The Battle for Generative Visibility: Semrush Study Reveals ChatGPT’s Brand Recommendations Remain Highly Unsettled
As generative artificial intelligence increasingly shifts from a novelty tool to a primary search utility, brands are scrambling to understand how they are represented in AI-generated answers. For years, Search Engine Optimization (SEO) was a predictable science governed by keywords, backlinks, and domain authority. Today, the rise of "Generative Engine Optimization" (GEO) has introduced a highly volatile, algorithmic black box.
A groundbreaking study by SEO and digital marketing intelligence platform Semrush has pulled back the curtain on this new frontier. The analysis reveals that ChatGPT is far from settled on which brands define most buying categories. By analyzing over a thousand product and service categories, Semrush found that only 15.2% of topic categories had a clear brand leader across a series of related buyer questions.
This suggests that even for the world’s most established enterprises, AI search visibility remains a highly competitive, volatile, and wide-open battlefield.
1. Main Facts: The AI Visibility Gap
The Semrush study highlights a critical disconnect between a brand’s traditional search dominance and its visibility within Large Language Models (LLMs) like ChatGPT. For marketers who assume that high rankings on Google automatically translate to recommendations in AI chat interfaces, the data serves as a wake-up call.
CHATGPT CATEGORY LEADERSHIP STATUS
┌────────────────────────────────────────────────────────┐
│ [███████] Clear Leader (15.2%) │
│ [███████████████] Emerging Leader (31.2%) │
│ [───────────────────────────] No Consistent Leader (53.7%)│
└────────────────────────────────────────────────────────┘The study examined 1,094 U.S. business and product categories, testing how ChatGPT responded to a sequence of five common buyer prompts for each topic. These prompts spanned the entire customer journey, covering:
- Definitions: What is the product/service?
- Comparisons: How do different brands compare?
- Alternatives: What are the substitute options?
- Use Cases: How and when is the product used?
- Purchase Decisions: Which brand should a customer buy?
To qualify as a "Category Owner," a brand had to meet a rigorous dual standard: it had to appear in at least four of the five generated responses and lead its closest competitor by at least five percentage points in overall appearance frequency.
Out of more than a thousand categories, only 166 (15.2%) met this benchmark.
The rest of the landscape remains highly fragmented. Nearly a third of the categories (31.2%) featured an "emerging leader"—a brand that appeared in at least three prompts but failed to secure a decisive, statistically significant lead over its competitors. Meanwhile, the absolute majority of categories—53.7%—had no consistent brand leader at all, with ChatGPT cycling through various competitors depending on how the prompt was phrased.
2. Chronology: From Blue Links to Conversational Synthesis
To understand how brand visibility became so fragmented, it is necessary to trace the rapid evolution of search technology over the past decade.
Traditional SEO Era SGE & AI Search Era
┌─────────────────────┐ ┌────────────────────┐
│ • 10 Blue Links │ ─────────> │ • Direct Answers │
│ • Keyword Matching │ │ • RAG Synthesis │
│ • Domain Authority │ │ • Entity Relations │
└─────────────────────┘ └────────────────────┘The Keyword and Indexing Era (Pre-2022)
For over two decades, search engine visibility was linear. Google indexed web pages, evaluated them based on relevance and authority (primarily backlinks), and presented users with "10 blue links." Marketers optimized for specific keywords, and search engine results pages (SERPs) remained relatively stable, changing gradually over weeks or months.
The Generative Shift (November 2022)
The public launch of OpenAI’s ChatGPT in late 2022 disrupted this paradigm. Instead of directing users to external websites, ChatGPT synthesized vast quantities of training data to answer queries directly. This shifted user behavior from navigating directories to engaging in conversational discovery.
The Rise of SearchGPT and Hybrid Engines (2024–Present)
As OpenAI integrated real-time web browsing capabilities and launched "ChatGPT Search," the line between traditional search engines and LLMs blurred. Marketers realized that AI models were pulling information dynamically through Retrieval-Augmented Generation (RAG).
This transition transformed search from an indexing system into a recommendation engine. Semrush’s study, conducted during this mature phase of generative search adoption, represents the first comprehensive assessment of how reliably these recommendation engines favor specific brands over time.
3. Supporting Data: A Deep Dive into Semrush’s Findings
The data collected by Semrush paints a complex picture of how ChatGPT handles brand associations. The findings indicate that AI visibility is not a static metric; instead, it is highly sensitive to search volume, query type, and traditional brand strength.
The High-Demand Paradox
One of the study’s most striking revelations is that the most competitive, high-volume search categories are also the most unsettled.
LEADERSHIP STABILITY BY TOPIC DEMAND
┌──────────────────────────────────────────────────┐
High │ [█████] 11.3% Clear Owner │
Demand │ │
├──────────────────────────────────────────────────┤
Low │ [█████████] 19.0% Clear Owner │
Demand │ │
└──────────────────────────────────────────────────┘Among the half of the categories that generated the most AI search activity, only 11.3% had a clear brand owner. In contrast, less competitive, niche categories were more stable, with 19% producing a dominant brand.
This is particularly critical for enterprise marketers because high-demand topics accounted for 98% of the total AI search volume in the study. In other words, nearly all actual buyer activity occurs in categories where no single brand has successfully captured ChatGPT’s loyalty.
The Disconnect with Traditional SEO Metrics
To understand what drives AI visibility, Semrush compared ChatGPT’s category ownership against three pillars of traditional SEO:
- Branded Search Volume: How often users search for the brand by name on traditional search engines.
