AI Search Is Quietly Rewriting Where Web Traffic Goes

If you've noticed a website you used to visit regularly feels quieter than it did a couple of years ago, you're not imagining it. A growing share of searches now end without a single click to any website at all, because an AI-generated summary at the top of the results page already answered the question. That shift has been building for a while, but 2026 is the year the numbers behind it got hard to ignore — and they're worth understanding whether you run a website, write online, or just want to know why the internet feels different than it used to.

The numbers behind the shift

Zero-click searches — where someone searches and gets their answer without clicking through to any site — now account for roughly 60% of Google queries, up sharply from a couple of years ago, per The Next Web. Google's own traffic to publishers reportedly dropped 33% globally in the year ending November 2025, and individual sites report steeper declines: HubSpot around 70-80%, Chegg 49%, some DMG Media queries down as much as 89%.

The click-through data tells a similar story. A Pew Research study from July 2025 found only 1% of users click a link cited inside an AI Overview, and organic results below one get clicked far less — 8% of the time versus 15% with no AI Overview present. An Ahrefs analysis of roughly 300,000 searches found organic click-through rates drop about 34.5% once an AI Overview appears. For chatbots specifically, a TollBit report found ChatGPT and Perplexity send 95-96% less referral traffic per interaction than a traditional Google search.

These figures come from different studies measuring different things — one about AI Overviews, another about chatbots, another a single publisher's experience — but they all point the same direction, consistently enough that NPR has described the effect on some publishers as an "extinction-level event."

It's not entirely one-sided

Google disputes the framing, arguing that AI Overviews "generate more clicks, not fewer" — pointing to categories of query where the AI answer creates interest it wouldn't have otherwise. There's a real, narrower point buried in that pushback: the traffic that does arrive from AI tools looks different, and in some ways better, than typical search traffic.

Adobe Analytics data cited in coverage of this trend found visitors referred by AI tools stay on a site about 8% longer, view roughly 12% more pages, and are about 23% less likely to bounce immediately, versus visitors from traditional search. Conversion rates from AI referral traffic still run lower overall, but the gap is narrowing, particularly for B2B sites — less of this traffic, but a larger share of people who actually wanted to be there.

That's small consolation for anyone whose overall numbers are down by half, but it changes how you should think about the shift: not "the internet has less interest in what you publish," but "casual, single-fact lookups no longer need to land on a website, and what's left skews toward people wanting something a one-paragraph summary can't give them."

What actually changes if you publish anything online

The practical response taking shape across the industry has a name — Answer Engine Optimization, or AEO — and it's less a replacement for SEO than an addition to it. A few things show up consistently in how sites are adapting:

Content that's easy for an AI system to extract cleanly — clear headings, direct answers stated early, well-structured lists — tends to get pulled into AI summaries more often than content where the answer is buried in the fifth paragraph. That's a genuine incentive to write more plainly.

The content hardest to summarize away also tends to survive this shift best: original reporting, direct personal experience, expert interviews, and analysis that isn't just a restatement of facts already in an AI model's training data. A page that only exists to answer "what is X" generically is exactly what a zero-click summary was built to replace.

And increasingly, sites are tracking AI referral traffic as its own category rather than lumping it in with search, since it behaves differently enough — in engagement, in conversion, in which pages it favors — to be worth measuring separately.

The honest takeaway

The open web isn't disappearing, but the economics of putting something on it have changed. A page that exists purely to capture a search query and monetize the click now competes directly with a machine that can answer that query without sending anyone anywhere. What holds up better is the stuff a summary genuinely can't replace — and figuring out which category your own work falls into is more useful than watching the traffic dashboard and hoping it recovers on its own.