For twenty years, a branded search was the one query a marketer could treat as predictable: someone types your company’s name, and they get you. Now a generated summary may sit above the results, shaping what a searcher sees first.
That changes the question of AI search visibility. It is no longer enough to ask whether a feature appeared. We also need to know whether the brand appeared in the answer, and whether the answer represented it accurately.
AI Overviews appear on 80%+ of branded queries, but the baselines differ
The clearest agreement in the available data is the destination: by September 29, 2026, both DemandSphere and Ahrefs reported AI Overviews on more than 80% of branded search results in their respective datasets.
The starting points are materially different. DemandSphere reported 26.12% on September 1. Ahrefs reported 65.1% in August. That is not evidence of a scandal or proof that one dataset is wrong. It is the honest state of measurement: the samples, dates, and coverage differ.
| Dataset | Starting figure | Late-September figure | Coverage and limits |
|---|---|---|---|
| DemandSphere, reported by Semrush | 26.12% on Sept. 1 | 82.06% on Sept. 29; peak of 90.48% on Sept. 27 | All markets and devices. DemandSphere has not disclosed the number of tracked keywords. |
| Ahrefs, reported by Semrush | 65.1% in August | 82.91% on Sept. 29 | U.S. desktop results only. |
These figures come from DemandSphere’s daily tracking and Ahrefs’ branded-search analysis. The similar late-September readings are notable, but the different coverage means the series should not be treated as a like-for-like comparison.
Other checks point in the same direction. Nectiv co-founder Chris Long reported AI Overviews for 93 of 100 enterprise brands he tested, as Search Engine Land reported. Search Engine Roundtable’s Barry Schwartz also saw the pattern across several large brands, with exceptions for Google’s own name and news publications.
Google has not announced a change to branded queries, and it has not said whether the observed shift is finished. The rate could still move.

The change did not happen in a straight line
DemandSphere’s daily series shows a sharp jump to 57.43% on September 18, followed by a fall to 31.15% the next day. The rate then moved between roughly 30% and 57% for another week, before reaching 69.21% on September 26 and staying high through the final days in its data. DemandSphere reported these as observed daily readings, not smoothing artifacts.
A single check could therefore have produced a very different picture depending on the day. Between September 1 and September 29, the reported rate moved by roughly 56 percentage points. This is a fast-moving search surface, not a stable benchmark. A marketer who checked once in September got a number that described that snapshot, not the whole month.
The central limitation: feature presence is not brand representation
DemandSphere measured whether an AI Overview appeared for a branded query. It did not measure whether the searched-for brand appeared or was cited inside the generated answer. Semrush makes this distinction directly, and it is essential.
An overview’s presence is a feature-availability metric. It does not tell you:
- Whether the answer names your brand, or how prominently.
- How the answer describes your products, services, or positioning.
- Which pages or sources support that description.
- Whether the answer introduces competitors or alternatives.
Those are different questions. The second set carries the commercial consequences.
If an AI Overview appears when someone searches your brand, does it name you? Describe you correctly? Cite your website, or a reseller, review site, or competitor? Introduce other options? None of the datasets above answers those questions. A marketer reading the headlines cannot tell which situation applies to their own brand.

A second experimental surface: AI Mode website panels
In an October 7 report, Search Engine Watch’s Sachin Patel described experimental website panels appearing on some branded searches in Google AI Mode, with direct links to sites. The feature is experimental, appears only on some queries, and reporting on it is still forming; its behavior may change.
The reported panels are generated automatically: publishers cannot request one or choose which links appear. The practical response is familiar site and entity hygiene, not a guaranteed recipe for inclusion. Keep the homepage title and meta description clear and descriptive, use a properly sized square favicon, maintain accurate Organization and WebSite schema on the homepage, and make site navigation logical with a shallow hierarchy and clear labels.
Together, these developments point in the same direction: more of the branded search experience is being assembled by the system rather than presented by the brand alone.
What to check now: a practical GSC comparison
Semrush recommends comparing branded-query click-through rate (CTR) in Google Search Console before and after AI Overviews began appearing. That is good advice. Use CTR rather than raw clicks because branded demand can rise and fall with your own marketing activity.
| GSC comparison | Date range | What to do |
|---|---|---|
| Before | Sept. 8–14, 2026 | Filter the Performance report to branded queries; record CTR. |
| After | Sept. 29–Oct. 5, 2026 | Apply the same filter and compare CTR. |
If you know when an overview first appeared for your terms, use a full week before and a full week after instead. Then search your brand name while signed out and read the overview as a customer might: note what it says, which pages it cites, and whether those pages are current.