- Organic Traffic: The volume of non-paid visitors a brand’s website receives.
- Authority Score: Semrush’s proprietary metric grading the quality and quantity of a domain’s backlink profile.
The results showed a sharp divergence between old and new search paradigms:
| Metric | Correlation with ChatGPT Category Ownership |
|---|---|
| Branded Search Volume | Strong & Consistent: Brands with high direct search volume are consistently recognized by ChatGPT as category leaders. |
| Organic Web Traffic | Weak/No Relationship: High overall traffic does not guarantee that ChatGPT will mention or recommend the brand. |
| Authority Score (Backlinks) | Weak/No Relationship: High domain authority has little bearing on whether an LLM associates a brand with a topic. |
This suggests that LLMs prioritize real-world brand awareness and conversational prominence over technical website optimization.
Citations vs. Mentions: The 21% Overlap
Another crucial metric in generative search is the use of inline citations (links pointing to source websites). Marketers often assume that if their website is cited as a source, their brand is being recommended. Semrush’s data disproves this assumption.
The study found that only 21% of the most-cited domains within a category were also the most frequently mentioned brands in ChatGPT’s text responses.
Citations vs. Brand Mentions Overlap
┌─────────────────────────┐
│ [█████] 21% Overlap │
│ [─────────────────────] │ 79% Disconnect
└─────────────────────────┘This indicates that ChatGPT frequently uses authoritative third-party websites (like directory sites, review platforms, or news outlets) as sources to cite, while actually naming and recommending entirely different competitor brands in its written response.
Volatility in the Ranks
While established category leaders (the 15.2%) enjoyed high stability—remaining in first place in more than 90% of month-over-month comparisons—the rest of the market was highly volatile. Across the emerging and unsettled categories, the leading brand changed nearly 2,000 times during the study period. This high rate of fluctuation highlights how easily narrow leads can be lost to competitors.
4. Industry Analysis: Why LLMs Behave Differently Than Search Engines
The findings of the Semrush study underscore a fundamental shift in how search algorithms process information. Traditional search engines rely on information retrieval, whereas generative AI engines rely on information synthesis.
The Role of Retrieval-Augmented Generation (RAG)
When a user asks ChatGPT a buyer-intent question, the system does not simply search an index for keywords. Instead, it uses RAG to pull live web data, feeds that data into the LLM, and synthesizes a natural language answer.
Because the system synthesizes data on the fly, small variations in how a buyer asks a question can lead the model to pull from different data sources, causing the recommended brand to change. This explains why 53.7% of categories have no consistent leader.
Why Branded Search Volume Matters to AI
The strong correlation between branded search volume and AI category ownership reveals how LLMs are trained. LLMs learn from patterns in human language. If a brand has a high branded search volume, it is frequently mentioned across the web in association with its product category.
This deep semantic connection is baked into the model’s neural weights during training. Consequently, when asked for a recommendation, the model naturally associates the high-volume brand with the category, regardless of how well the brand’s website is optimized for SEO.
5. Implications: Redefining the Marketing Playbook for the AI Era
The transition from traditional SERPs to AI-driven answers requires a fundamental shift in digital marketing strategies. Marketers can no longer rely solely on legacy SEO playbooks to secure visibility.
LEGACY SEO PLAYBOOK GENERATIVE GEO PLAYBOOK
┌──────────────────────────────┐ ┌──────────────────────────────────┐
│ • Optimize for keywords │ │ • Build brand equity & queries │
│ • Build domain backlinks │ ──> │ • Secure PR & third-party reviews│
│ • Track search engine ranks │ │ • Optimize for full buyer journey│
└──────────────────────────────┘ └──────────────────────────────────┘1. Shift from Keywords to Entity-Based Marketing
LLMs understand the world in terms of "entities" (concepts, brands, people, and places) and the relationships between them. To win in generative search, brands must establish a strong semantic connection with their product category. This is achieved not by keyword stuffing, but by securing mentions in authoritative, contextually relevant contexts across the web, such as industry reports, news articles, and high-quality forums.
2. Prioritize Brand Building and PR over Technical Optimization
Since branded search volume is the only traditional metric that consistently aligns with ChatGPT category ownership, top-of-funnel brand building is more critical than ever. Public relations, brand advertising, and community building drive users to search for a brand by name. This activity signals to search engines and LLM web crawlers that the brand is a trusted authority in its space.
3. Optimize for the Entire Conversational Journey
Because visibility shifts as buyers ask different questions within the same topic, tracking rankings for a single prompt is no longer an effective KPI. Marketers must map out the entire multi-step conversational journey of their target audience. Content must be designed to answer definitions, comparisons, alternatives, and use cases comprehensively, ensuring the brand remains visible throughout the decision-making process.
4. Target Third-Party Review and Comparison Hubs
Given the 21% overlap between citations and mentions, brands must ensure they are highly visible on the platforms ChatGPT uses for information gathering. If ChatGPT regularly cites review aggregators, comparison tables, or industry forums to answer buyer questions, then having a strong, positive presence on those third-party platforms is essential to being mentioned in the final AI-generated response.
Conclusion: A New Era of Organic Competition
The Semrush study confirms that the generative search landscape is still in its infancy, characterized by high volatility and a lack of dominant brand leaders in key commercial categories. For market leaders, this volatility represents a threat to their established dominance; for agile competitors, it offers a historic window of opportunity to capture market share.
As AI search engines continue to mature, the brands that succeed will be those that look beyond traditional SEO metrics, focusing instead on building real-world brand authority and establishing deep semantic connections across the digital ecosystem.
