Pew Research Center pulled the browsing records of 900 U.S. adults and looked at 68,879 real Google searches in March 2025. In 18% of those searches, an AI Overview sat at the top of the page. That single number reset how search traffic works, and most of the follow-on data confirms the shift is permanent. Here are the AI Overviews statistics worth memorizing before you plan a single piece of content for the rest of 2026.
I have kept every figure attached to its source and its year, because this topic attracts more recycled, un-sourced numbers than almost any other in marketing. Where studies disagree, I show you the spread instead of picking the number that sounds best.
How often AI Overviews actually appear
The honest answer is a range, not a point. Pew Research Center measured 18% of searches returning an AI Overview across its real-user panel. Semrush, working from a keyword panel of more than ten million terms, watched the trigger rate climb from 6.49% in January 2025 to a July peak of 24.61%, then settle back to 15.69% by November. Ahrefs, sampling its own keyword set, put the figure above 20% heading into 2026.
Those numbers are not in conflict. They measure different things. Pew watches what real people search, which skews toward common, navigational queries that often do not trigger an overview. Rank trackers like Semrush and Ahrefs sample broader keyword sets that lean informational, so they catch more overviews. Read the AI Overviews statistics as a band: somewhere between one in six and one in five Google searches now opens with an AI answer, and the ceiling during peak months has touched one in four.
Which searches trigger an AI Overview?
Query length is the strongest predictor Pew found. Only 8% of one-to-two-word searches produced an AI Overview, but 53% of searches ten words or longer did. Phrasing matters just as much. Searches that began with a question word (“who,” “what,” “why”) returned an overview 60% of the time, while full-sentence searches triggered one 36% of the time.
The intent mix is moving down the funnel, which should worry anyone selling something. Semrush tracked commercial-intent queries triggering overviews at 8.15% in October 2024, then 18.57% a year later. Transactional queries jumped from 1.98% to 13.94% over the same span. Google started with informational questions and is now answering the searches that used to send buyers to your site.
One more pattern from Semrush: when an AI Overview appears, other SERP features almost always come with it. Related Searches co-occurred 95.32% of the time and People Also Ask boxes 90.03% of the time. The ten blue links you used to compete for are getting pushed further down a busier page.
What AI Overviews do to your clicks
This is the part that pays your salary, so slow down here. Pew found that on visits to search pages carrying an AI summary, users clicked a traditional result only 8% of the time. On pages without a summary, they clicked 15% of the time. The overview roughly halves the odds that anyone leaves Google for your page.
Ahrefs quantified the damage by position. Its April 2025 study of 300,000 keywords found the presence of an AI Overview cut click-through rate for the number-one organic result by 34.5%. Its late-2025 update, comparing Google Search Console data across two years, put the top-position drop closer to 58%. Position two fell about 51%, position three about 46%. The decline reaches all the way down page one.
Then there is the citation trap. You might assume that being cited inside the overview rescues your traffic. Pew says otherwise: people clicked a source linked within the AI Overview on just 1% of visits. Getting quoted in the answer is a branding win, not a traffic win. Plan accordingly.
The citation surface: what AI Overviews pull from
Call the pages and entities eligible to be pulled into an AI answer your citation surface. It has three layers, and each of the AI Overviews statistics maps onto one of them.
The first layer is entity clarity. Pew found that .gov sites made up 6% of AI Overview citations while accounting for only 2% of standard organic results. Google trusts sources it can identify and verify, and government domains carry that trust by default. Your brand has to earn the same clarity through a clean entity footprint: a consistent name, a Knowledge Panel, corroborating profiles.
The second layer is extractable answers. Pew found 88% of overviews cited three or more sources and the median overview ran 67 words. Google is assembling short, factual spans from multiple pages. If your content buries the answer in paragraph nine, it never makes the cut. Write the direct answer near the top of the section, then support it.
The third layer is third-party corroboration. Wikipedia, YouTube, and Reddit together supplied about 15% of all cited sources. Google leans on places where many independent voices already agree. A single page on your own domain saying you are the best is worth less than three neutral sources saying the same thing.
Do people trust AI Overviews?
Trust is split, which is why the format keeps growing without a public backlash. Pew’s October 2025 survey found 65% of U.S. adults say they at least sometimes see AI summaries in search. Among people who have seen them, 20% called them extremely or very useful, 52% somewhat useful, and 28% not useful. On accuracy, 54% said they trust the information at least somewhat.
