In a study of more than 10 million keywords, Semrush found that Google showed an AI Overview for 24.61% of US searches in July 2025, then for 15.69% in November 2025. That swing is the first thing worth knowing about AI Overviews. There is no fixed rate you can quote, only a pattern of where they show up more often and where they barely show up at all.
If you run content for a brand, "does my industry get AI Overviews" is the wrong first question. The better one is which query types that trigger AI Overviews actually match the searches your own audience runs. A healthcare brand can have heavy coverage on symptom questions and almost none on "dentist near me." A finance brand can see strong coverage on tax questions and almost nothing on a stock ticker lookup. The pattern lives at the query level, not the industry label.
This piece walks through what the third-party studies actually measured, breaking down AI Overviews by industry and query type and flagging where the studies disagree, so you are not left quoting a number that means something different than you think.
When does Google actually show an AI Overview?
The short answer to when does Google show an AI Overview is qualitative, not a formula. Google says AI Overviews are meant to help with complicated, nuanced questions where an AI-written synthesis adds something beyond a normal results page, including questions that might otherwise take several searches to answer: exploring a concept, comparing options, working through a plan. Google also says plainly that AI Overviews often do not appear at all.
There is no published word count, no fixed list of trigger phrases, and no confirmed rule tied to search volume or cost per click. Every number in this piece comes from third-party tracking studies watching Google's behavior from the outside, not from Google's own trigger logic. Treat every figure below as an observed pattern in one dataset, on one date, not a permanent rule.
Which query types trigger AI Overviews most often
The clearest pattern across the big studies is that informational, explanatory queries lead. Semrush's analysis of 200,000 US keywords that had triggered an Overview between September 1 and 10, 2024 found informational keywords made up 80% of desktop triggers and 76% of mobile triggers, while transactional keywords were under 3% and navigational keywords were under 2%. That is a composition figure, the share of already-triggering queries that were informational, not the share of all informational searches that trigger an Overview, so it describes the shape of the results rather than the odds for any one query.
Question wording adds another layer. The same Semrush study found question-form keywords made up 35% of desktop and 32% of mobile triggering queries, most often starting with "how," "what," or "is." A January 2025 Cressive DX study of 4,000 US and UK queries went further and measured category-specific rates directly: about 85% of question-format keywords produced an Overview compared with about 43% of non-question keywords. Cressive is vendor research built around client-modeled queries, so treat that gap as directional rather than a universal split.
Longer, more specific queries are more exposed
SE Ranking tracked 100,013 US keywords across 20 niches in June 2024 and found a clean climb by word count: a one-word query triggered an Overview 0.80% of the time, a five-word query 11.13% of the time, and a ten-word query 19.10% of the time. Cressive's separate study found the same shape in tail terms: about 65% of long-tail queries triggered an Overview against about 34% of short-tail queries.
Search volume moves the same direction, in reverse. Semrush's 200,000-keyword sample found 82% of desktop Overviews and 76% of mobile Overviews landed on keywords with under 1,000 monthly searches. SE Ranking's volume breakdown showed high-volume head terms (100,000 or more monthly searches) triggering at just 0.23%, against double-digit rates in several lower-volume bands. The likely read is not that low volume causes an Overview. It is that low-volume keywords tend to be longer, more specific, and more informational, while big head terms skew navigational, branded, or already served by another SERP feature.
Commercial research and transactional action behave differently
Not every "buying" query behaves the same. Semrush's July 2026 study of over 600,000 keywords split commercial-intent searches (comparing options, evaluating a purchase) from transactional ones (ready to act now) and tracked them from November 2025 through April 2026. AI Overview presence on commercial-intent pages rose 71% over that window, while presence on transactional-intent pages fell 5%. Commercial queries out-triggered transactional queries in every industry in that dataset except Food & Drink, where transactional actually ran slightly ahead, 16% to 14%.
Finance showed the sharpest commercial-side growth, up 231% over the six months, followed by Computers & Electronics at 108% and Games at 77%. For a content team, this means a "best tool for X" or "X versus Y" research page is currently a stronger AI Overview candidate than a straight "buy X" page, even inside the same product category.
Industry and topic patterns, and why there is no single ranking
Studies rank industries differently because they measure different things, so treat any single industry list as one dataset's snapshot rather than a permanent order for AI Overviews by industry and query type. SE Ranking's 20-niche study found Relationships (26.62%) and Food and Beverage (24.78%) at the top and Real Estate (0.12%) and Cars (0.12%) at the bottom. Semrush's later 10-million-keyword study, measured in November 2025, put Science (25.96%) and Computers & Electronics (17.92%) ahead, with Real Estate, Shopping, and Arts & Entertainment each under 3%.
Both studies point the same direction even though the exact order differs: explanatory, conceptual topics (science, relationships, food, technology) draw more Overview coverage, while shopping, real estate, and other destination-driven categories draw less. Jobs & Education stood out in Semrush's commercial-intent data too, where more than 60% of the group's highest-value keywords triggered an Overview, the highest rate in that study.
