New analysis says AI Overviews appear in about 87% of queries. Ahrefs says 20.5%. Both numbers are correct.

The gap is not a contradiction. The two numbers reflect a measurement difference, and the difference matters more than either headline. Peec sampled commercial, buying-intent prompts. Ahrefs sampled 146 million broad keywords. Different inputs, different outputs, both useful once you read them correctly.

For brands deciding where to invest AI search effort, the actionable takeaway is buried under those numbers: AI Overviews are nearly ubiquitous on the searches that lead to purchases, even if they are sporadic across Google overall. The piece below covers what Peec measured, why the number disagrees with broader studies, and how to read AI Overview data without being misled by the headline.

What Peec actually found

Peec AI analyzed 500,000 prompts sampled in April 2026, then measured how often Google's AI Overviews appeared.

The headline number: AI Overviews showed up in about 87% of those prompts. Within that overall figure, the breakdown is sharper:

• Decision-stage prompts, the comparison and "best X for Y" searches buyers use when narrowing options, returned AI Overviews 88.5% of the time.

• Two-word queries returned them 64.6% of the time. Prompts of 11 to 15 words peaked at around 89%.

• Outside the EU, AI Overviews appeared on 90.3% of sampled searches. Within the EU, 76%. France registered at 0%, because Google has not launched AI Overviews or AI Mode there.

The longer and more commercially loaded the prompt, the more likely an AI Overview appears.

Why the number disagrees with broader studies

Other studies measured different things and got different numbers. None of them is wrong.

• Ahrefs analyzed 146 million keywords across a broad sample and found AI Overviews on 20.5%.

• A randomized field experiment last month found AI Overviews on 42% of queries.

• Earlier samples from Authoritas and SE Ranking landed near 30%.

• Peec landed at 87% on commercial prompts.

The gap is the sample. Peec excluded navigational queries (standalone brand names, log-in searches, anything where the user knows exactly where they want to go) and focused on commercial decision-stage prompts. Broad keyword studies include those navigational and short-tail queries, which rarely trigger an AI Overview. Strip them out and the rate climbs.

So the right reading of the 87% figure is not "AI Overviews appear on 87% of Google." It is "AI Overviews appear on 87% of the commercial buying-intent searches Peec sampled." Different sentence, different strategy.

The methodology caveat Peec disclosed

The honest detail worth carrying forward.

Peec used machine-learning classifiers to sort its 500,000 prompts by funnel stage and query type, then measured AI Overview presence within each bucket. Peec did not publish accuracy figures for those classifiers, which means the funnel-stage and query-type labels are model-assigned rather than hand-checked.

The 87% headline is solid. The sub-breakdowns by funnel stage depend on classifier accuracy that has not been independently verified. For brand-level decision-making, the macro pattern (commercial prompts trigger AI Overviews more often than navigational ones) holds. The exact 88.5% on decision-stage prompts is directionally right but should not be treated as a precise number.

How to read AI Overview studies without being misled

Three questions to ask before acting on any AI Overview presence figure.

1. What was the sample?

A number from 146 million broad keywords means something different from a number from 500,000 commercial prompts. Always check: how was the sample drawn, what types of queries were included, and what was excluded. A study that strips out navigational queries will always report a higher AI Overview presence rate than one that includes them, and neither is wrong.

2. How were queries classified?

If the study breaks down by funnel stage, query type, or vertical, ask how those categories were assigned. Hand-coded by humans? Inferred by classifier? If by classifier, has accuracy been disclosed? Sub-bucket numbers built on unverified classifiers are directional, not precise. Treat them as patterns to act on, not exact figures to plan against.

3. What is the geographic and temporal scope?

AI Overview presence is uneven across regions and changes month to month. A 90% number outside the EU and 0% in France are both true at the same time. A study from January 2026 may already be stale by mid-year because Google has rolled out AI Mode and changed how Overviews trigger. Always note the date and the geographic mix.

Asking those three questions takes a minute per study and prevents the most expensive mistake in AI search planning, which is acting on a stat that did not measure what you thought it measured. Our research on why AI citations may not be the best visibility metric to track for AI search goes deeper into why the measurement infrastructure across the industry is uneven, synthesizing 25+ studies covering hundreds of millions of citations.

What the Peec data means for your strategy

The strategic point under the variance is simple: AI Overviews are nearly ubiquitous on the searches that lead to purchases.

