Google moved AI Overviews for brand searches into the knowledge panel slot: who now writes your brand’s description, and how to take it back

What changed: AI Overviews for brand searches moved to the right column

Search for a big brand on Google desktop in the United States this week and the right-hand column no longer holds the familiar knowledge panel. In its place sits a panel labelled “AI Overview”: a generated paragraph about the company, a few fact rows, a rating and a list of sources. Search Engine Roundtable reported the change on 2 November 2026, writing that Google “has moved the AI Overviews for many brand-related keywords to the right-hand column of the search results page. Previously, they appeared at the top or middle of the main search results section.” Its examples were Apple, Zillow, Coca-Cola, Google and its own site.

Google has made no announcement. PPC Land’s account of 3 November notes that no help page update or statement accompanies the report, that Google has not said whether the layout is a test or permanent, and that every published example shows results set to the United States. We checked Google’s Search Central blog and the Search Status Dashboard on 9 November: nothing has been posted about brand queries or the layout, and no ranking update is in progress.

The panel’s contents are the part worth reading closely. PPC Land’s walk-through of the Zillow example shows the generated text saying the company was “founded in 2006 and headquartered in Seattle” while the fact row in the same panel says “Founded: 2004, Seattle, WA”. Zillow was founded in 2004 and opened its site to the public in 2006. Both statements are defensible; the panel simply does not agree with itself, and no one at the brand chose either sentence.

How many brand searches show an AI Overview now?

Most of them, and the climb happened in one month. Ahrefs analysed 232.8 million US desktop search-result crawls from July to September 2026, 92.5 million of them for branded keywords. The monthly share of branded keywords with an AI Overview went from 61.3% in July to 65.1% in August and 73.3% in September, and on 29 September alone the figure was 82.91%. The gap between branded and non-branded queries narrowed from 12.2 percentage points in July to 5 in September. Ahrefs’ previous study, in November 2025, had put branded queries at 13.1%.

DemandSphere’s daily tracking tells the same story with sharper edges. Its post of 1 October 2026 counts an AI Overview as present whenever the results page returns one for a tracked branded keyword, across all markets and devices, whether or not the brand is cited inside it. The share was 26.12% on 1 September, 57.43% on 18 September, 90.48% on 27 September and 82.06% on 29 September. DemandSphere does not state how many keywords it tracks, so read the shape of the curve rather than the exact decimals.

The two studies disagree on the bare brand-name rate (Ahrefs 63 of 100 on 30 September, Slate 93 of 100 over 30 September to 1 October) because they used different brand lists, locations and rerun rules. Slate’s own finding explains much of the gap: of 44 brand searches it repeated, 16 flipped between showing and not showing an AI Overview, a 36% flip rate. One check of your brand tells you very little. A schedule of checks tells you the trend.

Knowledge panel vs AI Overview: what a brand actually loses

A knowledge panel was never fully yours either. Google’s help documentation says its information “generates automatically based on public information on the web”, that Google does not manually create or delete panels, and that descriptions “come from various data sources and can’t be edited”. But a panel can be claimed: a verified owner, manager or contributor can suggest edits through the Feedback option, Google says it reviews verified feedback “within a few days”, and accepted changes are written into the Knowledge Graph. The same page also warns that “not all knowledge panels are claimable as of now”.

An AI Overview has none of that machinery. Google’s AI features documentation is explicit: there are “no additional requirements to appear in AI Overviews or AI Mode, nor other special optimizations necessary”, no special structured data, and no AI text file or markup to add. The only controls are the ones you already have for Search: robots.txt for crawling, and nosnippet, data-nosnippet, max-snippet or noindex to limit what is shown from your pages. Those controls limit what Google takes from your site. They do nothing about what it takes from everyone else’s.

