As generative artificial intelligence increasingly reshapes the digital marketing landscape, a groundbreaking study has revealed that the battle for brand visibility inside Large Language Models (LLMs) remains wide open.
According to a comprehensive data analysis conducted by search intelligence platform Semrush in partnership with Kevin Indig, founder of Growth Memo, only 15.2% of ChatGPT topic categories have a clear, dominant brand owner across related buyer questions.
This finding means that approximately 85% of commercial product and service categories in ChatGPT are currently up for grabs, with no single brand establishing a consistent foothold. For search engine optimization (SEO) professionals and digital marketers, the study signals a profound paradigm shift: appearing in a single AI prompt does not guarantee a brand will consistently show up across related questions in the same topic category.
1. Main Facts: The Fragmented State of AI Brand Ownership
The joint study by Semrush and Kevin Indig introduces a critical distinction between "prompt-level visibility" and "topic-level authority" within conversational AI environments. While traditional search engines like Google present a list of links that users can browse, ChatGPT synthesizes information to deliver direct answers. This conversational format makes brand mentions within the generated text highly valuable.
However, the research highlights that true brand dominance within these AI-generated responses is exceptionally rare.
Defining "Category Ownership" in the Age of AI
To quantify brand dominance, Semrush established a rigorous mathematical definition for what constitutes a "category owner." To earn this title, a brand had to meet the following criteria:
- High Share of Mentions: The brand must secure the highest share of verbal mentions within ChatGPT’s response text.
- Consistency: The brand must appear in at least four out of five related prompts within a specific topic category.
- Clear Margin of Victory: The brand must lead its closest runner-up by a margin of at least 5 percentage points.
Using this strict benchmark, only 15.2% of the analyzed categories had a clear owner. The remaining 84.8% of categories showed high volatility, with different brands appearing depending on how a question was phrased, or no single brand emerging as a dominant authority.
ChatGPT Category Ownership Distribution:
┌─────────────────────────────────────────┐
│ ░░░░░ Clear Brand Owner (15.2%) │
├─────────────────────────────────────────┤
│ ████████████████████████ (84.8%) │
│ No Dominant Brand / High Volatility │
└─────────────────────────────────────────┘The study also revealed a counterintuitive trend regarding search volume and brand dominance: the more popular a topic, the harder it is for a single brand to own it. Semrush divided the analyzed categories into two halves based on AI search demand:
- High-Demand Topics (Top Half): Only 11.3% of these highly competitive topics had a clear brand owner.
- Lower-Demand Topics (Bottom Half): In contrast, 19% of lower-demand, niche categories featured a dominant brand.
This discrepancy suggests that high-volume commercial topics are subject to intense informational noise and diverse training data, making it difficult for ChatGPT’s algorithms to settle on a single, definitive brand recommendation.
2. Chronology & Methodology: How the Research Was Conducted
To understand how ChatGPT establishes topic authority over time, Semrush and Kevin Indig conducted a longitudinal study tracking data over a six-month period.
CHRONOLOGY OF THE STUDY
│
├── January 2026: Commencement of data collection via Semrush AI Visibility Toolkit.
│ └── Initial mapping of 1,094 U.S. commercial categories.
│
├── February – May 2026: Ongoing monthly monitoring of 220,000+ URLs and 50,000+ brands.
│ └── Tracking of month-over-month shifts in brand mentions and citations.
│
└── June 2026: Conclusion of the monitoring window.
└── Final analysis of 600,000+ citations and 220,000+ unique domains.The Dataset by the Numbers
The scale of the study represents one of the most comprehensive investigations into LLM-based search behavior to date:
- Timeframe: January through June 2026
- Geography: United States
- Tools Used: Semrush AI Visibility Toolkit
- Brands Tracked: More than 50,000
- Domains Analyzed: 220,000
- Citations Analyzed: 600,000
- URLs Evaluated: 220,000
- Topic Categories: 1,094 unique U.S. market categories
The Five-Prompt Funnel Model
To simulate real-world buyer journeys, the researchers didn’t rely on single keyword searches. Instead, they designed a matrix of five distinct prompts for each of the 1,094 categories. These prompts mapped directly to common stages of a consumer’s decision-making process:
- Definitions: Explanatory queries about what a product class or service is (e.g., "What is cloud-native CRM software?").
- Comparisons: Head-to-head evaluations of solutions (e.g., "Compare CRM X and CRM Y for enterprise sales").
- Alternatives: Queries seeking options (e.g., "What are the best alternatives to HubSpot for small businesses?").
- Use Cases: Practical application questions (e.g., "How is project management software used in agile software development?").
- Buying Decisions: High-intent purchasing questions (e.g., "What is the best accounting tool for freelancers on a budget?").
By tracking brand performance across all five of these related queries, the study was able to evaluate whether ChatGPT views a brand as a holistic category authority or merely a transactional recommendation for a single, isolated query.
3. Supporting Data: The Disconnect Between SEO and AI Visibility
One of the most striking revelations of the study is that traditional digital marketing metrics have surprisingly little influence on how ChatGPT selects and presents brands to users.
The Failure of Traditional SEO Metrics
For decades, SEO professionals have relied on metrics like Domain Authority, backlink profiles, PageRank, and organic search volume to predict search engine rankings. However, the Semrush study found that broad SEO metrics had limited predictive value in determining who owns a ChatGPT topic.
A brand with a massive backlink profile and millions of organic visitors on Google might find itself completely omitted from ChatGPT’s conversational answers in favor of a smaller, more specialized competitor. This suggests that LLM retrieval mechanisms—often powered by vector databases, semantic search, and semantic similarity—weigh contextual relevance and direct answers far more heavily than traditional domain-level authority signals.
