Google's AI Summary Now Covers Most Buying Searches

What the study tracked

Semrush picked 400 keywords where someone is clearly shopping, then checked every month whether Google put an AI Overview on the results. The keywords came in five flavors:

• Alternatives: "[brand] alternatives"
• Brand versus brand: "[brand] vs [brand]"
• Best X for Y: "best accounting software for freelancers"
• Price or cost: "how much does [thing] cost"
• Reviews: "[brand] reviews"

A year ago, the spread was wide. Review searches got an AI Overview 30% of the time; brand-versus-brand searches got one 63% of the time. By August 2026 the lowest group sat above 82% and the highest hit 96%. Google stopped being picky about which buying question earns a summary.

Semrush also reports that the 12% of keywords with no AI Overview made up under 3% of total search volume. The searches people run most are the ones that got covered first.

Why this matters if you sell haircuts, not software

Fair objection: banking, software, and ecommerce are not your market. Nobody searches "best dog groomer alternatives" the way they search for accounting tools. The study says nothing direct about local service businesses, and I won't pretend it does.

Here's what it does show. Google treats "reviews," "cost," and "best" as questions it should answer itself. Those are the exact words your customers type before they call you. Google already summarizes searches for your business name, and this study suggests the buying-intent searches around it are heading the same way.

When the summary writes itself from whatever it can find, the question becomes what it finds about you. A summary of "how much does [your service] cost" pulls from pages that state a cost. If your site says "call for a quote" and a competitor's lists a range, the competitor is easier to quote.

What these searches reward

None of this is a trick. It is a list of questions your site either answers or doesn't.

Reviews. Reviews live mostly off your site, on Google and other platforms, so the on-site move is small: say plainly what customers tend to ask about, and make sure your business details match everywhere. Getting more Google reviews is still the biggest lever here.

Cost. Give a range or the factors that move the price. A page that says "most jobs fall between X and Y, and here is what changes it" gives a summary something to cite. A page that hides everything behind a form gives it nothing.

Best, for whom. State who you are best for, and who you are not. "We're a good fit for homeowners with older plumbing; for new construction, call someone else" is quotable. "We serve all your needs" is not.

Alternatives and comparisons. Customers compare you to the other option in town, or to doing it themselves. A short, honest comparison page beats leaving the comparison to someone else's listicle. AI search leans on listicles, so if you are absent from them, your own page is the counterweight.

Keep each answer under its own question heading, in plain sentences near the top of the page. That is the same structure behind Answer Engine Optimization (AEO), the practice of getting AI systems to quote you.

Do the clicks follow?

Unknown. The study counts how often the summary appears, not whether anyone clicks through or buys. A summary can cite you and still send nobody to your site. Zero-click behavior was already common before AI Overviews. Treat getting named as the goal and measure it directly: check your own Search Console, as covered in our post on whether AEO is working.

For scale, our own numbers: in the three months ending August 30, 2026, the Cat Grooming Directory earned 12,200 impressions inside Google's generative AI features report in Search Console, against about 1,400 clicks overall (a founder-provided screenshot; Search Console rounds these). Those are different measurements, so don't divide one by the other. The point is that the AI surface is already a place impressions happen.

Wasted effort to skip

• Chasing the panel's keywords. They are banking and software terms. Optimizing for them if you run a salon is wasted effort.
• Stuffing "best" and "reviews" into every heading. A summary written by a language model reads your page for meaning, not keyword density.
• Fake review pages. Invented testimonials are a legal problem under the FTC's fake review rule and a trust problem long before that.
• Treating one study as a forecast. Four hundred keywords in three industries is a snapshot. Check your own searches in a private window this week and write down what you see.

What to do this week

1. Search five buying questions about your service in a private window. Write down whether an AI Overview appears and who it names.
2. Pick the one question your site answers worst, usually cost, and write a plain page for it.
3. Re-run the same five searches in a month. That is your data, and it beats any panel.

If you would rather have someone run that check and fix the gaps each month, that is part of what getting found on Google and AI covers.

FAQ

Do AI Overviews show up on local searches like "plumber near me"?

The Semrush study did not track local searches, so it cannot say. We have a separate post on how AI Overviews handle "near me" searches. Check your own area in a private window.

What is an AI Overview?

An AI Overview is a summary Google writes at the top of some search results, drawing on pages it considers relevant. It sometimes links to its sources.

Does being in the AI Overview mean more clicks?

Not necessarily. The study measures how often the summary appears, not click-through or sales. Measure your own traffic and impressions in Search Console.

Should I publish a page with my prices?

If you can state a range or the factors that move the price, yes. A page that states a cost gives both customers and AI summaries something concrete to quote. If your pricing really is custom, say what drives it.

Is the Semrush study reliable?

It is a useful trend line from a reputable tool vendor, with clear limits: 400 keywords, three industries, U.S. desktop, one 12-month window. Use it as a direction, not a guarantee.

How often should I check what Google says about my business?

Monthly is enough for most small businesses. Use the same handful of searches each time so you can compare.