appeared in roughly one-in-five searches during a Pew Research Center study period, synthesizing information from multiple sources into neat, conversational paragraphs. Faster answers sound like pure upside, right?
But here's the thing nobody mentions in the feature announcement: That tidy summary is quietly hollowing out the internet it was trained to search.
The Vanishing Click
The numbers tell an uncomfortable story. When an AI summary appears, search sessions end on Google's page 26% of the time, versus just 16% for traditional results (Pew Research Center, 2025). Even when people do scroll past the summary, they're far less likely to click anything—8% versus 15% for searches without AI Overviews. And those citation links within the AI summary, the ones supposedly sending readers to sources? About 1% of users tap them (Pew Research Center, 2025).
This behavior fuels what's now called "zero-click" searches, which accounted for nearly 69% of all Google queries by mid-2025 (Similarweb via TechCrunch, 2025)—up from 56% just a year earlier. For two decades, the internet operated on a simple bargain: creators produce valuable content, search engines send them traffic. AI Overviews rewrote that deal overnight.
Publishers are experiencing something bizarre: visibility without visitors. Search impressions are climbing—your site appears in more results than ever—but actual clicks plummet (BrightEdge, 2025). It's like having a storefront on the busiest street in town, except everyone stops at the window display and never walks inside.
The data is merciless. When AI Overviews appear, the average click-through rate for top-ranking pages drops 34.5% (Ahrefs, 2025). For news queries specifically, desktop click-through rates crater by roughly 47.5% (Authoritas via Press Gazette, 2025). Translation: You can rank #1 and still lose nearly half your audience.
The Humans Behind the Stats
Chegg, the online education platform, watched non-subscriber traffic plunge 49% year-over-year and sued Google, alleging AI Overviews siphon demand directly from its content (Search Engine Land, 2025). Penske Media—parent company of Rolling Stone, Variety, and The Hollywood Reporter—reported affiliate revenue dropping by more than a third, a decline it attributes squarely to reduced Google referrals.
The trade group Digital Content Next, representing publishers from Condé Nast to Hearst, surveyed its members in May and June 2025. The results? Traffic losses from Google search hit 10-25% for most, with some weeks seeing news brands down 16% and non-news publishers down 17% (Digital Content Next, 2025). These aren't blips. They're structural.
Independent creators face even grimmer math. Consider travel bloggers who spent years building expertise on niche destinations—Canadian slang, hidden hiking trails, budget accommodation tricks. Some report traffic drops approaching 90%, forcing site closures (NPR, 2025). When your revenue comes from ad impressions, affiliate commissions, and converting casual visitors into newsletter subscribers, a 90% traffic loss isn't a setback. It's a death sentence.
DMG Media, which operates MailOnline and Metro, documented CTR drops as high as 89% on specific queries where AI Overviews dominated the page (Search Engine Journal, 2025). That's not a typo. Eighty-nine percent.
Why This Keeps Happening
Google maintains its AI features send "higher-quality clicks" to a "wider variety of websites"—but offers zero public data on AI Overview click-through rates. Meanwhile, the company's incentives point clearly toward keeping users inside the garden walls. Google invested approximately $75 billion in AI infrastructure in 2025 (Google/Sngular, 2025), a bet on AI-first experiences that answer questions on platform rather than sending people elsewhere.
The financial picture reinforces this. Network advertising revenue—the money Google shares with external publishers through AdSense and Ad Manager—declined 1% in Q2 2025, even as Search and YouTube advertising (revenue Google keeps entirely) surged (PPC Land, 2025). As organic traffic erodes, some advertisers now pay more in Google Ads just to reclaim lost visibility (AdExpert, 2025).
There's an informational cost, too. As AI Overviews increasingly favor a narrow set of sources—Reddit posts, YouTube videos, Wikipedia entries—the risk of quality degradation grows. Jason Kint of Digital Content Next frames it starkly: Google is "trading the public square for a walled garden built on monopoly profits."
Publishers face a brutal bind. Block Google's AI crawlers and you disappear from search entirely. Allow crawling and you're training the system that's taking your traffic. It's not a choice. It's extortion with a friendly interface.
The Measurement Black Box
Google Search Console, the primary analytics tool publishers use to understand their Google traffic, doesn't separate AI Overview clicks from traditional organic clicks (DCN, 2025). This opacity makes it nearly impossible for creators to know exactly how much AI Overviews hurt them versus seasonal trends, algorithm updates, or other variables.
Industry impact also varies. News queries trigger fewer AI Overviews than "how-to" searches; Google Discover's referral patterns add noise; some verticals (healthcare, education) see AI Overviews in 87% of relevant searches while e-commerce sees just 4% (BrightEdge, 2025).
But here's the thing: Multiple independent 2025 studies—from Pew Research Center to Ahrefs to Authoritas—all converge on the same finding. When AI summaries appear, fewer people click through to the sources that made those summaries possible in the first place.
What Happens Next
The fight for the open web is already underway. In Europe, independent publishers filed an antitrust complaint seeking fair opt-out provisions that don't penalize them in search results (Reuters, 2025). In the U.S., Penske Media and Chegg are challenging Google's right to repurpose content that diverts traffic away from the creators. Publishers want transparency on AI Overview click-through rates, fair licensing deals, and a real choice about how their work gets used.
Here's the uncomfortable question nobody in Mountain View wants to answer: What happens when the sources dry up? If creating quality content becomes economically unsustainable, what will tomorrow's AI summarize—yesterday's summaries of last week's summaries?
We're not talking about halting progress. We're asking whether "faster" and "better" mean the same thing. Speed that systematically defunds the production of reliable information isn't optimization. It's extraction masquerading as innovation.
The next time Google hands you the perfect answer in that neat little box at the top of the page, take a second. Scroll past it. Click the source. It's not charity—it's enlightened self-interest. Because the humans who research, verify, test, and create the content that powers these summaries? They need more than algorithmic credit. They need the click.
If the answer helped, tip the source—with your attention.