AI Overviews Now Cover 88% of Tracked Commercial Keywords

By Priya Raman, AI SEO, GEO & Martech Analytics. Last updated: 2026-10-22

The queries in this panel are the ones marketers actually fight over: "X alternatives", "brand vs brand", "best X for Y", pricing, and reviews, across banking, software, and ecommerce. These are decision-stage searches, the last click before a shortlist. A year ago, roughly half of them still opened with ten blue links. The Semrush panel says that era ended sometime in 2026. This explainer reports what the study measured, shows the full table, and marks exactly where the evidence stops, because the headline number is already traveling further than the data underneath it. It extends our earlier piece on how AI Overviews are changing search traffic, which covered the click side of the same shift.

What did Semrush actually measure?

A fixed panel, checked monthly. Semrush built 400 keywords across three industries (banking, software, ecommerce) and five commercial modifier groups: alternatives, brand versus brand, best X for Y, pricing and cost, and reviews. Using its US desktop data, it recorded each month from September 2025 through August 2026 whether each keyword triggered an AI Overview. The headline metric, keyword-weighted coverage, is the share of keywords in the panel that triggered one. A second metric weights each keyword by estimated monthly search volume. The study, written by Luke Harsel with Nikola Manojlović contributing, was published on Semrush's blog on October 21, 2026.

Two design facts bound everything downstream. The data is US desktop only, so mobile behavior and other markets are outside the sample. And the study records presence, not outcomes: it cannot say what the summaries did to traffic, because it never measured a click.

How did coverage move, group by group?

Reviews are the story inside the story. A year ago a "product review" search was the least likely commercial query to draw an AI summary; by August 2026 it was the second most likely, which means the research step of a purchase now usually begins inside Google's synthesis of other people's reviews. Brand versus brand queries, the classic comparison-page battleground, started highest and grew least in percentage terms, but at 86.59% they are saturated too. The industries converged as well: software at 91%, banking at 89%, and ecommerce at 85% by August 2026, with three of the fifteen industry-modifier cells (best-X-for-Y in banking, alternatives in software, best-X-for-Y in software) at 100%.

One flaw in the study is worth naming, because PPC Land caught it and headline writers will inherit it: Semrush's key takeaways describe the alternatives figure as "appearing on more than 82% of brand comparison SERPs," but alternatives and brand-versus-brand are separate groups in its own methodology. The 82.81% belongs to alternatives. The underlying table is consistent; the summary sentence is not.

What does the study not establish?

Four limits, each material. First, no outcome data: coverage is not clicks. The click evidence comes from other studies of mostly informational queries, including Pew Research Center's July 2025 finding that users clicked a link 8% of the time with an AI summary present versus 15% without, and Seer Interactive's September 2025 update reporting a 61% organic CTR fall across 3,119 informational queries, both as summarized in PPC Land's coverage. Those numbers describe a different query class than this panel and should not be bolted onto the 88%.

Second, panel arithmetic. Four hundred keywords across fifteen cells averages under 27 keywords per cell, and the study does not publish cell sizes. PPC Land's reconstruction suggests cells around 21 to 29 keywords, where one keyword is worth three to five points, so cell-level figures like the 100% readings should be read as "every tracked keyword in a small cell," not "every searcher."

Third, the volume weighting rests on Semrush's modeled search volume estimates, not measured counts. The 97.08% figure is therefore an estimate built on estimates, directionally interesting and precisely wrong by an unknown amount.

Fourth, the source. Semrush sells the AI visibility tracking the study recommends, the study closes with a trial invitation for it, and Semrush is now an Adobe company, with the study carrying that branding. None of that disqualifies the data, and the methodology is disclosed well enough to argue with, which is more than most vendor research offers. It does mean the framing ("ranking number one likely won't lead to as much traffic," a claim the panel itself contains no traffic data to support) should be discounted to what the tables actually show.

What should a marketing team change on Monday?

Less than the headline urges, and more than nothing. If your category sells through comparison, alternatives, and review content, assume the first impression on Google is now a synthesized answer and audit what that answer says about you and your rivals, the way the study's own Mailchimp versus Klaviyo screenshot illustrates: for one brand, its own page made its case; for the other, a third-party site did, and the summary cited YouTube and review sites alongside vendor pages. The operational habits follow from that asymmetry: keep comparison and pricing pages current, because the summaries quote them (our pricing transparency audit shows how often they cannot); earn presence on the third-party surfaces the answers cite; and track AI citations as a standing metric instead of discovering your position in a vendor study once a year. Pair that with the new paid layer arriving in the same answers economy, covered in our ChatGPT ads explainer, and the discipline is complete: earn the citation, measure the citation, and only then consider renting placement beside it.

FAQ

What does the 88% figure mean, exactly?

In August 2026, 88% of the 400 US commercial keywords in Semrush's fixed panel triggered an AI Overview on desktop, up from 45.75% in September 2025. It is a share of tracked keywords in a decision-stage panel, not a share of all Google searches, and not a traffic figure.

Which commercial queries are most covered by AI Overviews?

Best-X-for-Y queries, at 96.12% of tracked keywords in August 2026, followed by reviews at 89.04% and brand-versus-brand at 86.59%. Every modifier group in the panel ended the period above 80%.

Does the study show AI Overviews reduced website traffic?

No. The study records whether an AI Overview appeared for each keyword and contains no click, traffic, or conversion data. Traffic effects are documented in separate studies of mostly informational queries, such as Pew Research Center's July 2025 click analysis and Seer Interactive's September 2025 CTR update.

How much should I trust a vendor study like this?

The methodology is disclosed (fixed panel, US desktop, monthly checks), which makes the coverage trend credible as a panel measurement. The limits are the narrow panel, small cells, modeled volumes, and Semrush's commercial interest: it sells AI visibility tracking and is now owned by Adobe. Trust the direction, cite the panel size with the percentage, and verify outcomes in your own analytics.

What is the difference between keyword-weighted and volume-weighted coverage?

Keyword-weighted coverage counts each of the 400 keywords equally: 88% in August 2026. Volume-weighted coverage weights keywords by Semrush's estimated search volume: 97.08%, because the highest-demand commercial queries were the most likely to show an AI Overview. The volume figures depend on Semrush's modeled estimates.

Sources

• PPC Land: Google AI Overviews now hit 88% of tracked commercial keywords, Semrush finds (full methodology, tables, and limitations)
• Seer Interactive: AIO impact on Google CTR, September 2025 update (3,119 informational queries; organic CTR on AIO queries down 61%)
• PPC Land: Google disputes Pew study showing AI Overviews reduce clicks by half (Pew's 8% versus 15% click finding and Google's rebuttal)
• GTMStack: how Google AI Overviews are changing search traffic (the click-side evidence: CTR and zero-click studies)
• GTMStack: track AI citations (the standing measurement habit this study argues for)

GTMStack publishes original audits of the AI marketing stack. If you want the next one when it lands, the newsletter is the place to get it.