Citations vs. Mentions: The Invisible Gap
The study also identified a critical disconnect between the links ChatGPT displays as sources and the brands it actually recommends in its text.
┌────────────────────────────────────────────────────────┐
│ CHATGPT ANSWER TEXT (Mentions = Core Brand Authority) │
│ "We recommend Brand A for enterprise security..." │
└────────────────────────────────────────────────────────┘
│
▼ (Often Disconnected)
┌────────────────────────────────────────────────────────┐
│ FOOTNOTE / SOURCE LINKS (Citations = Referral Traffic)│
│ 1. [Link to independent blog post] │
│ 2. [Link to comparison directory] │
└────────────────────────────────────────────────────────┘Semrush measured category ownership based on brand mentions within the actual body text of ChatGPT’s answers, rather than the hyperlinked citations displayed at the bottom of the response. The data showed that the most-cited websites were rarely the most-mentioned brands.
This means that while a publisher’s blog post might be cited as a source link (generating referral traffic), the LLM’s conversational recommendation often favored a different, highly-visible brand discussed within that source material. For brand managers, this highlights the necessity of being talked about by authoritative sources, rather than simply trying to get their own domain ranked or cited directly.
Stability of the Crown: The Rich Get Richer
For the few brands that did manage to secure clear ownership of a topic, the news was highly encouraging. The study found that clear category owners stayed in first place in 90.4% of month-over-month comparisons. Once an LLM’s weights and retrieval-augmented generation (RAG) sources align to favor a dominant brand, that position is incredibly stable.
Conversely, weaker or unsettled categories experienced extreme volatility:
- In emerging or unsettled categories, the top-performing brand switched places in 1,950 out of 5,470 monthly comparisons.
- When a top brand held a narrow lead and eventually lost its first-place position, its average lead before dropping was just 1.3 percentage points.
- In contrast, when a dominant brand successfully maintained its top position month-over-month, its average lead was 2.9 percentage points—more than twice as large.
4. Official Responses and Expert Commentary
The findings have sparked intense discussion among search industry veterans, who point out that the rules of digital visibility are being rewritten in real-time.
Kevin Indig on the Limits of Classic SEO
Kevin Indig, founder of Growth Memo and co-author of the study, emphasized that marketers cannot rely on old playbooks to win in the era of conversational search.
"Traditional SEO metrics aren’t enough to explain who owns a topic. While they play their role, there’s more to it," Indig noted. "The algorithms powering LLMs look at semantic associations and brand sentiment across a vast web of documents. Simply having high domain authority doesn’t mean an AI will recognize your brand as the definitive solution to a complex user problem."
The Strategic Shift to AI Visibility
The research underscores why Semrush has actively expanded its product offerings to include specialized tools like the Semrush AI Visibility Toolkit. As search behaviors migrate from standard query boxes to interactive chat interfaces, tracking share of voice requires entirely new monitoring frameworks.
Because Search Engine Land is owned by Semrush, the publication noted its commitment to providing objective coverage of these emerging marketing technologies, stating:
"Unless otherwise noted, this page’s content was written by either an employee or a paid contractor of Semrush Inc."
The collective expert consensus points to a single reality: brands must transition their measurement frameworks from keyword rankings on Search Engine Result Pages (SERPs) to tracking semantic share of voice within LLM outputs.
5. Implications: How Marketers Must Adapt to "AI SEO"
The fact that 85% of ChatGPT topic categories lack a dominant brand owner represents a massive, time-sensitive window of opportunity for enterprise and mid-market companies alike. However, capturing this territory requires a fundamental shift in how content is produced and optimized.
1. Shift from Keyword Optimization to Semantic Entity Building
To be consistently mentioned across ChatGPT’s five-prompt funnel, a brand must be recognized as an "entity" with a strong semantic relationship to its category.
- Action: Instead of writing articles targeting isolated keywords, build comprehensive resource hubs that answer every stage of the buyer journey—from initial definition queries to direct alternative comparisons.
- Focus on Co-occurrence: Ensure your brand name frequently co-occurs with industry-standard terms, competitor names, and key product features across high-quality external sites, forums, and press releases.
2. Optimize for Retrieval-Augmented Generation (RAG)
ChatGPT and other modern LLMs do not rely solely on static training data; they use real-time web retrieval (RAG) to ground their answers in current information.
- Action: To win citations and mentions in real-time AI searches, publish original research, proprietary data, and authoritative whitepapers. When third-party review sites, industry blogs, and news outlets reference your unique data, they feed the retrieval mechanisms that ChatGPT queries to construct its answers.
3. Prioritize Off-Site Brand Sentiment and Mentions
Because ChatGPT’s conversational engine synthesizes opinions and reviews from across the web, your on-site SEO is only a fraction of the equation.
- Action: Double down on digital PR, forum participation (such as Reddit and Quora), and third-party review platforms (like G2, Capterra, or Trustpilot). If real users consistently recommend your brand as an alternative or solution in online discussions, LLMs will reflect that consensus in their conversational outputs.
Summary of Strategic Adjustments
| Traditional SEO Focus | Next-Gen AI SEO Focus |
|---|---|
| Keyword Search Volume | Topic-Level Semantic Authority |
| Domain Authority & PageRank | Entity Association & Co-occurrence |
| Direct Clicks & CTR | Share of Voice & Conversational Mentions |
| On-site Content Hubs | Off-site Digital PR & Forum Presence |
Ultimately, the Semrush study proves that the AI search landscape is currently a frontier with few established rulers. The brands that act quickly to build comprehensive topic-level authority today will likely secure the highly stable, dominant positions that prove incredibly difficult for competitors to dislodge in the future.