The citation data is the part that should change how a brand team spends its week. In Ahrefs’ 100-brand panel, Wikipedia was cited in 41 of the 57 AI Overviews with a source list, YouTube in 22, LinkedIn in 13, Yahoo Finance in 9. Competitors barely feature: only two of 58 answers discussed one, and only one competitor-owned site appeared across 57 source lists. The risk is less that a rival writes your description and more that a stale Wikipedia paragraph, a three-year-old product video or a forum thread does. For the limits of what nosnippet and the other controls can do about AI Overviews, we covered the mechanics last month.

Who does Google cite when someone searches your brand?

Slate’s study is the most detailed look so far at the sources behind brand AI Overviews. It ran 144 searches for 100 brand names from a logged-out US scraper on 30 September and 1 October 2026 and classified all 652 unique citations across the 93 AI Overviews it captured. The brand’s own website accounted for 29.3% of citations. News and financial media took 15.3%, video 13.7%, Wikipedia 8.7% and Google’s own Shopping and Maps results 8.4%. Reviews, Reddit and competitors made up about 5% combined. The brand’s site held the first citation in 48 of 93 answers (52%), but only 18% of second citations, and 13 brands supplied none of their own sources at all.

Slate also catalogued what went wrong. Fifteen answers led with negative news, including one that opened with a product recall. Thirteen confused the brand with another meaning of its name, which is why the study deliberately included brands called Clay, Gong, Loom, Ramp, Tide and Delta. Seven pulled sources from the wrong country, seven leaned on forums, and in one bank’s answer four of seven sources were user posts on a government grants forum. At least 22 answers quoted prices directly. None of this is a hallucination problem in the usual sense. The AI Overview is accurately summarising the web’s version of the brand, which is not the brand’s version.

What practitioners are saying about AI Overviews on brand searches

The reaction splits along a line worth noticing. Those who watched AI Overviews sit above the brand’s own result in September, as Ahrefs found in 48 of 63 cases, see the right column as an improvement: the brand’s homepage gets the top organic slot back. Those thinking about brand control see the opposite. In the Ahrefs write-up, Long’s LinkedIn post is quoted saying that “competitors/external sources can now control the narrative of your brand directly in the search results”, and Ann Smarty warns that the videos Google cites “may very well include outdated product information”. Both readings are right. The placement improved; the authorship did not change.

The commercial side is already moving. PPC Land’s second piece on 3 November relays Digiday reporting from 18 media buyers who said 10% to 30% of Google search budgets face new scrutiny as AI answers take more of the page, and notes that Google’s Performance Max help page now says campaigns “may filter out navigational and brand-specific searches in favor of dedicated brand Search campaigns”. Tinuiti’s Michelle Merklin put branded cost per click up 5% in the quarter. Separately, Search Engine Roundtable showed on 4 November that AI Mode can now carry up to four ads under a single answer. The brand search results page is being re-sold as well as rewritten.

How to control what Google’s AI Overview says about your brand

You cannot edit the panel, so you work on what feeds it. Google’s own documentation says the only requirements are that a page is indexed and eligible to show with a snippet, and that AI Overviews may use “query fan-out”, issuing several related searches to build one response. That means the overview for your brand name is assembled from the answers to questions like “what is [brand]”, “[brand] reviews”, “[brand] pricing” and “[brand] founded”. If your site does not answer those plainly, someone else’s page will. This is the same work that earns citations in ChatGPT, applied to your own name.