This comparison can help show whether branded search performance changed. It cannot, by itself, diagnose the content or accuracy of the AI answer.
AI search visibility needs a representation measure
The channel question is settled: branded search is now mediated by generated surfaces. The open question is representation. A rank tracker, a feature-availability count, or a CTR comparison cannot answer it alone.
A useful measurement program needs to track whether the brand appears in the overview, its prominence and sentiment, factual accuracy, cited source pages, and competitor presence. It should do so over time and across platforms. When the observed rate changes by roughly 56 percentage points in four weeks, and Google has made no announcement, a one-time check is not enough.
This is also a brand knowledge governance issue, not simply an SEO task. Marketing, legal, product, and content operations all have a role in keeping public facts consistent and correcting outdated or misleading descriptions.
Where CiteMetrix fits
CiteMetrix includes dedicated Google AI Overview tracking, distinct from traditional search and conversational platforms. Its analysis tools help examine Brand Perception and Model Sentiment, identify the exact Source Pages cited, including third-party pages, and compare Competitor Share of Voice across named competitors.
Accuracy & Hallucination Detection is brand-specific monitoring, not a general enterprise hallucination solution. It identifies five error types (Incorrect Fact, Outdated Info, Fabricated Claim, Misleading Context, and Critical Omission) at Minor, Moderate, or Critical severity, with a contextual fix for each. An overview describing a discontinued product or old positioning is an Outdated Info finding; omitting a material fact about the business is a Critical Omission. Included from Starter, the system tracks unverified facts separately from correct answers to help avoid false alarms.
CiteMetrix’s Source of Truth lets a team verify brand facts once, then generate llms.txt and schema.org markup deterministically from those facts, not written by AI. Only an Owner or Admin can publish verified facts; internal and live drift detection flag changes. This supports accurate brand representation because the entity signals used across search surfaces depend on consistent facts.
The platform’s nine core features are AI Coaching, Report Analysis, Source of Truth, Hallucination Detection, Mobile Dashboard, CiteMetrix Analyst, Content Engine, Content Change Tracker, and Site Crawl. Content Change Tracker keeps structured page snapshots so teams can connect a shift in AI descriptions to a page edit. Report Analysis offers 10+ templates, PDF and PowerPoint export, shareable links, and scheduled delivery for leadership reporting.
ModelScore’s four components make the changing inputs visible: Mention Score (45%; nine engines, weighted by sentiment and prominence), Brand Demand (20%; branded searches from Google Search Console), Authority Transfer (20%; AI-referred traffic from GA4 or Adobe Analytics), and Technical Readiness (15%; including schema, crawlability, and llms.txt). Brand Demand deserves attention: branded searches are precisely the input now mediated by these surfaces. Technical Readiness includes structured data, one of the reported signals behind the experimental AI Mode panels.
CiteMetrix reports per-platform results across nine engines: ChatGPT, Perplexity, Claude, Google AI Overview, Google Gemini, xAI Grok, Mistral, DeepSeek, and Copilot. All nine platforms are included on every plan, starting at $79/month with no add-ons: Everything Included, Nothing Extra.
The workflow is monitor → detect → diagnose → fix → verify. One helpful next read: Being Mentioned Isn’t the Same as Being Cited or Recommended.
Measure what the answer says, not just whether it appears
We know AI Overviews appeared on a large share of branded queries in late September. We do not yet know, from these datasets, how often the brand was included, accurately represented, or displaced by alternatives. That is the measurement gap with business meaning.
At CiteMetrix, we believe visibility should be verifiable: check the answer, inspect its sources, track changes over time, and tie findings back to facts and pages you can confirm. See what AI says about your brand → citemetrix.com
LinkedIn version
Your branded query used to be the predictable one. That assumption needs a second look.
DemandSphere measured AI Overviews on 26.12% of tracked branded queries on September 1 and 82.06% on September 29. Ahrefs also measured more than 80% on September 29, but its August starting point was 65.1%. DemandSphere covers all markets and devices, with an undisclosed keyword count; Ahrefs covers U.S. desktop results.
The limitation matters: these datasets measured whether an overview appeared, not whether the brand appeared in it, or whether it was described accurately.
Semrush recommends comparing branded-query CTR in Google Search Console for a full week before and after the change. Use CTR, not raw clicks, to account for shifting branded demand.
See what AI says about your brand → citemetrix.com
#AISearch #SEO #BrandVisibility

