AI Overviews on Branded Searches: What Google Tells People Who Look You Up
AI Overviews on branded searches went from the exception to the norm at the end of September. One tracker had Google showing an AI Overview on about 27% of branded queries for most of the month and on more than 80% by the last few days of it, so when somebody types your company's name into Google now, there's a good chance the first thing they read about you is a summary Google wrote from pages you may or may not have written yourself.
My reason for paying attention to this one is who's doing the searching. A person typing your brand name already knows who you are. They're checking your pricing, looking for the login page, reading reviews before a demo, or trying to decide between you and the company a friend mentioned. For most of us those were the clicks we could count on in the search report, and Google has now put its own answer in front of a lot of them.
The short version
DemandSphere tracked AI Overviews on about 26% to 27% of branded keywords in the first days of September, then 90.48% on September 27 and above 80% for the two days after. Chris Long found them on 93 of the 100 large brands he checked.
Google hasn't announced a change, and nobody has published what this does to branded click-through rate yet.
Google builds that summary from whatever it reads about you, and in my own study data Gemini cited brands' own websites less than any other engine did.
Search your own name, read the summary, and open every source it cites.
What changed with AI Overviews on branded searches in September
Two sets of numbers came out within a day of each other. Ray Grieselhuber at DemandSphere published daily tracking of branded keywords across all markets and devices (DemandSphere). AI Overview presence sat between about 26% and 35% for the first half of the month, started bouncing around on September 18 (57% one day, 31% the next), and then climbed to 69% on September 26 and 90.48% on September 27, staying above 80% on the 28th and 29th. Averaged out, that's about 27% for September 1 through 5 and about 84% for September 27 through 29. One detail in their method changes how you read that number, so I'd keep it in mind: an AI Overview counted as present whether or not it cited the brand being searched.
Separately, Chris Long, co-founder of Nectiv, tested 100 large brands and found AI Overviews on 93 of them, including Reddit, Salesforce, Amazon, Adobe, Samsung and Apple (Search Engine Roundtable). In most cases the overview showed up partway down the page, under the brand's own site. Adobe was the exception, with the overview sitting above its homepage. Searches for Google itself and for news publishers didn't get one. Long had noticed something similar back in July, and Search Engine Land's write-up notes that Google hasn't announced anything (Search Engine Land).
Google's AI answer showed up on most brand searches by month's end
Share of tracked branded keywords where Google showed an AI Overview, whether or not the overview cited the brand.
Sources: DemandSphere, branded keywords tracked daily Sept 1 to 29, 2026, all markets and devices ("Start vs. end" averages Sept 1 to 5 and Sept 27 to 29). Chris Long (Nectiv), 100 brands, via Search Engine Roundtable, Sept 30, 2026.
It's a little early to call it permanent. Google tests things constantly, and the back-and-forth in the middle of the month looks like testing to me. Still, three straight days above 80% on a large keyword set is enough that I'd plan as if it's staying.
Why branded searches are the ones to watch
Most of what we know about AI Overviews and clicks comes from informational searches. Ahrefs' February update looked at 300,000 keywords and found position-one click-through rate was 58% lower when an AI Overview was present (Ahrefs), and that study was about informational keywords specifically. I haven't seen anyone publish a comparable number for branded queries, and Long himself only said the expansion "could reduce click-through rates from branded searches."
My guess is the hit on branded clicks will be smaller than 58%, since a lot of those people want your site in particular and will scroll past the summary to get there. I could be wrong about that. The people I'd worry about are the ones who don't scroll. Somebody searching "[your brand] pricing" or "[your brand] reviews" or "is [your brand] legit" can get a full answer from Google without ever visiting you, and that answer becomes the version of your company they bring to the sales call.
There's a measurement wrinkle too. Matt Southern at Search Engine Journal pointed out that clicking an outside link in an AI Overview counts as a click in Search Console, while links that start a new Google search, like the follow-up questions that open AI Mode, don't count as clicks or impressions at all (Search Engine Journal). So if some of your brand searchers go from the overview into AI Mode and keep asking questions there, that whole conversation never shows up in your reports. If your branded clicks dip in October and nothing else changed, I'd look here first.
Where Google gets its description of you
Google's documentation says there are "no additional requirements to appear in AI Overviews or AI Mode" (Google Search Central). That's reasonable as far as it goes, and it still leaves out which pages Google will lean on when it describes your company, which is the part I care about for brand searches.