The behavioral tell is stronger than the opinion data. Pew found that people ended their browsing session entirely on 26% of visits to pages with an AI summary, compared with 16% of visits without one. The overview does not just take the click. It ends the search. The user got the answer and left the open web behind.
The scale problem: two billion users
Google confirmed on its Q2 2025 earnings call that AI Overviews had crossed 2 billion monthly users across more than 200 countries, up from 1.5 billion in May 2025. Half a billion new users in two months is not a pilot. It is a default behavior for a meaningful share of humanity.
That scale is why the traffic math has already shown up in publisher data. SparkToro, working with Similarweb clickstream panels, reported that by early 2026 the share of Google searches sending a click to the open web had fallen sharply, and attributed a large part of the acceleration to AI Overviews. When one in five searches answers itself and two billion people rely on it monthly, small percentage shifts move enormous volumes of traffic.
Can you still win anything when the overview appears?
The data says yes, but the prize changed shape. Since being cited inside an AI Overview recovers almost none of the click, according to Pew’s 1% figure, chasing the citation for traffic is a losing game. Chasing it for influence is not. Your brand name and your claim sit inside an answer that two billion people see monthly, which is a branding placement you could never buy at that scale.
There is also a defensive angle worth understanding. Pew found overviews cluster on informational, question-shaped queries and appear far less on short, navigational ones. If a meaningful slice of your traffic comes from branded or transactional searches, those queries trigger overviews less often, so that traffic is more insulated. The strategic move is to know which of your money queries carry an overview and which do not, then defend the exposed ones by becoming a cited source and protect the insulated ones by strengthening branded demand. The AI Overviews statistics do not say the click is gone everywhere. They say it is gone unevenly, and the teams that map the unevenness make better decisions than the ones reacting to a single headline number.
Which industries see the most AI Overviews
AI Overviews are not spread evenly across topics, and knowing where they cluster tells you how exposed your category is. Semrush’s November 2025 data found the most AI-Overview-saturated industries were Science at 25.96%, Computers and Electronics at 17.92%, and People and Society at 17.29%. The least affected were Real Estate, Shopping, and Arts and Entertainment, all under 3%.
The pattern follows intent. Informational and research-heavy categories draw the most overviews, because those are the questions Google feels confident answering directly. Transactional categories like shopping still send more clicks, for now, because Google has less incentive to answer a buying query itself. If you operate in a high-saturation category, the 58% click drop is your baseline reality. If you sit in a low-saturation one, treat the trend as a warning shot, because the commercial-intent migration Semrush tracked shows Google steadily expanding into exactly those categories.
There is a paid-search wrinkle worth noting too. Semrush found Google Ads at the bottom of AI Overview result pages jumped to 25% of those SERPs by late 2025, up from under 1% in March 2025. Google is not just answering more queries itself. It is monetizing the pages where it does, which signals the format is permanent revenue infrastructure, not an experiment it might roll back.
What the numbers mean for your traffic model
Put the AI Overviews statistics together and they describe a single structural shift, not a set of disconnected data points. One in five searches now opens with an answer, that answer halves your click odds and cuts your top-position traffic by up to 58%, being cited inside it recovers almost none of that traffic, and two billion people rely on the format monthly. A traffic model built on ranking first and collecting the click is quietly obsolete for a large and growing share of queries.
The brands adapting fastest have stopped treating organic traffic as the only scoreboard. They track citation share, the percentage of their money queries where an AI Overview names them as a source, alongside branded search growth and revenue from channels Google does not mediate. Traffic can fall while influence rises, and the AI Overviews statistics reward the teams that measure influence directly instead of mourning the click.
Audit your citation surface first
Do not start by rewriting content. Start by measuring what Google can already see. Run your top twenty commercial queries and note which ones trigger an AI Overview and which sources it cites. If your competitors show up and you do not, the gap is almost always in the citation surface: weak entity clarity, answers buried too deep, or too little third-party corroboration.
Fix entity clarity first, because it gates the other two. Then rewrite your highest-intent pages so the answer sits in the first two sentences of each section. The AI Overviews statistics all point the same direction. The click is getting scarcer, the citation is becoming the prize, and the brands that win the citation are the ones Google can identify, extract, and verify without effort. Build for that, or watch the 58% drop find your best pages next.