Google AI Overviews are becoming the default search experience
Google’s AI-generated summaries are no longer a novelty. They now appear in 43% of US searches, according to a new report from market intelligence firm Similarweb. A year ago, that figure was just 15%. The jump signals a fundamental shift in how the world’s most popular search engine presents information.
The data comes from Similarweb’s 2026 Generative AI Landscape report, which tracks the rise of AI across search and standalone platforms. It’s a staggering number, but it’s only part of the story.
What exactly are AI Overviews and AI Mode?
AI Overviews are the generated summaries that appear directly within conventional Google results. They sit above the traditional blue links, offering a direct answer to a query. Then there’s AI Mode, a more conversational interface designed for longer questions and follow-up prompts. Users can move seamlessly from an AI Overview into AI Mode, carrying the context of their original search with them.
Google says AI Mode relies on a technique called query fan-out. The system breaks a question into related subtopics and runs multiple searches in parallel. It then synthesizes the results into a single, coherent response with supporting links. It’s a clever approach, but it raises a big question: what happens to the websites that used to get that traffic?
The numbers behind the AI search boom
Similarweb’s estimates show that visits to Google’s AI Mode web experience nearly doubled, jumping from 126 million in June 2025 to 279 million in May 2026. Google itself reported in May that AI Mode had surpassed one billion monthly users. That figure counts users, not visits, so it’s not directly comparable. Still, the trajectory is unmistakable.
The 43% figure and the AI Mode numbers measure different things. The former tracks the share of US searches displaying AI Overviews. The latter reflects activity within Google’s conversational service. Both point to the same conclusion: AI is now central to the search experience.
There’s another telling metric. Similarweb recorded a 5.4% increase in the average length of Google searches after AI Mode launched. People are typing longer, more natural-language queries. They’re searching the way they’d talk to an assistant, not the way they’d keyword-stuff a query in 2015.
“People are now adopting a new, more natural way to search and discover,” said Ethan Smith, chief executive of digital marketing firm Graphite, in the report.
How Similarweb gets its numbers
It’s worth being clear about the methodology. Similarweb’s figures are estimates, not direct counts from Google or OpenAI. They draw on first-party analytics, anonymized device information, external data partnerships, and publicly available web data. The company models those inputs to estimate traffic and user behavior. It’s solid data, but it’s not the whole picture.
The broader generative AI market is booming too. Generative AI websites received an average of 9.5 billion monthly visits between June 2025 and May 2026, up 70% year-over-year. Monthly unique visitors rose 57% to 655 million, and mobile app downloads climbed 58% to 4.4 billion. Those figures cover standalone services like ChatGPT, Gemini, Claude, and Perplexity.
Publishers worry about disappearing referral traffic
Here’s the tension. Publishers have long raised concerns about referral traffic when their content gets absorbed into an AI-generated answer without producing a click. If a user gets their answer directly in the search results, why would they visit the source?
Google has responded by tweaking how external sources appear. In May, the company announced direct links within responses, article suggestions, website previews, and more prominent references to original material. The stated goal: give users more context about linked pages and make relevant websites easier to spot. Google didn’t release traffic data showing whether these changes actually helped publishers.
Similarweb has previously identified news publishers among the sectors hit hardest by traffic changes from AI-based search tools. The concern is real, and the data backs it up.
Citations don’t always mean clicks
Here’s a fascinating wrinkle. Similarweb found that the share of ChatGPT responses containing web citations increased more than fivefold over the past year. It rose from about 1.3% in June 2025 to 6.8% in May 2026, based on US desktop activity. But a citation is not a click. The measurement covers responses that display source references, not whether users actually opened them.
Citation rates also vary wildly by sector. Travel, retail, and sports responses show higher rates, likely because prices, availability, and results change so frequently. AI needs fresh data for those queries, and it needs to show its sources.
There’s a deeper mismatch at play. About 65% of URLs cited by ChatGPT were located two or three folders below the main domain. That means articles, product pages, and other detailed content. Folder-depth two alone accounted for 41.7% of cited URLs. Yet 58.8% of AI referral traffic landed on homepages.
In plain English: the pages that inform an AI answer are often not the pages users visit. Detailed content feeds the response, but the link people click takes them to a company’s main site. That’s a crucial distinction for anyone tracking the value of their content.
ChatGPT’s search update changed the game
ChatGPT’s May 7 search update gave brand links greater prominence within generated responses. The effect was immediate. Similarweb recorded a 157.7% week-over-week increase in referral traffic following the update. Homepage referrals jumped 354.7%. Before the update, about 26% to 32% of ChatGPT referral visits landed on homepages. After May 7, that proportion rose to roughly 60%, based on desktop activity measured between April 30 and May 20, 2026.
The firm observed the increases after the interface change but didn’t establish causation. Still, the correlation is hard to ignore. When AI platforms make brand links more visible, users click them more often.
Cloudflare’s answer: pay per crawl
As AI services scrape the web at scale, publishers are looking for new ways to control access. Cloudflare is testing a Pay per Crawl service that lets website operators permit, charge, or block individual AI crawlers. The private-beta program allows publishers to set a fee that an authenticated crawler must pay before accessing content.
The system can return an HTTP 402 “Payment Required” response when paid access applies. Cloudflare records completed paid requests, charges the crawler operator, and distributes the proceeds to the website owner. It’s a bold experiment in turning content access into a commercial transaction.
Cloudflare has also made configurable HTTP 402 responses available to paying customers through its AI Crawl Control service. Website operators can use the response to provide licensing terms or contact details to crawlers. Automated payments through Pay per Crawl remain in beta, but the infrastructure is taking shape.
The measures give publishers another lever to pull as AI-generated answers become more common across search and standalone AI platforms. Whether they’ll be enough to offset lost referral traffic remains an open question.
For now, the numbers tell a clear story. AI is not a side feature of search anymore. It’s the main event. And everyone from publishers to marketers to everyday users is still figuring out what that means.
