AI Overviews: From Experiment to Default Search Experience
Google AI Overviews are AI-generated summaries that appear at the top of search results, synthesizing information from multiple sources into a direct answer, and they now show in over four out of ten queries, turning Google Search from a gateway to external websites into a destination where users stay to consume information inside Google’s own interface.
The most important change is not the technology but the balance of power. When AI Overviews appeared in 15% of searches, they were an experiment; at 43%, they are the default search experience. That shift moves attention, trust and traffic away from publisher sites and toward Google’s own AI search results. According to Similarweb, “Google’s AI Overviews appeared in 43% of searches by May 2026, up sharply from 15% a year earlier”. In practice, the classic page of ten blue links is now secondary. Users are trained to look first at the AI box, skim the summary, and only click links if something is missing or they need depth.
How User Behavior Is Changing: Longer Queries, Fewer Clicks
As AI Overviews expand, user behavior is changing in ways that favor Google and sideline publishers. Similarweb’s data shows people are moving from short, keyword-based queries to longer, more conversational prompts aimed at triggering richer AI search results. Instead of typing “best electric cars,” users ask multi-sentence questions or describe scenarios, mimicking how they interact with conversational tools like ChatGPT and other assistants.
At the same time, Google is routing more people into AI Mode, a chat-style interface layered on top of search. Visits to AI Mode jumped from 126 million in June 2025 to 279 million by May 2026, a clear sign that search sessions are stretching out inside Google’s ecosystem rather than on publisher sites. Rather than presenting lists of hyperlinks, Google provides synthesized answers and invites follow-up questions without requiring users to leave the page. This design is deliberate: keep users inside Google, weaken the habit of clicking through, and turn search into a sticky conversation instead of a quick pass-through to the open web.
Web Traffic Changes: AI as the New Middleman of the Internet
The rise of Google AI Overviews is not a neutral usability upgrade; it is a structural change in how traffic flows across the web. For more than two decades, search engines acted as matchmakers between users and publishers. Now, Google itself compiles information and answers questions directly, reducing the need for users to click through to original sources. News organizations, educational platforms and other content sites report declining referral traffic because AI-generated summaries satisfy user intent inside the results page.
Publishers are right to be worried. Similarweb identified shrinking referral traffic as one of the biggest impacts of AI-powered search, especially for news outlets that rely on page views for advertising and subscriptions. The problem is not limited to Google. AI assistants that summarize content across the web often include citations, but they do not always drive clicks. As of May 2026, only 6.8% of desktop ChatGPT queries contained citations that directed users to external sites. When AI platforms keep users in their own interfaces, they capture the value of the publisher’s work while returning only a fraction of the audience.
Pushback and New Guardrails: Blocking Crawlers, Adding Links
Faced with these web traffic changes, some infrastructure players are starting to push back on behalf of publishers. Cloudflare now offers website owners tools to block AI crawlers unless AI companies agree to a licensing deal, an early attempt to build a marketplace where access to content is conditional on compensation. This is a clear signal that the industry no longer accepts unrestricted scraping as the cost of visibility.
AI platforms are also slowly improving how they handle citations. Similarweb reports that the share of AI-generated answers that include source references has risen more than fivefold over the past year. And citation quality matters: after a May 7 search update, ChatGPT made links more prominent in its responses, and the share of desktop searches that led to visits to external web pages jumped from about 25% in March to nearly 60% by May 30. These changes show that design choices can restore some traffic to publishers. But they also expose a hard truth: in an AI-first search world, links are a feature that platforms can dial up or down, and publishers have limited influence over that dial.
Rethinking SEO: From Keywords to AI Summary Optimization
The expansion of Google AI Overviews means that traditional search engine optimization focused on ranking blue links is no longer enough. As queries become longer and more conversational, and as natural language interaction replaces keyword string searches, content needs to be written in a way that AI systems can understand, summarize and trust. The goal is no longer only “ranking high”; it is being accurately represented in AI search results, where many users will now stop their journey.
For publishers, this transition is one of the most significant structural changes since search engines first appeared. AI-generated answers becoming the primary interface for information forces a strategic shift: build content that is clear, factual and context-rich so it can be cited and summarized, while developing business models that do not depend entirely on referral traffic from traditional search. The uncomfortable reality is that AI Overviews tilt the field toward platforms that own the discovery layer. The practical response is not to wait for protection, but to assume AI will mediate most user journeys and design content, products and revenue streams for that future.
