AI Overviews Jumped From 30% to 89% of Review Searches in One Year
Google AI Overviews appeared on 88% of 400 tracked US commercial keywords in August 2026, up from 45.75% in September 2025, according to a Semrush study covered by PPC Land. The biggest jump hit "reviews" queries, which went from 30% to 89% coverage. Weighted by search volume, coverage reached 97%, so comparison and review pages are now the most exposed pages you own.
For years the comfortable assumption in SEO was that AI Overviews were an informational-query problem. The "what is" and "how to" stuff. Publishers took the hit, and anyone selling software or credit cards or coffee makers could mostly shrug, because the money keywords (reviews, comparisons, "best X for Y") still looked like a normal page of blue links.
That assumption seems to be gone now. And honestly, it went faster than I expected.
Where the 88% actually comes from
Semrush tracked a fixed panel of 400 commercial keywords across three industries (banking, software and ecommerce) and five modifier groups: alternatives, brand vs. brand, best X for Y, pricing, and reviews. US desktop only. They checked the same keywords monthly from September 2025 through August 2026.
The headline number is keyword-weighted, meaning the share of keywords that triggered an AI Overview. Here's how each modifier moved over the year:
- Reviews: 30.14% to 89.04% (+58.9 points)
- Best X for Y: 42.72% to 96.12% (+53.4 points)
- Pricing/cost: 43.59% to 82.05% (+38.5 points)
- Alternatives: 48.44% to 82.81% (+34.4 points)
- Brand vs. brand: 63.41% to 86.59% (+23.2 points)
Then there's the volume-weighted number, which is the one I'd actually pay attention to. Weight each keyword by its estimated monthly searches and coverage hits 97.08%. The 12% of keywords that still show no AI Overview account for under 3% of search volume. Put differently: the keywords Google leaves alone are mostly the ones nobody searches.
This also isn't a one-off. Semrush's earlier study published in July, which used a much bigger set of 600,000+ keywords, found AI Overviews on commercial-intent results grew 71% between November 2025 and April 2026, while transactional results actually dipped 5%. Two different methods, same direction. Google is leaning into the research phase of buying and (for now) leaving the "buy now" queries mostly alone.
The fine print most coverage will skip
I want to be fair about the data, because there's a real weakness in it.
Four hundred keywords split across 15 industry-and-modifier cells works out to fewer than 27 keywords per cell on average. PPC Land backed out the likely cell sizes from the percentages and landed on roughly 20 to 30 keywords each. Ecommerce pricing going from 4.76% to 71.43%, for example, looks a lot like 1 of 21 keywords becoming 15 of 21. At that size, a single keyword flipping moves a cell by 3 to 5 points. So the per-industry, per-modifier numbers are directional at best.
It's also desktop only, US only, and it measures whether an AI Overview showed up. Not clicks, not traffic, not conversions. And, small thing but worth saying, Semrush sells the tracking tools its recommendations point you toward. That doesn't make the data wrong. It just means you should check it against your own Search Console before you rebuild anything.
The overall trend, though, I'd believe. Every one of the 15 cells went up. That's hard to fake with noise.
Why this hurts more than the informational hit did
When AI Overviews ate "how to" traffic, a lot of that traffic was low-value anyway. Commercial modifiers are different. A person typing "hubspot vs salesforce" or "chase checking account review" is close to a decision, and the page they land on often sets the shortlist.
The click data we already have isn't encouraging. Ahrefs found a 58% CTR reduction for top-ranking pages when an AI Overview appears, based on 300,000 keywords. Seer Interactive measured organic CTR on AIO queries falling 61% across 3,119 queries, and paid CTR on those same queries dropped 68%. Both of those studies leaned informational, so applying them to commercial queries is a guess on my part. But I'd be surprised if comparison shoppers click more generously than people asking how to boil an egg.
The Seer data has one detail that matters a lot here. Brands cited inside the AI Overview got about 35% higher organic CTR (0.70% vs. 0.52%) than brands that weren't cited. Seer is careful to say they can't prove the citation causes the lift, since strong brands probably earn both. Still, it's the clearest signal we have about where the remaining clicks go.
Ranking first on a review query used to mean you owned the shortlist. Now it mostly means you're eligible to be quoted.
Who's actually getting quoted
Semrush's own example is a "mailchimp vs klaviyo" Overview. It pulls Klaviyo's own comparison guide for the "choose Klaviyo if" recommendation, Zapier for the Mailchimp side, plus a YouTube video and a third-party review site. So the vendor's own page got cited, on a query about its own brand, which (to be fair) is the easiest possible version of this game.
The broader pattern is less flattering for brand sites. In Semrush's AI Visibility Index covering 126 million prompts, AI Overviews cited an average of 9.2 sources per response, and YouTube was the single largest source with 35.2 million mentions, more than double Reddit. That lines up with what we saw when AI Overviews hit 83% of brand searches and Wikipedia ended up writing a lot of them. The citations go to whoever has the clearest, most extractable answer, not necessarily whoever ranks first.
Google's own guidance on AI features says the usual SEO best practices still apply and that AI Overviews use "query fan-out," running multiple related searches to find supporting pages. Which, in practice, means your comparison page competes against everything Google finds for the sub-questions too. Pricing, integrations, "is it good for small teams." If you don't answer those, somebody's YouTube video will.
A 30-minute Search Console check before you rebuild anything
I wouldn't tear down your review pages based on a 400-keyword panel. I would spend half an hour finding out whether it describes your site.
- Filter Search Console queries by modifier. Use a regex filter on Queries for
review|vs|alternative|best|price|pricing|cost. Export 16 months so you can compare the same quarter year over year. - Compare CTR at stable position. Look only at queries where average position held within roughly one spot year over year. If CTR on those dropped 30% or more while position didn't move, that's the AI Overview tax showing up in your own numbers. If it dropped under 10%, you're probably fine for now.
- Check who's cited on your top 10. Search your ten highest-impression commercial queries in an incognito window and write down every source in the Overview. If you're cited on fewer than 3 of 10, that's your gap.
- Rewrite the top of the page, not the whole page. Put a plain verdict in the first two or three sentences ("Choose X if you need Y; choose Z if your budget is under $N"). Add a short pricing table with a date on it. Overviews quote things that are easy to quote.
- Make one video. Given YouTube's share of citations, a ten-minute comparison video on your biggest "vs" query is probably the cheapest new citation source you can get. It doesn't need to be polished.
One aside on the paid side: the July Semrush study found Google Ads and AI Overviews showing up on the same results page about twice as often as a year earlier. If organic CTR on your comparison terms is sliding, check whether your "vs" and "alternatives" search campaigns are quietly picking up the slack. Some of that lost organic traffic may be getting resold to you.
Where I think this goes by spring
My guess: by the end of Q1 2027, Semrush or someone like them will publish a panel showing transactional queries ("buy," "coupon," "near me") above 50% AI Overview coverage too. The research phase got absorbed in about twelve months. I don't see why the checkout phase would take much longer, especially with ads now sitting on the same pages as the Overview.
And, slightly off topic, this whole thing makes me think the old "rank for everything, convert some of it" model was always a bit fragile. We just didn't notice because the clicks kept coming.
I don't think review pages are dead. I think they're turning into source material, and the brands that do fine will probably be the ones who accept that and write them to be quoted, not just read.
