Ahrefs tracked 100 of the world's biggest brands and watched AI Overviews jump from 5 branded results in early July to 63 by September 30, 2026. A second study found the same spike. The answer sitting on your own brand name is now built from sources you don't own.
2026-10-10 · by Moe Ameen
Two independent studies published in early October 2026 found that Google's AI Overviews have spread across most brand-name searches — including searches for a company's own exact name. Ahrefs published its analysis on October 2, 2026. It checked Google's U.S. results for the exact names of 100 of the world's biggest brands, drawn from Interbrand's Best Global Brands 2025 list, every week from July through September. On July 1 only 5 of 98 usable brand-name queries showed an AI Overview; by September 30 that had risen to 63 of 100.
Ahrefs' headline figure comes from a far larger dataset: 232.8 million U.S. desktop search crawls between July and September, of which 92.5 million were branded-keyword crawls. The share of branded queries triggering an AI Overview climbed from 61.3% in July to 65.1% in August to 73.3% in September, and on September 29 reached 82.91% — the "83% of branded searches" in the report's title. Non-branded keywords sat slightly higher, around 78.3% in September. A second firm, DemandSphere, published its own tracking on October 1 and found the same late-September spike: branded queries with an AI Overview went from 26.12% on September 1 to 90.48% on September 27, settling near 82% on September 29. DemandSphere's day-by-day data shows that share staying below 60% until around September 26 before jumping sharply — Ahrefs only reports monthly averages, so it can't confirm the exact date, but both data sets point to the same late-September acceleration.
The placement matters as much as the frequency. Ahrefs found that about three in four branded AI Overviews appeared in position one — above the brand's own website in the results. In its 100-brand panel, 48 of the 63 AI Overviews sat in the top slot, with a median position of one. And the sources those answers cited were overwhelmingly third parties the brand does not control: across the branded AI Overviews with visible source lists, the most-cited external domains were Wikipedia (about 72%), YouTube (about 39%), and LinkedIn (about 23%).
Both firms were careful about the limits. Neither measured actual click loss, and neither fully documented how it classified a "branded" keyword, so the exact percentages differ between samples. Ahrefs explicitly flagged that the current situation may not persist — Google adjusts AI Overview triggering frequently. Treat the direction as the durable finding and the precise percentages as a late-September 2026 snapshot.
This finding reframes the job. For years the brand-name search was the one query you reliably owned — your own site ranked first for your own name. Now an AI Overview often sits above it, and the studies show it is stitched together from third-party sources, with YouTube and LinkedIn among the top three. You cannot edit Google's answer. What you can do is make sure the platforms that answer cites are full of current, on-brand content about you — and that is a publishing-cadence problem, which is exactly what [Kompozy](/) is built to solve.
Here is the concrete move. Take the questions people actually search alongside your brand name and turn each into a set of content that lives on the surfaces AI Overviews cite: a captioned [Persona Short](/glossary/persona-shorts) and a longer Persona HeyGen cut for YouTube, a text post and a document-style carousel for LinkedIn, plus a blog article on your own domain — all generated from one brief and kept in your voice by the [Persona Brief](/glossary/persona-brief). The persona video formats render in the studio off a script you approve first; the text post, carousel, and blog article can run on [Autopilot](/glossary/autopilot)'s recurring cadence behind its quality gates instead. Either way, everything schedules across the eight social platforms plus blog and email, so your brand's footprint on YouTube and LinkedIn stays fresh instead of a single stale upload. Kompozy does not change what Google's AI says — no tool honestly can — but it is the fastest way to flood the sources that AI reads about you with content you actually control. For the playbook, see [how AI Overviews pull from social media sources](/guides/google-ai-overviews-social-media-sources) and [how to get your content cited in a Google AI Overview](/how-to/get-your-content-cited-in-google-ai-overviews).
Yes. Ahrefs' October 2026 study found AI Overviews on 82.91% of U.S. desktop branded search results by September 29, 2026, and its 100-brand panel rose from 5 branded AI Overviews on July 1 to 63 by September 30. A second firm, DemandSphere, found the same late-September spike. The percentages are a snapshot and Google can change triggering, but the direction is consistent across both studies.
Across branded AI Overviews with visible source lists, the most-cited external domains were Wikipedia (about 72%), YouTube (about 39%), and LinkedIn (about 23%). In other words the AI summary of a brand is often built from third-party platforms rather than the brand's own site, which is why a thin presence on those surfaces lets others narrate your brand.
You cannot edit Google's answer directly, but AI Overviews cite sources — so the practical lever is owning more of those sources. Build a current, consistent, on-brand footprint on the platforms these answers reach for, especially YouTube and LinkedIn. A content engine like Kompozy generates captioned avatar video, carousels, posts, and a blog article from one brief in your voice, keeps the text and carousel formats running on a recurring autopilot cadence, and schedules everything across those platforms — so what the AI reads about you stays fresh and on-message.