Google AI Overviews Cut Publisher Clicks as 2026 Core Updates Hit

Two kinds of upheaval in one year

For most of the past two decades, "SEO news" meant tracking Google's ranking updates: which sites rose, which fell, and what the company seemed to reward. In 2026 that kind of news has kept coming, and the updates have been rough. But it now competes with a bigger change. A growing share of search sessions end on Google's own AI-generated answers, so where a page ranks says less than it used to about how much traffic it gets. This year's evidence suggests the second change matters more than the first for publishers, and that the gap between the two is growing.

The algorithm calendar: fast and volatile

Google's Search Status Dashboard lists a busy 2026. A Discover-only update ran from February 5. A short spam update came on March 24, followed by a broad core update on March 27. A second core update started May 21, and spam updates followed in June, August and September.11 The September spam update ran for more than 13 days, much longer than the two- to three-day spam rollouts earlier in the year.11

As usual, Google said little about the core updates. It called the May rollout "a regular update designed to better surface relevant, satisfying content for searchers from all types of sites."13 It told site owners there was "nothing new or special" they needed to do if they were already making people-first content.16 Industry trackers described something far from routine. Semrush's volatility sensor reached 9.5 out of 10 during the March rollout. About 24.1% of pages that had ranked in the top 10 fell out of the top 100 entirely, compared with 14.7% after the December 2025 update.14 Search Engine Roundtable's Barry Schwartz said the May update felt bigger than March's.13 One analysis cited Glenn Gabe's view that May was heavier still.15 The two updates came only 43 days apart, one of the shortest gaps in recent memory.14

Accounts of who gained and who lost line up fairly well. Recaps of the May update describe gains for topical depth, first-hand experience and original testing, and continued losses for aggregators, directories and mass-produced summary content.15 Others list thin informational pages, AI-generated material and aggregators among the hardest hit.20 Google's stated position, as summarized by one agency, is that it targets scaled content abuse, not AI involvement as such.14 One company's case study of its own May losses shows what a core update usually does. Impressions fell about 25% and average position slipped about 1.5 spots, but click-through rate did not change.12 A core update moves a page up or down the list. It does not change how likely people are to click.

Ranking no longer guarantees the click

The AI layer above the results does change that likelihood. The causal evidence got much stronger this year. A randomized field experiment with 1,065 desktop Chrome users, posted to SSRN in April and revised in June, found that showing an AI Overview cut outbound organic clicks by 39.8% and raised zero-click searches by 34.5%. Clicks on sponsored results did not change.23 The effect came almost entirely from Overviews placed at the very top of the page, which the authors say describes more than 87% of appearances.23

A second preregistered experiment, from researchers at the University of Pennsylvania and Northeastern University, was posted to arXiv on August 18. It tested AI Mode directly. When participants' searches were routed into AI Mode, the share of searches ending in a click to an outside site fell by 18.8 percentage points. Removing AI features raised that share by 8.8 points among participants who complied.410 AI Mode also reduced the share of users clicking on news sites, Reddit and Wikipedia. Trust in information on Google fell, as did self-reported satisfaction.104 Trade coverage agrees on an important caveat. The test made AI Mode the effective default, raising its use from 0.6% of searches to almost 95%, so it shows what an AI-first Google would look like rather than how people choose to use it today.103

The observational data points the same way, though the numbers vary, and the variation is worth taking seriously. Ahrefs reported a 58% lower click-through rate for top-ranking pages on keywords that trigger an Overview, up from 34.5% in April 2025.21 Pew found that users clicked a traditional result in 8% of searches with an Overview, compared with 15% without one, and clicked a link inside the summary only 1% of the time.22 One German SEO analysis warns that these figures measure different things and are often mixed up. The Ahrefs number describes click-through on the top-ranked page, not average traffic loss. Seer's long-running data even showed tentative stabilization in early 2026.27 That warning is fair. Still, studies using different methods all find the same direction of effect, and none finds the opposite.