I have some data on this, with a caveat attached. In the AI Verdict Study I ran 432 buyer conversations through ChatGPT, Claude, Gemini and Perplexity and logged every source they cited, about 32,000 in all (where AI engines go to decide about your brand). Gemini cited the brand's own website less than the other three, 11.4% of its classified citations, and in 33 of its 108 conversations it never cited the brand's own site once. It also leaned on Reddit and YouTube more than anyone else. Now, the caveat is that I tested Gemini 2.5 Flash through the API, which isn't the system that writes AI Overviews, so treat that as a hint about how Google's models read the web and not as a measurement of AI Overviews themselves.
Gemini read your own site the least
How often each engine cited the website of the brand the buyer was asking about, across 432 buyer conversations. Gemini is the Google model in the set.
Source: AI Verdict Study v1, jarredsmith.com. 32,089 citations from 432 search-on conversations, Sept 18 to 20, 2026; shares are of the 76% of citations that could be classified. Engines tested through their APIs (gpt-5, claude-haiku-4-5, gemini-2.5-flash, sonar-pro), which aren't the same systems as Google's AI Overviews.
Two other findings from that study apply here. When a brand's own site did get cited, help center and support articles made up 29% of those citations, more than the blog, pricing and comparison pages put together, and on the question about a product's limitations they were 46%. The article you wrote explaining how to work around a missing feature is a pretty likely source for whatever Google says about your weaknesses. The other finding is that the descriptions moved with whatever the engine happened to read. With search turned off, Claude told the buyer that Saucony "runs narrow," and with search on it cited Saucony's wide sizing. I'd expect Google's summary of you to be only as current as the pages it pulls on a given day.
How to check what AI Overviews say about your brand
None of this needs a tool to start. Here's what I'd do, roughly in this order.
Number one, search your own brand the way customers do. Try your name by itself, then your name plus "pricing," "reviews," "alternatives," "login," your name versus your closest competitor, and "is [brand] legit." Do it on a phone and on a laptop, signed out, and screenshot every AI Overview you get. I'd repeat it on a few different days, because the September numbers jumped around from one day to the next and a single check only tells you about that day.
Two, open every source the overview cites and sort them into yours and not yours. For your own pages, read the sentence Google pulled and ask whether you'd be comfortable hearing it read out loud to a prospect. For everything else (a Reddit thread, an old review, an affiliate list, a competitor's comparison page), write down what it says and whether it's still true.
Three, fix your own pages first, starting with the help center. The articles about what your product doesn't do should say who it's not for and what to do instead, in plain terms. Your about page, pricing page and FAQ should answer the exact questions people type after your name, in sentences that still make sense when Google lifts one out by itself.
Four, go after what's out of date elsewhere. You can't edit a Reddit thread, but you can update your review profiles, ask the publisher of a stale roundup to refresh it, and publish your own honest comparison page so the engine has a version from you to read. If an overview gets a fact flat wrong, use the feedback link on the overview and also fix the page it got the error from, since the feedback by itself won't stop Google from reading that page again.
Then last, put a note in your analytics. Mark the week of September 27 in Search Console and GA4, so that when somebody asks in a couple of months why branded clicks moved, you've got the date.
Six-question brand search check
Answer for your own brand. You'll get a score and a short list of what to do first. Nothing you click here is stored or sent anywhere.
0 of 6 answered
Recommendations draw on the AI Verdict Study (jarredsmith.com, Sept 2026) and Search Engine Journal on how Search Console counts AI Overview clicks.
If you'd like a structured way to do the first two steps, the free AI Visibility Quick Check walks through nine checks and three buyer prompts in about 20 minutes. The $199 Audit Kit goes further and includes a version of the Verdict Study buyer conversation you can run by hand on all four engines.
Where this fits with the rest of AI visibility
I've written before about why I think the share of AI answers your brand shows up in is the number to watch (The Click That Didn't Happen). Branded AI Overviews make that more concrete, since now even the people who already know your name get Google's description of you before they get yours. For more on Google specifically, my post on showing up in Google AI Mode covers how AI Mode and AI Overviews pull from different sources, and an older one on what happens when AI gets your brand wrong tells the story of a company that found out from a prospect. The longer version of why the engines describe some brands well and others badly is a big part of what I wrote Explainable about.
I'd like to hear what you see when you search your own brand. Is the overview above your site or below it, what's it citing, and does it get anything wrong? Send me a note through the contact page or tag me on LinkedIn, especially if what you're seeing doesn't match what I've described here.
Jarred Smith is the author of Explainable: Why AI Recommends Some Brands & Ignores Others, an Amazon bestseller on AEO, GEO, and SEO. He's a marketing leader with nearly 20 years of experience across healthcare, public media, retail, and environmental services. Find him at jarredsmith.com.