By Stuart Kerr, Technology Correspondent, LiveAIWire
Google’s own search referrals to publishers fell by a third globally in the year to November 2025, and inside AI Mode specifically, roughly 93 percent of searches now end without a single click to an outside website. That is not a worst-case estimate from an angry publisher. It comes from Chartbeat data covering more than 2,500 news sites, published through the Reuters Institute’s own 2026 industry survey, alongside a separate Semrush analysis of 69 million US desktop search sessions measuring AI Mode specifically.
Since AI Mode moved from limited testing into a mainstream Google Search feature, the gap between it and the older AI Overviews format has become one of the more important distinctions in the SEO world, and one that gets blurred constantly. AI Overviews still show a summary above a list of familiar blue links. AI Mode replaces the results page with a conversational interface that answers follow-up questions on its own, and Google has been actively steering more traffic toward it throughout 2026.
Table of Contents
AI Overviews and AI Mode Are Not the Same Thing
The distinction matters because the two features behave very differently for the sites whose content they draw on. When an AI Overview appears, a user can still see and click a traditional result underneath it, and citation data suggests roughly 43 percent of those searches end with no click at all. Inside AI Mode, that figure climbs to around 93 percent, because the format itself is built to keep the user inside the conversation rather than send them elsewhere. Google has openly described AI Mode as suited to longer, more complex queries, and its own executives have said users type two to three times as many words into it as they do into ordinary search.
What This Means for You
If your business depends on organic search traffic, the practical shift is this: ranking well is no longer the same thing as being visible, and being visible is no longer the same thing as getting a click. A Seer Interactive study covering 42 organisations and more than 25 million impressions found that brands actually cited by name inside an AI-generated answer saw 35 percent more organic clicks and 91 percent more paid clicks than brands that were not cited at all. The practical goal has shifted from ranking first to being the source Google’s AI decides to name, which changes how content should be written and structured, not just where it should rank.
The Numbers Publishers Are Actually Seeing
The scale varies enormously by publisher and content type. Educational platform Chegg told regulators its non-subscriber traffic fell 49 percent in a single year, a decline central to the antitrust lawsuit it filed against Google. HubSpot has separately estimated organic traffic losses of 70 to 80 percent, and CNN has reported a decline in the 27 to 38 percent range. Reuters Institute’s 2026 survey of 280 media leaders found that publishers now expect search referrals to fall by a further 43 percent on average over the next three years, with roughly one in five expecting losses beyond 75 percent.
An earlier UK-focused study by SEO platform Authoritas, submitted as part of a formal complaint to Britain’s Competition and Markets Authority, found a 47.5 percent clickthrough drop on desktop and 37.7 percent on mobile when AI Overviews appeared, and warned the picture would get worse once AI Mode rolled out more broadly. Google disputed the study’s methodology at the time, telling reporters it continues to send billions of clicks to websites daily and has not observed the dramatic traffic collapse being described.
Google’s Defence, and Why This Is Now a Legal Fight
Google’s consistent public position is that AI features create new ways for content to be discovered rather than simply removing existing ones, and that publishers focused only on raw click volume are measuring the wrong thing. Regulators and courts are now being asked to settle the disagreement directly. The UK’s Competition and Markets Authority has said it plans to give publishers more control over how their content appears in AI-powered search, and in the US, Google has been fighting Penske Media’s antitrust claims over AI Overviews in court, arguing in a January 2026 filing that the feature is a lawful product improvement rather than anti-competitive conduct.
The core problem publisher groups keep raising is that there is currently no clean way to opt out. A publisher can block their content from training future AI models, but Google has stated plainly that this does not prevent the same content from being used to generate answers inside AI Overviews or AI Mode, since it treats that use as grounding rather than training. Opting out of AI features entirely generally means opting out of Google Search visibility altogether, which for most publishers is not a real choice at all.
Not Every Publisher Is Losing
The picture is not uniformly bleak. Digiday’s reporting on AI Mode’s winners and losers found that publishers who had already diversified away from Google referrals, building direct subscriber relationships through newsletters, apps and memberships, were far less rattled by the shift than those still dependent on search traffic alone. Branded searches have also behaved differently to generic informational ones, with several 2026 studies finding clickthrough rates on branded queries rising even as informational query traffic collapsed, since users searching for a specific company by name tend to click through regardless of what summary Google shows first.
News/Media Alliance president Danielle Coffey has described how thoroughly the underlying economics have shifted for her members. Where roughly half to four-fifths of member traffic once came from Google, that figure has fallen to somewhere between a fifth and a third, according to the organisation’s own internal analysis. That kind of structural shift explains why LiveAIWire’s earlier look at AI Mode’s Canvas and Search Live features flagged publisher anxiety even before the traffic data caught up with it.
What Publishers Are Actually Doing About It
The adaptation playbook that keeps recurring across 2026 coverage is consistent: build content that answers a question so completely it becomes the obvious source to cite by name, invest in original reporting and analysis that a summarising model cannot easily reproduce, and reduce dependence on search referrals through direct audience relationships. This is not simply a defensive posture. As LiveAIWire’s reporting on algorithmic framing and emotional manipulation has explored elsewhere, the systems making these decisions about what to summarise and what to cite are shaping perception in ways that go beyond simple ranking, and understanding that framing is becoming as important to a publisher’s strategy as understanding keywords once was.
None of this is happening in isolation from the rest of Google’s AI strategy. The company’s aggressive push into AI Mode sits alongside the kind of infrastructure spending LiveAIWire has tracked in its coverage of the broader AI arms race among Google, Amazon and Meta, and the competitive pressure from rival answer engines detailed in LiveAIWire’s comparison of ChatGPT, Gemini and Claude is part of why Google has moved so quickly to make its own AI-native search experience the default rather than an optional feature. For publishers, the strategic question for the rest of 2026 is no longer whether AI Mode will keep growing. It is whether their content is built to be the thing it chooses to cite.
About the Author
Stuart Kerr is Technology Correspondent at LiveAIWire, covering artificial intelligence, emerging technology, and their impact on business, society, and everyday life. LiveAIWire publishes original AI journalism every weekday at liveaiwire.com.