Estimates of how often Overviews appear differ even more. They range from about 13% of queries26 to more than 40% by May, according to Similarweb.45 For branded searches, coverage seems to be climbing quickly. DemandSphere data reportedly showed Overviews on more than 80% of tracked branded queries by late September, up from about 26% at the start of that month.7

What publishers are actually losing

Publisher data shows the damage most clearly. The Reuters Institute, using Chartbeat data from more than 2,500 news sites, found Google organic traffic down 33% globally and 38% in the US between November 2024 and November 2025. Publishers expect a further 43% decline over three years.25 Smaller sites have fared worst. Search referrals fell 60% for small publishers, 47% for mid-sized ones and 22% for large ones.24 Growtika data cited by Playwire shows ten major tech outlets dropping from 112 million monthly US visits from Google to under 50 million in about two years. Digital Trends alone reportedly fell 97%.29

The UK picture adds an important detail. A study of eight leading publishers by the Association of Online Publishers found organic Google referrals falling 7.1% in a single quarter. If that pace holds, the study projects that Google-delivered pageviews will halve by Q3 2027.6 News publishers declined about 5% per quarter, while consumer and B2B publishers lost 18%. How-to content lost more than 20%.6 The pattern makes sense. Evergreen explainers are the easiest content for an AI to summarize, while breaking news reporting is harder to replace.

Publishers are planning around these numbers. One large lifestyle publisher told Digiday it models AI Mode reaching about 80% of Google searches by the end of 2027, and prepares for a further 30–40% drop in Google referrals.9

Google's response and the regulatory squeeze

Google's public account differs from the independent data. At I/O 2026, Sundar Pichai said AI Overviews had more than 2.5 billion monthly users and AI Mode more than a billion, and called it the company's biggest search upgrade ever. Google has not published independently verifiable click-through data for AI Mode.2 In May, the company added inline links, hover previews, "Subscribed" labels and "Further Exploration" suggestions, which it presented as ways to send traffic back to publishers.2127 Google's own earnings give some support to the critics. Search revenue on Google's own pages grew 19%, while the Google Network business, which serves ads on partner sites, fell 4%.27

Regulators have moved faster in Britain than anywhere else. On June 3, the Competition and Markets Authority ordered Google to let publishers opt out of AI Overviews, AI Mode and AI summaries in Discover without losing their organic rankings. Google must also attribute content with clear links, and publishers can opt out of fine-tuning.3334 Google has nine months to comply.35 Industry reaction is divided. The News Media Association called the order a significant step.38 Raptive's Paul Bannister called the toggle "nonsensical" and said it does not stop Google from using publishers' content.31 The controls also apply to all AI features at once, so a publisher cannot leave AI Overviews while staying in AI Mode.35

In US courts, publishers have had less success so far. A tracker reports that Google's motions to dismiss the Chegg and Penske Media cases were granted on September 30, without prejudice.36 The same tracker notes that in mid-September Google began testing payments to publishers for content used in AI surfaces.36

Rival AI engines are no replacement yet

Traffic from AI assistants is growing, but from a small base. ChatGPT's referrals jumped about 157% in a week after OpenAI made brand links more prominent on May 7, and most of that new traffic landed on homepages.4345 Ahrefs data still puts ChatGPT at about 0.21% of site traffic, compared with nearly 40% for Google.48 ChatGPT's share of generative-AI web traffic also fell to about 53% as Gemini and Claude grew.41 Chatbot referrals still account for under 1% of publisher pageviews.45

The takeaway

The core updates show that Google is still rebalancing rankings toward expertise and depth, and sites can respond to that. The AI layer is a different kind of problem. Even pages that rank well get fewer clicks, which is why one analysis notes that only 17–54% of Overview citations now come from top-10 results.15 For publishers, this means ranking and citation now need to be tracked separately. Neither is likely to bring back the traffic levels the old search model delivered.