For decades, marketers tracked share of voice to measure how much of a market’s attention their brand owned compared to competitors. That worked well when people found products through ads, search results and articles. But a growing number of people now ask AI tools directly for recommendations instead. This shift has created a new metric called Share of Model. Here is what it means, how it works and why marketers are starting to track it.
What Does Share of Model Mean?
Share of Model, often shortened to SoM, measures how often, how prominently and how favorably a brand appears in AI generated answers, compared to its competitors. It is the AI era counterpart to share of voice.
Share of voice tracked how much of a category’s advertising or organic search presence a brand owned. Share of Model tracks something different. It measures how much of an AI model’s output a brand owns when someone asks ChatGPT, Gemini, Claude or Perplexity a question related to that category.
Share of Model asks one direct question. When someone asks an AI tool for a recommendation in your industry, does your brand show up in the answer?
Why This Metric Exists Now
Large language models have changed how people search for products and services. Many people no longer type keywords into Google and browse several websites. Instead, they ask an AI tool a direct question and receive a complete answer, sometimes without ever clicking through to a website afterward.
This creates a real measurement gap. A business can rank well on Google and still miss entirely from an AI generated answer to a similar question. This happens because AI tools select and mention brands using different signals than traditional search rankings use. Share of Model exists to close that measurement gap.
How Is Share of Model Calculated?
The basic calculation is straightforward. You ask a representative set of category related questions to major AI tools. Then you count how often an AI mentions your brand and compare that to the total number of brand mentions across all the answers.
Here is an example. Say you sell CRM software. An AI tool recommends brands 100 times across a set of relevant prompts. Your brand appears in 25 of those recommendations. Your Share of Model for that topic is 25 percent.
A keyword ranking stays fixed. You are either ranked first or you are not. Share of Model works differently, it is probabilistic. An AI model might mention your brand in 80 percent of responses to one question and only 20 percent of responses to a closely related question. Traditional rankings do not work this way, your position stays fixed until something changes it.
Two Ways to Calculate Share of Model
Different platforms calculate Share of Model in two distinct ways. It helps to know both before you compare any numbers.
Brand Mention Share looks at your total mentions as a percentage of all category mentions:
Total Mentions of Your Brand ÷ Total Mentions of All Category Brands × 100
Query Inclusion Rate, sometimes called a win rate, looks at how many prompts included your brand at all. It does not matter how many total mentions appeared:
Prompts Featuring Your Brand ÷ Total Prompts Tested × 100
These two numbers can tell different stories. A brand might appear in a smaller share of total mentions but still show up in most of the prompts tested or the reverse can happen. When you compare Share of Model data, always check which formula the source used. Mixing the two can create a misleading picture.
It also helps to understand where an AI model’s answer actually comes from. Some AI tools generate answers purely from patterns they learned during training. Marketers sometimes call this parametric memory. Other tools use Retrieval Augmented Generation or RAG. These tools search the live web before answering, a method Perplexity, Gemini with Search and ChatGPT Search all use.
This distinction matters. Fresh mentions, like a recent press release or a new review, tend to move the needle faster on RAG based tools. Broader industry recognition takes longer to shift results on models that rely only on parametric memory.
Results can also shift based on a few factors. How you phrase a question matters. Which AI tool you use matters. Even your conversation history leading up to the question can matter. Test a range of question phrasings, rather than just one fixed prompt per topic, to get a more reliable picture of your actual visibility.
The Three Parts of Share of Model
Current industry breakdowns generally split Share of Model tracking into three components:
- Mention frequency. This is the raw percentage of times your brand appears in an AI response for your target topics and questions.
- Prominence. This looks at where your brand appears in the answer. The AI might name it first, list it among several options or mention it only briefly.
- Sentiment. This measures whether the AI describes your brand in a positive, neutral or negative way, not just whether it mentions your brand at all.
Share of Model vs Share of Voice: The Key Difference
Share of voice measured presence in advertising and traditional organic search results. Share of Model measures presence inside AI generated answers instead, a fundamentally different discovery channel.
They are not competing metrics, they are complementary. Share of voice still matters for traditional search visibility. Share of Model measures a newer, separate channel, one traditional metrics were never built to track. A brand can have strong share of voice in Google search results and still have very low Share of Model. This happens when AI tools rarely mention that brand by name.
How to Start Measuring Your Own Share of Model
You do not need an expensive platform to get a basic sense of where you stand. Here is a simple starting approach:
- Write down a list of real questions a potential customer might ask an AI tool about your type of product or service.
- Ask those exact questions directly in ChatGPT, Gemini, Claude and Perplexity.
- Record whether your brand appears and if so, how the AI positions and describes it.
- Repeat this regularly, since AI answers can change as models update and as your content changes.
Several specialized tools have emerged for more consistent, ongoing tracking. These include Otterly.ai, Peec AI, Brand24 AI and other AI visibility platforms. Each one runs repeated prompts against multiple AI tools and reports mention frequency and sentiment over time. This is a fast-moving and increasingly competitive tool category, so check each provider’s own website for current features and pricing before you choose one.
What Improves Your Share of Model
The same practices that support Answer Engine Optimization also tend to improve Share of Model, based on current industry guidance. Write clear, well-structured content that directly answers real questions. Keep your information accurate and current. Earn mentions on third party sites like industry publications, review platforms and community forums, since brand mentions on external sites also appear to influence how often AI models recognize and recommend a brand.
Simply repeating your brand name or keywords does not reliably improve this metric. AI models respond more to the depth and clarity of the information available about your brand across the web, not to repetition. This connects closely to what we cover in our guide to Answer Engine Optimization (AEO), the same foundational work supports both.
Final Thoughts
Share of Model is a genuinely new way to measure brand visibility. AI tools increasingly make recommendations directly now, instead of just linking to search results and this metric was built specifically for that shift. It does not replace share of voice entirely, but it fills a real gap that traditional metrics cannot see. Knowing where your brand currently stands in AI generated answers is quickly becoming as important as knowing where it stands on a traditional search results page. If you want help building content that improves your visibility inside AI answers, get in touch with our team.
FAQ Section:
What is Share of Model in simple terms?
It is a metric that measures how often, how prominently and how favorably AI tools like ChatGPT, Gemini and Claude mention a brand, compared to competitors.
How is Share of Model different from share of voice?
Share of voice measures presence in advertising and traditional search results. Share of Model measures presence specifically inside AI generated answers, a separate and newer discovery channel.
How do you calculate Share of Model?
It is usually calculated in one of two ways: Brand Mention Share (your brand’s mentions divided by all category brand mentions) or Query Inclusion Rate (the percentage of relevant test prompts where your brand appears at all).
Does ranking well on Google guarantee a strong Share of Model?
No. A brand can rank well on Google and still have low Share of Model. AI tools select and mention brands using different signals than traditional search rankings use.
Comments are closed