Share of Model (SoM): The New Metric for AI Search Visibility
As search behavior shifts from "searching and clicking" to "asking and reading," a new metric is emerging to measure brand visibility in the AI era: Share of Model (SoM).
What is Share of Model (SoM)?
Share of Model (SoM) is the percentage of times a brand is mentioned or recommended by Generative AI models (like ChatGPT, Gemini, or Perplexity) in response to category-relevant prompts, relative to its competitors.
Think of it as a brand's "presence" within an AI's knowledge base. It measures how effectively an LLM perceives, understands, and cites a brand when answering user queries. If a user asks an AI, "What are the best CRM tools for small businesses?", the AI's response—and whether it includes your brand—determines your SoM.
While traditional Share of Voice (SoV) focuses on the volume of media exposure (how much you spend or how often you rank on SERPs), Share of Model focuses on citation and recommendation. It is both a qualitative and quantitative measure of whether an AI trusts your brand enough to present it as an answer.
Share of Voice (SoV) vs. Share of Model (SoM)
What it measures
Brand visibility in ads, media, and search rankings
Brand mentions and recommendations in AI-generated answers
Primary focus
Impressions, clicks, rankings
Citations, recommendations, context
End user
Human searchers
AI engines (ChatGPT, Perplexity, etc.)
Success indicator
High ranking, traffic volume
Being part of the synthesized answer
Optimization approach
Traditional SEO (keywords, backlinks)
Generative Engine Optimization (GEO)
The key difference is the shift from volume-based visibility to contextual presence. In traditional search, being on page one was enough. In AI search, you need to be the answer—or at least part of it.
Why Share of Voice is Evolving
The traditional concept of Share of Voice is losing relevance as the fundamental mechanism of information discovery changes. In the Search Engine Results Page (SERP) era, users were presented with 10 blue links. Being visible (even in position 3 or 4) meant capturing a share of the traffic.
In the Generative Engine era, AI models often provide a single, synthesized answer. This creates a "Winner-Takes-Most" dynamic: if your brand isn't part of the synthesized response, your visibility is dramatically reduced.
Key Shifts Driving the Rise of SoM:
Zero-Click Search: Users are getting answers directly on the platform without visiting websites. Google's AI Overviews, ChatGPT's responses, and Perplexity's summaries all keep users within the platform.
Authority Filtering: AI models prioritize trusted sources. High SoV (buzz) doesn't guarantee high SoM if the AI perceives your content as low-quality or overly promotional.
Recommendation over Listing: AI doesn't just list options; it recommends solutions based on context, user intent, and perceived authority.
"In the AI era, visibility is increasingly binary: you are either part of the answer, or you're significantly less visible. Share of Model measures your presence in this new landscape."
How to Measure Share of Model
Measuring SoM requires a shift from tracking keywords to tracking prompts. The process involves testing AI models with specific scenarios to see how they respond.
1. Define Your Target Prompts
Instead of keywords like "best running shoes," use conversational prompts that mimic real user behavior:
"I need a running shoe for marathon training under $150. What do you recommend?"
"Compare Nike and Adidas for flat feet."
"What's the most durable running shoe for heavy runners?"
2. The Measurement Formula
To calculate SoM, you can use a simplified formula based on a set of test prompts:
SoM = (Your Brand Mentions / Total Brand Mentions in Category) × 100
However, a more nuanced approach includes these factors:
Visibility: Did the brand appear? (Yes/No)
Share of Discussion: How much text was dedicated to the brand relative to competitors?
Sentiment: Was the mention positive, neutral, or negative?
Recommendation: Was the brand explicitly recommended as the best choice?
3. Manual vs. Automated Tracking
Manual Tracking: Periodically asking ChatGPT, Perplexity, or other AI platforms key questions and recording the results.
Limitations: Time-consuming, subject to AI response variability, and difficult to scale. Best practice is to test at least 20-30 prompts per category and run tests weekly to account for model updates.
