AI Overviews Now Sit on 88% of Buying Searches

You search "best invoicing software for freelancers," and before a single blue link there's a paragraph with the answer, a few names, and a few sources. Most of the people who type your category's version of that query get the same treatment now. The question for you is whether your name is in that paragraph.

What the study measured

Semrush picked 400 keywords that signal someone is close to buying. Five kinds of phrasing: alternatives ("X alternatives"), brand versus brand, "best X for Y," pricing and cost, and reviews. It checked each keyword every month for a year and recorded whether an AI Overview appeared.

The headline numbers, per Semrush:

• Share of keywords with an AI Overview: 45.75% in September 2025, 88% in August 2026.
• Weighted by search volume: 97.08% in August 2026. The 12% of keywords still without an Overview made up under 3% of volume.
• Reviews queries: 30.14% to nearly 90%, the biggest jump.
• "Best X for Y" queries: 96.12% by August 2026, the highest of the five.
• Brand-vs-brand queries started highest (63.41%) and grew least, to over 86%.

PPC Land's write-up is where "nearly doubled" comes from: 88 divided by 45.75 is about 1.92.

Read the limits before you panic

Three limits, all stated by the coverage of the study.

The panel is 400 keywords. That's a sample someone chose, not Google's whole index. Semrush doesn't publish how many keywords sit in each modifier group.

The industries are banking, ecommerce, and software. A plumber, a groomer, or a dentist is not in the data. Local searches follow different rules; our post on why AI Overviews skip "near me" searches covers how.

It's U.S. desktop only. Phones are where most local searching happens, and the study says nothing about them.

So the fair reading is narrow. For online buying research in these categories, an AI summary is now the default first screen. If your business sells something people compare before buying (software, courses, products shipped anywhere, professional services priced online), you're in the zone this data describes.

Why reviews and "best for" queries matter most

Those two groups are where a small business has the most to gain or lose.

A "best X for Y" search asks the engine to pick. The Overview has to name candidates, and it builds the list from pages that already rank things, which fits the pattern in our look at why AI search leans on listicles. If the pages it reads never mention you, you're not a candidate.

A reviews search asks what people say. The Overview summarizes review sites, forums, and comparison pages. Your own site is one input among several, and often not the loudest.

What to do about it

Nothing here is exotic. Semrush's own recommendations are mostly housekeeping, and they hold up.

1. Run your own queries. Search your category with each of the five phrasings: "[your thing] alternatives," "best [your thing] for [your customer]," "[your thing] cost," "[your thing] reviews," and "[you] vs [competitor]." Screenshot what the Overview says and which sources it lists. That's a ten-minute exercise and it's the only data about your business in this whole post.
2. Find the sources it cites. If three roundup sites and a forum thread keep showing up, that's the map. Your job is to be accurate on those pages, not to out-write them.
3. Keep your comparison and pricing pages current. Dated numbers get skipped or, worse, quoted wrong. State what's included in plain lists a summarizer can lift cleanly.
4. Ask for reviews where the Overview looks. If it leans on a particular review site for your category, a steady trickle of honest reviews there counts for more than a pile on a site it ignores. Our guide to getting more Google reviews covers the asking part.
5. Track mentions, not just rankings. A page can hold position three and still be absent from the paragraph above it. This is the core of Answer Engine Optimization (AEO): ranking and being quoted are two separate results.

Shortcuts that backfire

• Stuffing "best" and "alternatives" into every page. The Overview reads the content, not the keyword density. A page claiming to be "the best" with nothing to back it up is the easiest kind to ignore.
• Writing fake comparison pages. An "X vs. Y" page that quietly rigs the result for yourself reads as what it is, and it's a trust problem if a human finds it first.
• Buying reviews. Beyond the moral problem, the FTC has a rule on fake reviews, and Overviews summarize review sentiment. Fakes get you the wrong summary.
• Treating one study as a law. Semrush's expectation that the remaining gaps will keep closing is a projection, labeled as one in the study. It may hold. It's not a measurement.

If maintaining comparison pages, review profiles, and a monthly check of what AI says about you sounds like a job you don't have the hours for, that's the kind of work our search and AI visibility service handles, starting with an audit of where you stand now.

FAQ

Do AI Overviews show up on 88% of all Google searches?

No. The 88% is the share of 400 hand-picked commercial keywords in banking, ecommerce, and software that triggered an Overview in Semrush's August 2026 U.S. desktop data. It's not a figure for all searches, and it says nothing about local or informational queries.

Does this apply to my local business?

Not directly. The panel had no local service keywords. If customers compare your category online before buying, the pattern is worth checking for your own queries, but don't assume the 88% applies to "near me" searches.

What does "keyword-weighted" versus "volume-weighted" mean?

Keyword-weighted counts every keyword equally: 88% of the 400 had an Overview. Volume-weighted gives popular keywords more weight: 97.08% of searches, by volume, landed on a keyword with one. The popular queries are almost all covered.

Which kind of search grew fastest?

Review searches, per Semrush: from 30.14% of tracked keywords to nearly 90% over the year. "Best X for Y" searches ended highest at 96.12%.

Should I stop caring about regular rankings?

No. Overviews pull from pages that rank and from sources they trust. Rankings still feed the pipeline. The change is that rank alone no longer tells you whether you're named in the answer.

How do I check whether I'm in the Overview?

Search your category using the five phrasings above, in a logged-out browser window, and note which businesses and sources appear. Repeat monthly. Results vary by person and by day, so look for patterns across several searches, not one screenshot.