Your Brand Search Now Opens With an AI Overview You Didn't Write

The search for your own company's name used to be the one Google result you could ignore. You ranked first without trying, and everything on the first screen led back to your own pages. That stopped being true over the summer. AI Overviews, the AI-written summary Google places above the regular results, now appear on most searches that include a brand name, and when one shows up on a big brand's name, it usually sits above the brand's own site (Ahrefs).

That makes your own name the first AI answer worth checking, ahead of the "best tool for X" questions where most AI-visibility work goes. Buyers who met you in an AI answer often go there to look you up, and what is most likely to be wrong in it is old information rather than a competitor. Most of what it cites, you can fix or ask to have fixed.

Buyers who find you in ChatGPT look you up on Google

Most people who use ChatGPT still use Google: Similarweb puts the share at 95 percent, a figure that held from September 2025 through May 2026 (Search Engine Journal). SEJ describes the path: someone asks ChatGPT for a recommendation, then searches Google for the brand it named, and the visit lands in Search Console as ordinary organic traffic. Similarweb lists a rise in branded searches among the metrics that should move when AI is feeding you demand.

We made the same argument in August: more people searching your name is the click an AI answer swallowed coming back one step later. As a count of interest, that still holds. What changed is what those searchers read when they arrive.

Ahrefs checked 232.8 million US desktop results pages from July to September. The share of branded searches showing an AI Overview averaged 61.3 percent in July and reached 82.91 percent on 29 September. A separate weekly check on the bare names of 100 brands from Interbrand's Best Global Brands ranking found Overviews on 5 of 98 checked on 1 July and on 63 of 100 by 30 September. In 48 of those 63, the Overview held position one, above the brand's own site (Ahrefs).

So a buyer who found you through one AI answer is now likely to read a second one before reaching your site, and Google writes it from whatever it found about you. Ahrefs thinks lost clicks are probably not the main concern on branded searches. What the summary says about you matters more.

The bigger risk in that summary is stale information

The summary draws on a short list of other people's pages, and that list is less hostile than the first reactions suggested. Chris Long, whose check of 100 enterprise brand names prompted the Ahrefs study, warned that "competitors/external sources can now control the narrative of your brand directly in the search results." Ahrefs' own numbers disagree about the competitors. Of the 58 branded Overviews whose full text it could read, two discussed a rival: Pepsi's mentioned Coca-Cola and the cola wars, and Intel's named TSMC and NVIDIA. Of the 57 lists of source links the Overviews showed, one pointed to a verified competitor's site (Ahrefs).

The sources that deserve attention are the rest: Wikipedia appeared in 41 of those 57 lists, YouTube in 22 and LinkedIn in 13. Ann Smarty noted that the cited videos were not from the brands' official accounts and "may very well include outdated product information" (Ahrefs). That is the failure to check for. An AI summary is built from what has been written about you, and a lot of that was written about the company you used to be: the pricing in a two-year-old review video, the product name you retired. Practitioner guidance in the marketingskills library adds that a brand's place in AI answers moves slowly, even after its own pages change (marketingskills).

Keeping those facts current is also the part of AI search that job ads almost never ask for. In the 4 October crawl of Promptafire's jobs index, 49 distinct marketing roles at 34 companies mention AI search work, mostly under the acronyms AEO and GEO, both shorthand for optimizing to show up in AI answers. Three of them frame the work as making sure AI answers describe the company accurately, and one of those is a recruiting role worried about what candidates hear. The rest describe visibility: getting found, cited and recommended.

Correct it at the pages it cites

The fix starts with reading the list of cited pages, which takes minutes.

The check may well come back clean: Ahrefs calls Wikipedia "a fairly neutral source of information for most brands", and competitors barely featured (Ahrefs). If yours is accurate, keep the date and the cited pages as next month's baseline. If it isn't, give more weight to sources that keep coming back, since Ahrefs warns that a source cited once could be a fluke (Ahrefs). Run monthly, and across ChatGPT and Perplexity too, the same check becomes a log that traces each wrong fact back to the page it came from.

Then work through the sources, starting with the ones you control. If one of your own cited pages still shows an old price or a dropped feature, update it, because AI answers repeat those mistakes (Ahrefs). Make your LinkedIn company page, review-site profiles and directory listings describe you with the same category, segment and differentiators as your About page, which one practitioner reference puts at an afternoon of edits (marketingskills).

YouTube is less out of reach than Ahrefs suggests. It calls YouTube and LinkedIn much harder for brands to influence than their own sites, which is fair for a reviewer's video and much less so for your own channel, because an AI reads the text around a video (title, captions, description) rather than the footage (marketingskills). The same reference records an eight-day-old video with under 100 views as the first source cited in the Overview for "best email marketing software", so a current official video can give the summary something newer to cite.

For other people's articles and videos, send the author the correct fact and a source they can check, such as your pricing page or changelog, and ask them to fix the fact rather than their opinion. Ahrefs tried this with more than 20 publishers; a few agreed, and the corrections later showed up in AI answers (Ahrefs).

Wikipedia has rules of its own. Its conflict-of-interest guideline strongly discourages people connected to a company from editing its article directly. Disclose who you are and propose the correction on the article's talk page, its discussion page, with a published source behind it. Volunteers review the request and can decline it (Wikipedia).

What a study of global brands can't tell you about yours

All of this rests on one vendor's measurements of some of the world's biggest brands, which limits how far it travels. The citation counts come from 57 Overviews on US desktop, gathered by a company that sells Overview tracking, and Ahrefs itself says Google keeps changing when Overviews appear, with no guarantee the pattern holds. Practitioner guidance warns against borrowing anyone else's source list for AI answers generally, because the mix differs by category and shifts over time (marketingskills). The brands in the Ahrefs panel were big enough that Wikipedia led their citations. If your company has no Wikipedia article, something else is filling that slot, and the only source list worth acting on is the one under your own Overview.

Sources

• AI Overviews Currently Appear for 83% of Branded Searches, Ahrefs, 2 October 2026
• AI Visibility Data Paralysis: How to Get Unstuck, Ahrefs, 30 September 2026
• People Aren't Leaving Google For AI. They're Using Both, Search Engine Journal, 4 September 2026
• Positioning and Consensus in AI Answers, marketingskills ai-seo reference, added 2 October 2026
• YouTube Videos That Get Cited by AI, marketingskills ai-seo reference
• Wikipedia: Conflict of interest, Wikipedia guideline
• Promptafire jobs index, marketing postings in the 4 October 2026 crawl