1. Audit the brand queries that carry revenue. Pick 20: the bare brand name, each product name, and brand plus pricing, reviews, alternatives, login, support and careers. Run them logged out, on desktop and mobile, in each market you sell in. Screenshot every AI Overview and record whether one appeared, because Slate’s 36% flip rate means a single check is noise.
2. Mark every claim. For each overview, label each sentence accurate, stale, missing or wrong. Note which page Google cited for it. In Slate’s data the brand’s site was the first citation only 52% of the time, so expect to find Wikipedia, YouTube, LinkedIn or a news story behind the sentences you dislike.
3. Fix the source, not the symptom. Google’s knowledge panel help says descriptions can only be changed by correcting the source; the same logic holds for AI Overviews. Update the Wikipedia or Wikidata entry within their rules, refresh or unlist outdated videos, correct directory and review-site profiles, and ask your PR team to get a current line into the coverage Google keeps citing.
4. Publish the answer pages Google is searching for. An About page with founding date, headquarters, leadership and what you do in the first two sentences. A pricing page that states prices. A comparison page for “[brand] vs” and “[brand] alternatives” before a competitor or affiliate owns that query. An FAQ for the questions in your support inbox. Clear headings, direct first sentences, indexable and snippet-eligible.
5. Disambiguate the name. If your brand is also a word, add Organization structured data with sameAs links to your official profiles so Google can tell the company from the dictionary entry, and make the distinguishing details obvious on every profile you own.
6. Keep the Google-owned feeds current. Google Shopping and Maps were 8.4% of Slate’s citations. Your Business Profile hours, categories and posts, and your Merchant Center product data, are brand content now, and they are the parts Google lets you edit directly.
7. Claim what is still claimable. Where a knowledge panel remains, claim it and keep its suggested edits current. Set up a Google Search profile if you are eligible, since Search Engine Roundtable found its own brand only triggered an AI Overview because it had one. Watch for the “official places” section Google said on 15 October it is testing, and make sure the page it should point to exists and is unambiguous.
8. Report, then re-test. Use the thumbs-down feedback on an inaccurate overview with a specific correction and the source URL, then re-run the query set in two to four weeks after recrawl. Google says recrawling “can take anywhere from several days to several months”.

How to measure it: what Search Console shows, and what it does not

Search Console will not tell you whether your brand search shows an AI Overview or what it says. Google’s documentation states that appearances in AI features are reported inside the Performance report’s “Web” search type, blended with everything else, and on 15 October Google told Search Engine Roundtable only that it would “introduce additional metrics over time”. The generative AI report that reached all sites at the end of August carries impressions but not clicks or queries, according to PPC Land. So the measurement has to be built by you.

• A brand query panel, run on a schedule. The same 20 queries, same markets, logged out, weekly. Record presence, position (right column, top or mid-page), the first three citations and any claim you marked wrong. Ahrefs’ Site Explorer branded-intent filter with the AI Overviews SERP feature, or Brand Radar’s AI responses report, automates the presence check; the claim review still needs a person.
• Brand click-through in Search Console. Filter the Performance report to your brand queries and watch CTR and position by device from late October onwards. If the right-column layout has reached your market, your homepage should regain position one while CTR tells you whether the panel is absorbing the click.
• Homepage and About page landing traffic in GA4, segmented by brand versus non-brand source where you can, so a change in panel content shows up as a change in the questions people arrive with.
• The same panel on ChatGPT, AI Mode and Perplexity. The brand description problem is not Google-only, and OpenAI’s unannounced business tools suggest first-party mention and click data for ChatGPT may be coming. Until then, track it yourself.

We fold all of this into one monthly view, with brand AI Overview presence and accuracy alongside Search Console and GA4, as part of reporting. The point is not the dashboard. It is that the brand team sees the sentence Google wrote about them before a customer does.

What this means for brand search in 2026

For a decade the brand search results page was the one Google result a company could reasonably say it owned: homepage at the top, sitelinks beneath, a knowledge panel on the right built from data it could at least argue with. In six weeks, between Chris Long’s 30 September test and the 2 November layout change, the right column became a generated summary that the brand cannot claim and that, in the only detailed sample so far, draws 71% of its evidence from pages the brand does not control. Google’s placement decision made the main results cleaner. It did not hand back the authorship.

The practical conclusion is the one Ahrefs reached about AI Overviews generally: expect them on branded queries for the foreseeable future. Treat your brand’s AI Overview as a page you publish indirectly. Decide what its first sentence should say, then make sure that sentence exists, dated and unambiguous, on every page Google is likely to cite. The brands that do this now will find the summary agreeing with them. The rest will find out what the web thinks of them one query at a time, and so will their customers. Our SEO work increasingly starts here, with the brand name, before any other keyword.