Health, finance, and other YMYL topics are not uniformly covered
It is tempting to assume Google is cautious across the board on health, finance, legal, and political topics. The evidence says otherwise: coverage inside YMYL categories varies as much as coverage across industries does. SE Ranking's dedicated YMYL study, run on 1,200 US keywords split evenly across health, finance, legal, and politics, found legal queries triggered an Overview 77.67% of the time and health queries 65.33% of the time, while finance came in at 41.67% and politics at just 16.67%. No Overviews appeared at all for the political sample's election-related terms.
BrightEdge's 18-month tracking of healthcare and finance keywords, running from May 2024 through December 2025, tells a more granular version of the same story. By December 2025, healthcare had climbed to 88% overall coverage, with symptom and treatment queries at 93% and general education queries at 74%, while local healthcare queries sat at just 11%. Finance reached only 21% overall in the same window, but that low number hides a wide range inside it: finance educational queries reached 67%, while real-time stock-ticker queries sat at 8%. A separate BrightEdge finance analysis, published in January 2026, found "what is" style finance questions at 91% and finance calculator queries at just 9%. The lesson holds across both: the task inside the topic, not the topic's YMYL label, decides the rate.
Local search: two studies that look contradictory but are not
Local search produces the widest-looking gap in the evidence, and it comes down to what each study counted as a "local" query. A seoClarity tracking run across 432,000 keywords found local-keyword AI Overview prevalence had fallen to just 0.01% by September 2025. A Whitespark study that manually tested 540 queries across plumbers, lawyers, dentists, optometrists, medical clinics, and real-estate agents in three cities found Overviews across 68% of its sample on average.
The two are not measuring the same thing. Whitespark broke its queries into pure local intent ("plumbers in Phoenix"), informational intent ("how much do plumbers charge in Phoenix"), and hybrid intent ("do I need a plumber or can I fix this myself"). Pure local-intent queries triggered an Overview just 15% of the time in that sample, while informational local queries hit 92% and hybrid queries hit 97%. The local pack showed the mirror pattern, appearing for 93% of pure local-intent queries and only 6% of informational ones. A "near me" style search still favors the map pack. A question wrapped around a local need favors an Overview.
Recipe queries are an emerging, wording-dependent case
Recipe search does not have a solid prevalence number behind it yet, so treat any claim about recipe Overview rates with caution. What exists is a set of observed examples reported by Search Engine Journal in June 2025: a search for "chicken cordon bleu" triggered a recipe-style AI Overview on desktop, and "healthy dinner ideas" did too, but adding the word "recipe" to the query, as in "chicken cordon bleu recipe," instead produced Google's standard recipe-rich-results treatment. The desktop examples did not always repeat on mobile. This is a wording and device-dependent pattern from observed examples, not a measured rate, and it should be described that way.
The query types least likely to trigger an Overview
A few patterns show up at the bottom of nearly every study reviewed here:
Direct navigational searches, where someone wants one known destination, triggered Overviews just 0.09% of the time in Advanced Web Ranking's 8,000-keyword study, and brand-login queries sat at 0% to 4% in BrightEdge's finance data. Live or real-time data queries, like stock-ticker lookups, sat at 7% to 8% across BrightEdge's finance datasets. Calculator and tool queries came in at 9% in BrightEdge's January 2026 finance analysis. Pure local-intent queries hit just 15% in Whitespark's local-business sample, and broad local-keyword tracking from seoClarity found rates under 1%. Shopping, real estate, and arts and entertainment keywords each sat under 3% in Semrush's November 2025 industry breakdown.
The shared thread across every one of these is the task, not the topic. A user who wants a direct destination, a live number, a calculator, or a specific nearby business is not asking Google to synthesize anything, and it shows up as the mirror image of the query types that trigger AI Overviews most reliably.
What this means for your own content
Before you decide where AI Overviews matter for your brand, the real question is which searches get AI Overviews among the terms you actually target, so sort your own queries by task rather than by category. A query asking a reader to understand, compare, or plan something is the strongest candidate. A query asking for a destination, a live number, a tool, or a nearby business is the weakest one, even inside a topic area that scores high overall.
This is also where a lot of teams get stuck: they know which of their pages should be AI Overview candidates in theory, but have no way to check whether Google actually shows an Overview for their own tracked terms, or whether their pages get cited when it does. DeepSmith's AI visibility tracking covers Google AI Overviews alongside nine other engines, reporting mention rate, citation rate, and which of your own pages are actually pulled into an answer, so you can see the real pattern for your queries instead of guessing from an industry-wide study.
Once you know which of your own query types are genuinely Overview-prone, the next question is what to do about the ones that already qualify. A chunk of your traffic likely comes through the People Also Ask questions stacked under an Overview, which has its own playbook. The same goes for shopping and comparison queries, with a separate playbook of its own. This piece is the map; those pieces are the next steps.
