How often AI Overviews appear now: the adoption data

Two data points define the current curve. In December 2025, Ahrefs measured AI Overviews on 34.5 percent of the queries in its tracking set. By March 2026 the same measurement returned 48 percent. Between those dates sits a 58 percent relative increase in roughly one quarter, which is not the pace of a feature drifting into more results. It is the pace of a company turning a dial.

The composition matters as much as the total. The growth concentrated in informational and how-to queries, where Ahrefs now finds Overviews on more than 70 percent of results pages. Ask Google what something is, how to do something, or which option fits a situation, and the odds are strong that an AI-composed answer sits above every link. Those are precisely the queries buyers use while forming opinions and shortlists, before they know which brand names to type.

This article is about how often Overviews appear. What they do to your clicks when they appear is a separate dataset with its own drama, organic click-through falling around 61 percent under an Overview while cited brands gain roughly 35 percent, and we keep that analysis in AI Overviews and your clicks. Adoption times click impact is the real exposure formula, and both factors are rising.

Ahrefs says 48 percent. Similarweb says about 43 percent. Neither is wrong, because there is no single true rate to measure. AI Overview frequency depends heavily on what you search, and every research panel samples a different universe of queries. A keyword set weighted toward informational questions will read high, since those queries trigger Overviews more than 70 percent of the time. A panel reflecting real user behavior includes navigational searches, people typing a brand name to reach a website, which almost never trigger Overviews and drag the average down.

Geography, device and account state add more variance. Google rolls features out by country and language on different schedules, mobile and desktop can differ, and logged-in users may see different treatments than clean crawlers. Measurement tooling matters too: rank trackers detect Overviews by parsing results pages at scale, and parsing changes whenever Google changes markup, which happened repeatedly through 2025.

The practical reading rule: trust trend lines within one dataset over absolute numbers across datasets. Ahrefs' own series moving from 34.5 to 48 percent in a quarter is a clean like-for-like comparison, and that is the finding that matters. For what an Overview actually is and how Google assembles one, start with Google AI Overviews explained.

Percentage of queries understates the exposure, because the queries with Overviews are not average queries. Navigational searches, a large share of raw volume, were never contestable marketing surface: someone typing your competitor's name was going to their site regardless. The queries that grow categories are the informational and comparative ones, where a buyer does not yet know the answer. Those are exactly where Overview coverage passed 70 percent. Weight the adoption rate by commercial relevance and the effective coverage of contestable queries is far higher than the headline 48.

The second implication is positional. An Overview does not share the page politely; it occupies the top, pushes the first organic link down, and resolves many sessions on the spot. When it appears on half of all queries and most of the valuable ones, ranking first under it is winning the second screen. The contest that matters moves inside the answer: which sources it cites, which brands it names. Being that citation is a different discipline from ranking, with its own craft, covered in how to show up in AI Overviews.

Third, the trajectory has no visible ceiling. Nothing in Google's 2026 behavior suggests 48 percent is the stopping point: the quarter-long jump from 34.5 percent happened after a year of public confidence in the feature, and Google keeps expanding the query classes eligible for AI treatment. Planning for the current rate is planning for the past. The sensible baseline assumption is that any question your buyers ask in natural language will eventually get an AI answer first.

The current numbers land differently with the rollout history behind them. Google previewed the concept as Search Generative Experience in 2023, an opt-in lab feature that most of the industry treated as a curiosity. AI Overviews launched properly in the United States in May 2024, stumbled through a famously rocky first month of odd answers, and then did the thing incumbents do: iterated quietly, expanded by country and query class, and kept shipping. By late 2025 the feature had spread across more than a hundred countries and most major languages, and the December 2025 baseline of 34.5 percent already represented the largest change to the results page in Google's history.

Then came the first quarter of 2026 and the jump to 48 percent. The significance of that jump is what it reveals about Google's internal confidence. A company with Google's ad revenue does not expand a results-page format by 13 points of query coverage in a quarter unless its quality metrics, its legal comfort and its monetization models have all signed off. The expansion is Google telling you, in deployment behavior rather than press releases, that AI-composed answers are the product now and the ranked list is the appendix.

History also explains why so many teams are late. Every previous results-page feature, snippets, knowledge panels, local packs, grew slowly enough for annual planning cycles to absorb. This one moved faster than a budget cycle. A strategy document written in November 2025 with the 34.5 percent figure in it was materially wrong by March. Whatever cadence you review search strategy on, this dataset argues for shortening it.

Exposure is not evenly distributed, and you can estimate yours from the shape of your funnel. Businesses acquired through informational content, guides, definitions, how-tos, comparisons, sit in the most exposed class, because that is where Overview coverage passed 70 percent. SaaS, B2B services, education, health information and finance content all live here. If your traffic reports show most organic sessions entering through blog and resource pages, the majority of your acquisition surface now renders under an AI answer.

Transactional and local queries currently trigger Overviews less often, which gives e-commerce and local businesses a longer runway on the classic results page. But the runway shortens from both ends: Google keeps widening the eligible query classes, and buyers keep moving their research questions, the ones that decide which product page they eventually visit, into AI surfaces. The research layer of a transactional funnel is already an answer layer even where the final click is not.

The audit implication is to classify your own queries rather than trust category averages. Pull your top converting organic queries, bucket them by intent, and check Overview presence per bucket. Most teams find the exposure concentrated exactly where their content investment has been heaviest, which is uncomfortable and useful to know precisely.

AI Overviews are a summary above results. AI Mode, which Google has been expanding through 2025 and 2026, is a full conversational search experience: a chat interface with follow-up questions, deeper synthesis and fewer visible links. The adoption story of Overviews is best read as the first phase of a longer migration, in which Google moves users from a page of links, to a page with an answer on top, to a conversation that is the page.

For visibility work this changes the citation economics again. In a conversational interface, being uncited approaches being invisible, because there is no ranked list underneath to catch the spillover. The same content properties that win Overview citations, answer-first structure, verifiable specifics, clean extraction, carry into AI Mode, but the measurement has to follow the user into the conversation. We track that surface separately in our Google AI Mode guide.

First, audit your exposure rather than assuming it. Take your top 25 buying questions, the things a customer would ask Google in a full sentence, and check which ones trigger an Overview today. In most B2B and SaaS categories the informational majority will, given the 70 percent coverage of how-to and informational queries. For each one that does, record whether the Overview cites you, cites a competitor, or cites third-party sources talking about neither of you.

Second, fix the losses in citation order. Overviews cite from Google's organic index, so pages need to be indexed and competitive first, and then structured for extraction: the direct answer near the top, statistics with named sources, question-shaped headings, valid schema. Ahrefs' May 2026 study of 1,885 pages found schema markup correlates with AI citation, and the Princeton GEO study measured up to a 40 percent visibility lift from statistics, quotations and cited sources.

Third, put the check on a schedule, because both the adoption rate and the citations behind any single answer keep moving. This is the loop Reachroller automates: it tracks your question set, stores each answer with its citations, scores your mention rate over time, and generates a publish-ready fix page for questions you lose. Starter is $29 per month, and the honest caveat stands: ChatGPT tracking is live today, Google surfaces and the other engines are rolling out. Either way, run the loop somewhere, because a number that moved from 34.5 to 48 in one quarter will not wait for your next annual planning cycle.