AI Overviews Hit 80% of Branded Searches in Major SEO Shift

The first week of October 2026 delivered two developments that, taken together, define where search optimization now stands. Google's AI Overviews suddenly appeared on the vast majority of branded searches, ending the assumption that brand queries were the last safe source of organic clicks. Days earlier, Google rewrote its official guidance on AI-generated content to demand manual human fact-checking before anything gets published. Both moves landed without warning, and neither came with an explanation.

AI Overviews swallow branded search

For most of September, DemandSphere's daily tracking showed AI Overviews on roughly a quarter to a third of the branded queries it monitors. Then the curve broke. Presence hit 69% on September 26, peaked at 90.48% on September 27, and settled above 80% through the end of the month — a tripling in under ten days, with no corresponding announcement from Google2126. A parallel test by Chris Long, co-founder of Nectiv, found AI Overviews for 93 of 100 enterprise brands he checked via SerpApi, including Reddit, Salesforce, Mailchimp and Facebook2122.

Not every dataset agrees on the starting point. Ahrefs, which confined its measurement to U.S. desktop results, had branded queries at 65.1% in August and 82.91% on September 29 — so it saw a smaller climb to roughly the same destination22. The direction, however, is not in dispute: by the end of September, an AI-generated summary showed up for most branded searches.

There is an important caveat baked into the headline number. DemandSphere counted an AI Overview whenever one appeared for a branded query, regardless of whether the brand itself was cited or described correctly inside the answer2225. Presence is not the same as control. A brand can trigger Google's AI layer and still lose the narrative to third-party pages, stale help documentation or generic category content — and most of the observed overviews appeared mid-page or lower rather than above the official site2425.

The exclusions are telling. Google's own brand name and news publishers reportedly did not show overviews for their branded searches, according to checks by Search Engine Roundtable's Barry Schwartz2224. That pattern suggests a deliberate carve-out for properties Google either owns or is politically careful with, which makes the absence of an announcement more conspicuous, not less.

The click math brands are now doing

Branded queries were supposed to be immune. A person typing a company name already knows where they want to go, and earlier Amsive research actually found branded click-through rates rose 18.68% when an AI Overview was present — the one category where the AI layer helped rather than hurt26. Nobody has yet published a click-through figure for branded queries under the new conditions, and Search Engine Land says plainly that the branded CTR impact remains unknown2124.

But the surrounding evidence points one way. When Google shows an AI summary, users click a traditional result just 8% of the time versus 15% without one, and only 1% click links inside the summary itself, according to Pew Research Center's study of 68,000 real queries1120. Ahrefs found the top-ranking page loses 58% of its average click-through rate on keywords with an Overview, worsening from a 34.5% drop measured eight months earlier1224. Seer Interactive measured a 61% organic CTR decline on informational queries with Overviews, while BrightEdge puts overall CTR on Overview-bearing searches at 83% zero-click2326.

My reading of the branded surge is that Google is now testing whether it can absorb the navigational layer of search — the part of the journey that used to end at your homepage — into its own summary. Even if mid-page placement spares most brand clicks today, the pattern fits everything else Google shipped this quarter: a sponsored shopping grid styled like free listings, AI Overview links that push users toward AI Mode rather than web pages, and an AI contribution pilot that reportedly pays some publishers under 0.1% of the ad revenue their content supports424.

Google tightens the rules on AI content

On October 1, Google updated its Search Central guidance on generative AI content to say it is "critical to manually fact-check and review all AI-generated content for accuracy and trustworthiness before publishing"2729. The new text explains why: generative models don't retrieve facts but predict likely word sequences from training data, so outputs may contain hallucinations2931.

The quiet change is the one that matters operationally. The review requirement now explicitly attaches to metadata — title elements, meta descriptions, structured data and image alt text — the four places where AI is used at the highest volume and checked the least2930. A hallucinated price or availability value in Product markup is a factual error living in code rather than on the visible page, which is precisely where editorial review tends not to reach29.

Google announced no new penalty or ranking signal with the edit, and the scaled content abuse policy is unchanged2932. It should be read as documentation, not enforcement. But the timing is hard to ignore: the September 2026 spam update began rolling out on September 24 and was still in progress as the guidance changed, and Google separately warned against fabricated author profiles in its people-first content guidance in the same week429. Whether the spam update and the new wording are connected is speculation — Google hasn't said so — but the combined message is coherent either way: mass-produced, unchecked AI output is the target, and a human review step is now the written expectation2932.

Google finally puts numbers on crawling and indexing

The same week, at Search Central Live Deep Dive Europe in Barcelona, Google analyst Gary Illyes showed internal timing data across crawling, indexing and serving — the first time Google has attached its own numbers to processes SEOs have always estimated3334.

The typical figures are faster than folklore suggests: about 20 hours to discover a new URL, roughly 24 hours to process a sitemap, about 1.5 hours for end-to-end indexing once critical processes run, and a tight 24-to-25-hour window for robots.txt updates3335. The slow tails are where the story turns. Site moves typically take one to three months, core update recovery takes three to six months — stretching to a year when a site must wait for the next core update — and several rows list "never" as the worst case3436.

The recurring word in those slowest columns is "quality." A page that sits unindexed for months is, on this evidence, more likely a page Google has judged not worth the effort than one stuck in a technical queue3334. That reframes a lot of client work: fast technical fixes don't help if Google doesn't think the page merits them, and resubmitting a sitemap changes little when sitemap processing itself runs about a day3335.

AEO graduates from buzzword to operating discipline

Around these Google announcements, the answer engine optimization industry consolidated. AEO is now its own practice with its own tooling, its own metrics and its own market reports — defined as structuring content so AI systems can extract and cite the answer rather than merely find the page1117.

The honest numbers cut both ways. AI referral traffic is still small — roughly 1% of total site traffic, about 0.16% in the UK — but it grew roughly 16-fold between 2024 and 2026, and HubSpot reported a 527% year-over-year increase in AI-referred sessions across its customer base while organic search fell 27%1320. ChatGPT dominates that flow with somewhere between 64% and 87% of AI referral share depending on the study, with Gemini a distant second131518.

Meanwhile, the counterintuitive finding from Seer's 5.47-million-query dataset is that queries without AI Overviews got more valuable: CTR on those queries rose from 2.8% to 3.8%, meaning the traffic still willing to click is higher-intent traffic20. The opportunity is bifurcating — win the citation for informational queries, convert ruthlessly on the commercial queries where the AI layer hasn't yet arrived2026.

Notably, some widely promoted AEO tactics lack evidence. Google has stated plainly that structured data is not required for AI Overviews and that no special schema exists for them, and analysts warn against treating llms.txt or Markdown formatting as automatic citation tactics128.

The week's real lesson

Strip away the individual headlines and one theme remains. Google spent early October simultaneously expanding its AI answer layer into the last defensible corner of organic search and raising the editorial bar for the content that layer feeds on. Branded search is no longer a guaranteed click path, unchecked AI content now carries written-in policy risk, and quality — not technical hygiene — determines whether Google bothers with your pages at all252934.

For SEO teams, the practical agenda follows directly: audit which branded queries now trigger overviews and what those summaries say about you2125; extend human review to metadata and schema, not just body copy2930; and build answer-structured, fact-checked content that survives extraction — because the systems now writing the first sentence of your brand story are the same ones deciding whether your pages were worth finding at all.