Skip to main content
← Back to all blog articles

AI Overviews on Commercial Searches: Which Pages to Update Now

Sergey Filatyev 7 min read
AI Overviews on Commercial Searches: Which Pages to Update Now

A year ago, Google showed an AI Overview on roughly half of the commercial searches Semrush tracks. Now it shows one on almost all of them. A Semrush study published on October 21, 2026 found AI Overviews on 88% of the buying-stage queries in its panel, compared with 45.75% in September 2025. The biggest gains came on searches about reviews, prices and comparisons, which is where a buyer decides who to contact.

Below I go through what was measured, which conclusions carry over to smaller markets such as Ukraine and which don’t, and which pages on your site deserve a second look. One caveat up front: the study covers US desktop results only. It has no data for Ukraine.

88%
of commercial queries show an AI Overview
August 2026, up from 45.75% in September 2025
97.08%
when weighted by search volume
same keyword panel
96.12%
of "best X for Y" queries
highest of the five groups
Source: Semrush, 400 commercial keywords, US desktop

What was measured

Semrush used a fixed panel of 400 commercial keywords across three industries: banking, ecommerce and software. Every month from September 2025 to August 2026 it checked whether Google returned an AI Overview for each one. The keywords fall into five groups by phrasing:

  • alternatives to a specific product;
  • “best X for Y”;
  • price and cost;
  • reviews;
  • brand vs. brand comparisons.

The headline 88% counts every keyword equally. Weight each keyword by its monthly search volume and the share rises to 97.08%. The authors say the keywords that still lack an AI Overview account for under 3% of total volume.

What the study leaves out: clicks, conversions, mobile results and other countries. It’s also worth knowing that Semrush sells tools for tracking visibility in AI answers. That doesn’t make the numbers wrong, but I’d treat a takeaway like “AI Overviews are on every query now” with some caution.

Where the growth is largest

Share of queries where Google shows an AI Overview
August 2026, % September 2025, %
  • All 400 keywords 88 %
    45,75 %
  • Ecommerce 85,11 %
    32,62 %
  • Price and cost, ecommerce 71,43 %
    4,76 %
  • Banking reviews 89,66 %
    13,79 %
  • Brand vs. brand 86,59 %
    63,41 %

Source: Semrush, October 2026

If you run an online store, two things stand out. Ecommerce started with the fewest AI Overviews of the three industries and has nearly caught up within a year. And product price queries, which barely triggered an AI answer in September 2025, now trigger one most of the time.

Brand comparisons already had the most AI Overviews a year ago, so they grew the least. The other groups have closed the gap.

What about “buy” queries

An earlier Semrush study, published on July 2, 2026, used a different setup: more than 600,000 US keywords from November 2025 to April 2026. Over those six months, the share of commercial results pages with an AI Overview grew by 71% on average. For transactional pages, where the searcher is already looking for a place to check out, the share fell by 5%. The same study notes that Google Ads and AI Overviews now appear on the same results page about twice as often as a year earlier.

The two studies measure different things, so their percentages shouldn’t be added up or compared directly. The direction matches, though: AI answers show up while people are still choosing, and less often once they’re ready to pay.

My read for businesses: part of the job that pricing, review and comparison pages used to do now happens on the results page itself. A buyer can form a view of you and your competitor before opening a single site, and that view is built from whichever pages Google found and was able to cite.

What Google says about appearing in AI answers

Google’s documentation on AI Overviews and AI Mode is short:

  • the page has to be indexed and eligible to show with a snippet;
  • there are no extra technical requirements, no special schema.org markup and no separate AI files;
  • to build a response, the system may run several related searches across subtopics, which Google calls query fan-out. So the linked pages can include ones that don’t rank near the top for that query in classic results.

Google’s suggestions include not blocking crawling at the CDN or hosting level, keeping important content in text, making sure structured data matches what’s visible on the page, and keeping Merchant Center and Business Profile details current.

In other words, Google asks for no separate “AI optimization.” It comes down to how clear and complete your pages are.

Which pages to review first

The five query groups in the study map neatly onto page types that almost every commercial site has. The table below is our working framework, not a Google requirement.

Buyer’s queryPage that answers itWhat to check
”how much does … cost”pricing, price list, product pagethe price is written as text, and it’s clear what’s included and what isn’t
”… reviews”reviews, case studiesreviews with dates and names, links to third-party review sites
”… or …”, ”… vs …“comparison pagehonest criteria and who each option suits
”best … for …“category or roundup pagea short note on who it’s for, filters by use case
”… alternatives”category or alternatives pagea list of similar products or services and how they differ

A few practical notes on the table.

Put the price in text, not in an image or behind a call. A page that says “price on request” gives the system nothing to cite. If the price depends on volume, give a range and explain what moves it. Semrush’s own recommendations also include keeping pricing and comparison pages current and well structured.

Reviews don’t live only on your site. Semrush observed AI Overviews citing YouTube videos, forum threads, review sites, brand sites and competitors. That makes your Google Business Profile and third-party review platforms at least as important as the testimonials block on your homepage. Ask customers for reviews where the platform’s rules allow it.

Write the comparisons yourself. Say you sell gas and electric boilers: a page on choosing one for a house under 150 m², with a table of criteria, answers the buyer’s question better than two separate product pages. It’s a made-up example, but the logic works in any niche.

Use a structure that’s easy to parse. Headings phrased as questions, short answers under them, tables instead of long paragraphs. If your site has product or organization markup, check that the prices and names in it match what’s on the page.

If your catalog only reveals prices and specs after a click or inside images, that’s a site structure problem rather than a copy problem. Rebuilding the template is usually cheaper than patching every product page, and that’s where online store development with a clean catalog structure comes in.

How to see what’s happening on your site

Search Console now has a separate Generative AI performance report in the Performance group. According to Google’s help page, it has been available to all sites worldwide since August 31, 2026. It shows impressions in AI Overviews and AI Mode only, with no clicks, position or CTR. You can break the data down by page, country, date and device. If your site gets few impressions in AI answers, the report may not appear at all.

The same impressions are already included in the standard Performance report under the Web search type, so don’t add the two reports together.

Here’s the order we suggest:

  1. Open the Generative AI performance report, filter by country and see which pages get impressions.
  2. Open the standard report for those same pages and compare clicks over recent months.
  3. In Google Analytics, check leads and purchases from organic search. Google says clicks from results pages with AI Overviews tend to be “higher quality,” but check that against your own data.
  4. Once a month, run 10–15 key queries in your niche by hand and note who the AI answer mentions and which sites it links to.

Should you opt out of AI answers

Search Console has a Search generative AI setting (Settings → Search generative AI) with three options: include your site, which is the default; exclude it; or, for a child property, inherit the parent’s setting. Per the help page, exclusion usually takes effect within 1–2 days. It doesn’t affect regular results, and model training is controlled separately through Google-Extended.

Our view: for most commercial sites, opting out makes little sense. An answer to a pricing or comparison query will still appear, just with links to your competitors. The exception is paid or licensed content that you sell as text.

What to do next

  • Check that your site has dedicated pages for pricing, reviews and comparisons, written as text.
  • Open the Generative AI performance report in Search Console if it’s available for your site.
  • Update your Google Business Profile and Merchant Center data.
  • Don’t draw conclusions from US numbers alone; look at your own impressions and leads.

Want to know which of your pages Google can cite and which it can’t? Send us your site’s address and we’ll tell you where to start.

Author: Sergey Filatyev, founder of the Veb-Dev web studio.

Tags

  • AI Overviews
  • commercial search
  • ecommerce SEO
  • Search Console
  • AI Mode
  • pricing page