Automated Tracking: Platforms designed for Generative Engine Optimization (GEO) automate this process by running thousands of prompt variations to provide a statistically significant SoM score.
How to Improve Your Share of Model
Improving your Share of Model requires Generative Engine Optimization (GEO)—structuring your content so that AI models can easily parse, understand, and cite it.
Key strategies include:
Target Prompt Analysis: Identify the specific questions your audience is asking AI. Instead of optimizing for "CRM software," optimize for "What CRM should I use for a 10-person sales team with limited budget?"
Entity Authority: Consistently associate your brand with specific topics, features, and solutions. The more your content demonstrates expertise in a narrow domain, the more likely AI will recognize you as an authority.
Citation-Ready Content Structure: Write in formats that LLMs prefer—clear headings, logical flow, data-backed claims, and direct answers to common questions. Each section should be able to stand alone as a citable unit.
E-E-A-T Signals: Demonstrate Experience, Expertise, Authoritativeness, and Trustworthiness through author credentials, data sources, and real-world examples.
Tools like Deca approach this by analyzing target prompts and helping create content specifically structured for AI citation, turning your brand into a trusted data source that AI models are more likely to reference.
Conclusion
Share of Model represents a fundamental shift in how we measure brand visibility. As consumers increasingly rely on AI to curate their choices, brands that don't optimize for SoM risk becoming significantly less visible in the answers that matter most.
Key Takeaways:
SoM measures brand presence in AI-generated answers, not just search rankings
The shift from SoV to SoM reflects the evolution from click-based to answer-based search
Measuring SoM requires prompt-based testing across major AI platforms
Improving SoM demands a strategic shift to GEO and citation-ready content
Start measuring your SoM today by identifying your category's most common user prompts and testing how major AI platforms respond. Understanding your current position is the first step toward improving it.
Frequently Asked Questions (FAQs)
1. What is the difference between Share of Voice and Share of Model?
Share of Voice (SoV) measures visibility in traditional media and search engine rankings, focusing on impressions and clicks. Share of Model (SoM) measures how often a brand is mentioned or recommended by Generative AI models in response to user prompts, focusing on citation and presence in AI-generated answers.
2. How can I calculate my brand's Share of Model?
Select a set of relevant prompts (questions your customers ask), run them through AI models like ChatGPT or Perplexity, and track how often your brand is mentioned compared to competitors. The basic formula is: (Your Brand Mentions / Total Mentions in Category) × 100. For accuracy, test at least 20-30 prompts and repeat weekly.
3. Why is Share of Model important for my marketing strategy?
As search behavior shifts toward AI-generated answers (like Google's AI Overviews and ChatGPT), traditional organic traffic patterns are changing. SoM measures your visibility in this new "zero-click" environment. If you don't have a measurable SoM, you may not appear in the AI answers that users are reading instead of clicking links.
4. Can I improve my Share of Model?
Yes, through Generative Engine Optimization (GEO). This involves creating high-quality, authoritative content structured for AI readability. Strategies include using clear entity definitions, citing data sources, demonstrating E-E-A-T, and directly answering common user questions.
5. Which AI models should I track for SoM?
Track the models most relevant to your audience. Currently, the major ones are OpenAI's GPT-4 (ChatGPT), Google's Gemini, Perplexity, and Claude. Since search engines like Bing and Google integrate these models, tracking them covers the majority of AI search intent.
6. How long does it take to improve Share of Model?
Unlike traditional SEO which can take 3-6 months, SoM improvements can sometimes be observed within weeks, as AI models update their knowledge more frequently. However, building consistent authority that results in high SoM typically requires 2-3 months of strategic content creation.
7. Is Share of Model relevant for small brands?
Yes. SoM is particularly valuable for niche brands because AI models often provide more detailed, contextual answers than traditional search. A small brand with deep expertise in a specific area can achieve high SoM in its niche, even if it can't compete with larger brands on traditional SoV metrics.
References
Last updated