Most Categories Still Have No AI Brand Leader: Inside Semrush's 126-Million-Prompt Study
Semrush studied 126M AI prompts and found 85% of categories have no clear brand leader in ChatGPT. The mention-citation gap is why.
Semrush analyzed 126 million US AI search prompts and found that only 15.2% of buying categories have a clear brand leader in ChatGPT. Across 1,094 categories, more than half had no consistent leader at all. For CMOs, the takeaway is blunt: AI category ownership is still up for grabs — but the brands getting cited are often not the brands getting named.
Key takeaways
- Semrush's 2026 AI Visibility Index examined 126 million US prompts across ChatGPT, Gemini, Google AI Mode, and AI Overviews. Only 15.2% of 1,094 categories had a clear brand leader.
- In the highest-demand categories — which account for 98% of AI search volume — just 11.3% had a dominant brand. The most valuable real estate is the most contested.
- Mentions and citations are different currencies. Only 21% of the most-cited domains in a category were also the most-mentioned brand.
- Branded search volume was the only traditional SEO metric that tracked category ownership. Organic traffic and authority scores barely correlated.
- Winning is a topic-level game played across an entire buyer journey, not a single-prompt ranking.
The land grab is real: 85% of categories have no AI leader
To qualify as a category owner in the Semrush study, a brand had to appear in at least four of five common buyer prompts — definitions, comparisons, alternatives, use cases, and purchase decisions — and lead its nearest competitor by at least five percentage points. Only 166 of 1,094 categories cleared that bar, as MarTech reported from the analysis.
Another 31.2% had an emerging leader with no decisive edge. The remaining 53.7% were open contests. For an incumbent, that is a warning; for a challenger, it is an opening. Established organic authority does not automatically transfer into AI answers, which means the field resets faster than most category leaders assume.
Appearing once is not owning the category
The most important structural finding is that AI visibility shifts as buyers ask different questions inside the same topic. A brand can win the "best X" comparison and vanish on "X alternatives." That is why Semrush frames category ownership as performance across a full buyer journey rather than a rank on one prompt.
The pattern intensifies where it matters most. Among the half of categories generating the most AI search activity, only 11.3% had a clear owner, versus 19% in less competitive topics. Because those high-demand categories represent 98% of measured AI search volume, the overwhelming majority of real buyer activity happens where no single brand consistently appears.
The mention–citation gap most brands measure wrong
Semrush treats two metrics as structurally separate: a mention, when an AI system names a brand inside its answer, and a citation, when the system quotes that brand's site as a source. A brand can be mentioned constantly and cited almost nowhere — or cited everywhere while rarely being the subject of the conversation.
The overlap is thinner than most dashboards imply. Only 21% of the most-cited domains in a category were also the most-mentioned brand. Across platforms, ppc.land's breakdown of the index shows mention-source overlap running as high as 64% on Google AI Overviews and as low as 30% on Gemini. This is precisely why founder Jaxon Parrott argues that share of voice is the wrong metric and share of citation is the right one — the two move independently, and optimizing the wrong one wastes budget.
The Universal 36 and the wide-open middle
Semrush identified a "Universal 36" — the only brands that appeared in the top-100 most-mentioned list on every platform, every month of the study window. The list is dominated by default destinations: YouTube, Google, Reddit, Amazon, Apple, Netflix, and a handful of category giants. Mention volume falls off a cliff even inside that group, from YouTube's 47.9 million mentions to Nintendo's 1.5 million.
That concentration at the very top is the exception, not the rule. Below the Universal 36, the field fragments fast — which is the same message as the 85% of unowned categories. For a mid-market brand, competing with YouTube for universal presence is pointless; owning a specific, high-intent category is achievable.
Branded search, not organic traffic, predicts ownership
Semrush compared category ownership against conventional SEO signals. Branded search volume was the only metric that consistently aligned with owning a category. Organic traffic and Authority Score showed little relationship to whether ChatGPT reliably associated a brand with a topic.
This reframes the work. AI engines appear to reward brands that people already search for by name — a demand signal — over brands that merely rank. Building that named demand is closer to public relations and citation architecture than to traditional link-building, and it explains why so many high-traffic sites remain invisible in AI answers for their own categories.
How the engines diverge: citation density is uneven
The four platforms do not cite alike. Per ppc.land, ChatGPT averaged 15.4 sources per response — the highest — leaning on Reddit and Wikipedia. Google AI Mode followed at 11.4, AI Overviews at 9.2, and Gemini cited just 3.3 sources per response, the most selective by a wide margin. Independent coverage of the index reached the same read on which brands actually get recommended: citation behavior varies enough by engine that a single-platform strategy leaves visibility on the table.
| AI engine | Avg sources cited per answer | Avg brands mentioned per answer |
|---|---|---|
| ChatGPT | 15.4 | 3.3 |
| Google AI Mode | 11.4 | 5.2 |
| Google AI Overviews | 9.2 | 3.7 |
| Gemini | 3.3 | 4.7 |
Source: Semrush AI Visibility Index 2026, via ppc.land.
The "Source Surplus" brands illustrate the citation economy directly: Medical News Today, Wikipedia, Healthline, and WebMD are cited far more than they are mentioned, functioning as infrastructure AI systems use to answer questions about other brands. The lesson for CMOs is to know which game a given engine rewards — being the answer, or being the source behind it.
The Machine Relations lens on an unsettled category
The Semrush data is a snapshot of what practitioners call Machine Relations — the discipline, coined by Jaxon Parrott, of making a brand legible to the machines that now mediate discovery. Independent research on source selection, including how ChatGPT actually chooses which sources to cite, points the same direction: citations flow to trusted third-party surfaces, not to brand-owned pages by default.
AuthorityTech's analysis of the same category-ownership gap reaches a compatible conclusion: AI search is additive, and the brands closing the gap are the ones building earned, corroborated authority across sources engines already cite — not the ones publishing more owned content.
The CMO playbook for a contested category
Three moves follow from the data. First, measure at the topic level across the full set of buyer prompts, not one flagship query — ownership is a journey, not a rank. Second, separate mentions from citations in reporting, because a brand can be winning one and losing the other. Third, invest in named demand and third-party corroboration, since branded search — not organic traffic — is what tracked ownership.
The window is the point. With 85% of categories unowned, most brands are one focused visibility program away from leading a topic that no competitor has locked down. Teams can start by auditing where they stand today with a free AI visibility audit.
Frequently asked questions
How many categories have a clear brand leader in ChatGPT? Semrush found only 15.2% of 1,094 studied categories had a clear leader — a brand appearing in at least four of five buyer prompts and leading by five-plus points. Another 31.2% had emerging leaders; 53.7% had none.
What is the difference between a mention and a citation in AI search? A mention is when an AI names a brand inside its answer. A citation is when it quotes that brand's website as a source. Semrush found only 21% of the most-cited domains were also the most-mentioned brand, so the two must be measured separately.
Which AI engine cites the most sources per answer? ChatGPT led at 15.4 sources per response, followed by Google AI Mode (11.4) and AI Overviews (9.2). Gemini was the most selective at 3.3 sources per response.
Does traditional SEO authority predict AI category ownership? Only partly. Branded search volume aligned with ownership, but organic traffic and Authority Score showed little relationship — suggesting named demand and earned citations matter more than rankings for AI visibility.