If your visibility work targets only informational queries, you are missing the searches buyers run when they are about to spend money. Decision-stage queries (comparisons, "best X for Y," "X vs Y", buying guides) are where the AI Overview is most likely to be the first thing the buyer sees, which means it is also where citation in the Overview has the largest influence on which brands the buyer considers.

Three concrete moves follow.

First, audit your visibility on commercial decision-stage queries specifically, not just informational ones. Pulling AI Overview presence on your top 50 commercial queries gives a far more useful picture than a blended average across all your tracked terms. The case studies of brands that grew durable AI visibility started here.

Second, prioritize content that earns citation on those commercial queries. Comparisons, vs-pages, buying guides, and "best for [use case]" content do the heavy lifting in the buying funnel, and they are where AI Overview citation produces the most downstream revenue.

Third, watch AI Mode as it links to AI Overviews. Google is now connecting them so a follow-up question moves a user from an Overview into AI Mode without leaving Search. The AI Overview is increasingly the entry point to a longer AI-mediated session, which raises the value of being cited in the first answer.

Read the next AI Overview number with the methodology in mind

The Peec study is genuinely useful once you read it correctly. AI Overviews are nearly ubiquitous on commercial decision-stage searches, even though they appear far less often across Google overall. Both facts coexist, and the strategy follows from holding both.

The general rule for every AI search study that crosses your desk: check the sample, check the classification, check the scope. Most disagreements between studies are not disagreements at all. The differences come from different samples measuring different things, and the right action falls out of reading them correctly.

The cleanest first step is a baseline view of your brand's AI Overview presence on the commercial queries that actually move revenue, separated from informational coverage. See how Passionfruit's GEO service builds that view on top of a solid SEO foundation, look at the cross-platform tracking inside Passionfruit Labs, and talk to the team about where your numbers stand.

Frequently asked questions

Do AI Overviews really appear on 87% of Google searches?

No. Peec AI found AI Overviews on 87% of the 500,000 commercial, buying-intent prompts it sampled, not 87% of all Google searches. Broader studies covering all query types report lower rates: Ahrefs found 20.5% across 146 million keywords, Authoritas and SE Ranking landed near 30%, and a randomized field experiment found 42%. Both numbers are correct because they measured different samples. The 87% figure is specifically about commercial decision-stage queries.

Why do AI Overview studies disagree with each other?

Mostly because they measure different samples. Studies that include navigational queries (standalone brand names, log-in searches) report lower AI Overview presence because those queries rarely trigger an Overview. Studies that focus on commercial or decision-stage prompts report higher presence because those queries trigger Overviews far more often. Geography matters too: AI Overviews appear on about 90% of searches outside the EU and 76% inside, with France at 0% because Google has not launched the feature there.

What types of queries trigger AI Overviews most often?

Commercial decision-stage prompts (comparisons, "best X for Y," "X vs Y," buying guides) and longer prompts. In Peec's sample, decision-stage prompts triggered AI Overviews 88.5% of the time, two-word queries triggered them 64.6% of the time, and prompts of 11 to 15 words peaked near 89%. The more commercially loaded and detailed the query, the more likely an AI Overview appears.

What should marketers do with the Peec data?

Audit AI Overview presence on commercial decision-stage queries specifically, not a blended average across all tracked terms. Prioritize content that earns citation in those commercial Overviews (comparisons, vs-pages, buying guides, "best for" content), because that is where citation has the largest influence on which brands the buyer considers. And watch AI Mode, which Google is now linking to AI Overviews so the Overview becomes the entry point to a longer AI-mediated session.

How should I read the next AI Overview study I see?

Ask three questions before acting on the headline. What was the sample (broad keywords vs commercial prompts)? How were queries classified (hand-coded or model-assigned)? What is the geographic and temporal scope? Most disagreements between studies are not disagreements; they are different samples measuring different things. Reading the methodology takes a minute per study and prevents acting on a number that did not measure what you thought it measured.

Are AI Overviews and AI Mode connected?

Yes, and increasingly so. Google is linking AI Mode to AI Overviews so that a follow-up question moves a user from an Overview into AI Mode without leaving Search. AI Mode passed 1 billion monthly users in its first year. The practical effect is that the AI Overview is becoming the entry point to a longer AI-mediated session, which raises the strategic value of being cited in the first answer.