AI NewsGo-to-marketReported
AI Overviews now show on 90% of 'reviews' searches, Semrush finds
Semrush tracked 400 commercial keywords across banking, software and ecommerce for a year and found that every one of five buying-intent modifier groups now triggers an AI Overview between 82% and 96% of the time.
.png)
Image: Semrush
Why it mattersA commercial comparison page an SEO team targeted for an organic click now sits under an AI Overview most of the time, so the summary at the top of the page is the thing the brand needs to show up in.
Google now shows an AI Overview on 97% of US commercial searches by volume, by Semrush's own measurement of 400 keywords tracked for a year. The share for searches containing the word "reviews" rose from 30% at the start of the panel to 90% at the end, which was the biggest jump of any modifier group in the study.
Semrush published the findings in a blog post titled "AI Overviews have nearly doubled across five commercial search modifiers". The analyst, Luke Harsel, built a fixed panel of 400 keywords in five modifier groups ("alternatives", "best X for Y", "price/cost", "reviews", and "brand vs brand") across three industry categories (banking, ecommerce, and software), and used US desktop data to record whether an AI Overview appeared each month from September 2025 through August 2026.
What the five groups show
Every one of the five groups finished the year above 80% keyword-weighted coverage, and the gap between them narrowed. In September 2025, Semrush says, the range ran from 30.14% for "reviews" to 63.41% for "brand vs brand". By August 2026 the range was 82.05% to 96.12%, with "best X for Y" at the top of the ranking.
The 59-point jump for "reviews" moved it from the modifier least likely to trigger an AI Overview to the second most likely. "Best X for Y" rose by 53.4 points. "Brand vs brand" grew the least, by 23.18 points, which Semrush attributes to that modifier starting the panel with the highest coverage.
Industry made less and less difference as the panel ran. In September 2025 the gap between the lowest-coverage industry (ecommerce at 32.62%) and the highest (software at 68%) was more than 35 points. By August 2026 all three industries sat between 85% and 91%, and the gap was less than six points. Three of the 15 industry-and-modifier combinations reached 100% keyword-weighted coverage, meaning every keyword in those cells triggered an AI Overview.
Weighted by volume, the share is 97%
Semrush also recomputed the coverage by weighting each keyword by its estimated monthly search volume. On that measure, 97.08% of estimated commercial searches in August 2026 surfaced an AI Overview. The 12% of keywords in the panel that did not trigger an AI Overview accounted for less than 3% of total search volume, which the post reads as evidence that the highest-demand commercial queries are also the ones most likely to carry an AI-generated answer above the organic results.
Two things to keep in mind when reading this. First, Semrush sells an AI visibility product and the post ends with an ad for a free trial. Attribute every figure to Semrush rather than restating it as a general measurement. Second, the panel is 400 keywords in three industries on US desktop; outside that scope the numbers may look different.
For an SEO team, the practical read is that a comparison page, pricing guide, or alternatives roundup is now summarised by an AI Overview more often than a reader clicks the organic result. One Semrush example shows an AI Overview citing Klaviyo's own guide for the "Choose Klaviyo if..." paragraph, and citing a Zapier guide for the "Choose Mailchimp if..." paragraph in the same answer. The brand's own guide has to be strong enough to be quoted, and the third-party pages that describe the brand have to be accurate too.
Source
Semrush: AI Overviews have nearly doubled across five commercial search modifiers, by Luke Harsel.
This item was written by an AI system from the linked source. Reveneau is responsible for what it publishes.
Get AI News in your inbox
New developer tools, model and agent releases, and how teams are actually using them to release software. Short, and only when there is something worth reading.


