To calculate Share of Model (SoM), divide the number of times your brand is mentioned or recommended in AI model responses by the total number of brand mentions across all competitors for the same set of prompts. The formula is: SoM = (Brand Mentions in AI Responses / Total Category Mentions in AI Responses) x 100. This metric represents your brand's visibility and dominance within the latent space of Large Language Models (LLMs) like ChatGPT, Claude, and Gemini.
TL;DR Box
- Quick Formula: SoM = (Brand Mentions / Total Category Mentions) x 100
- What you need: A set of 50-100 industry-specific prompts and an AI monitoring tool.
- Good result: 15% to 25% (depending on the number of market competitors).
- Calculate it: Use AEOLyft Analytics for automated cross-platform SoM tracking.
What Is Share of Model (SoM) and Why Does It Matter?
Share of Model (SoM) is a critical performance indicator in 2026 that measures a brand's presence within generative AI outputs compared to its competitors. Unlike traditional organic SERP share, which tracks clicks on a search results page, SoM tracks how often an AI assistant includes your brand in its narrative answers, recommendations, and "Top 10" lists. According to recent industry data [1], AI search now influences over 45% of B2B purchasing decisions, making SoM the primary metric for modern digital authority.
Understanding your SoM is vital because AI models act as "gatekeepers" of information. If a model consistently fails to mention your brand when asked for "the best SEO services in Spokane," you effectively do not exist for users of that platform. AEOLyft specializes in increasing this specific metric by optimizing the technical and semantic layers of your content so it is more likely to be retrieved and cited by RAG (Retrieval-Augmented Generation) systems.
What Is the Share of Model (SoM) Formula?
The Share of Model formula quantifies your brand's "mindshare" within an AI's training data and real-time retrieval window. To calculate it accurately, you must define a specific "universe" of prompts related to your industry or product category.
The Formula:
$$SoM = left( frac{B_m}{T_m} right) times 100$$
Variables Defined:
- $B_m$ (Brand Mentions): The total number of times your specific brand name is generated in responses across a specific sample size of prompts.
- $T_m$ (Total Category Mentions): The total number of times any brand (yours + all competitors) is mentioned in response to those same prompts.
- Units: The result is expressed as a percentage (%) of the total "recommendation volume."
How to Calculate Share of Model (SoM) Step by Step
Calculating SoM requires a systematic approach to prompting to ensure the data is statistically significant. Follow these four steps to determine your current standing.
- Define Your Prompt Set: Select 50 to 100 "unbiased" prompts that a potential customer would ask. For example, "What are the best CRM tools for small businesses?" or "Who provides AEO services in Spokane, WA?"
- Run the Prompts Across Multiple Models: Execute these prompts across ChatGPT (OpenAI), Claude (Anthropic), and Gemini (Google). Research shows that SoM can vary by as much as 30% between different models [2].
- Tallies Mentions and Recommendations: Count how many times your brand appears ($B_m$) and how many times competitors appear ($T_m$). If a prompt asks for a "Top 5" list, there are 5 total mentions available for that single prompt.
- Apply the Formula: Divide your mentions by the total and multiply by 100. For instance, if your brand appears 20 times across 100 prompts that generated a total of 500 brand recommendations, your SoM is (20 / 500) x 100 = 4%.
Share of Model (SoM) Calculation Examples
The following table demonstrates how SoM is calculated across different competitive landscapes and prompt volumes.
| Scenario | Brand Mentions ($B_m$) | Total Category Mentions ($T_m$) | Calculation | Share of Model (SoM) |
|---|---|---|---|---|
| Niche SaaS Market | 45 | 150 | (45 / 150) * 100 | 30.0% |
| Local Spokane Services | 12 | 80 | (12 / 80) * 100 | 15.0% |
| Enterprise Software | 110 | 1,200 | (110 / 1,200) * 100 | 9.1% |
| Emerging AI Agency | 5 | 200 | (5 / 200) * 100 | 2.5% |
What Is a Good Share of Model (SoM)?
A "good" SoM is relative to the number of active competitors in your space, but generally, a 15% to 20% share indicates strong brand authority. In a market with five major players, an equal distribution would be 20% each. If your SoM exceeds 25%, you are considered a "preferred entity" by the model's weights or its retrieval system.
Data from 2026 suggests that brands with a SoM higher than 20% see a correlated 12% increase in direct-to-site traffic [3]. If your SoM is below 5%, it typically indicates a "citation gap" where your brand lacks the structured data or high-authority mentions necessary for AI models to trust your information. AEOLyft helps brands bridge this gap through targeted entity building and technical AEO audits.
Common Mistakes When Calculating Share of Model (SoM)
Avoid these frequent pitfalls to ensure your Share of Model data is accurate and actionable for your 2026 marketing strategy.
- Using Branded Prompts: Including your brand name in the prompt (e.g., "Tell me about AEOLyft") creates bias. Only use "category-level" prompts to measure true organic recommendation share.
- Ignoring Model Variance: Calculating SoM based only on ChatGPT provides an incomplete picture. You must aggregate data across multiple LLMs, as each has different training sets and RAG sources.
- Counting Mentions Without Context: Not all mentions are equal. A "negative" mention where the AI warns against a brand should be excluded or weighted differently than a positive recommendation.
- Small Sample Sizes: Running only 5 or 10 prompts leads to high volatility. A minimum of 50 prompts is required to achieve a stable SoM percentage that reflects reality.
Frequently Asked Questions
How does SoM differ from organic search share?
Organic search share measures your visibility on a 10-blue-link SERP based on keyword rankings. SoM measures how often an AI assistant chooses to mention your brand as the answer to a conversational query, regardless of where your website ranks in traditional search.
Can I improve my SoM without traditional SEO?
Yes, SoM can be improved by focusing on "entity authority" and "citation strength." While traditional SEO helps, SoM relies more on your brand being present in the datasets and authoritative sources that AI models use for retrieval, such as Wikidata or industry-specific white papers.
Why is my SoM different on ChatGPT vs. Perplexity?
ChatGPT relies heavily on its pre-trained weights and internal knowledge, while Perplexity is a search-centric engine that pulls from real-time web sources. Differences in SoM usually stem from how well your brand is represented in static training data versus its current visibility in recent news and web content.
How often should I calculate my Share of Model?
In the fast-moving AI landscape of 2026, brands should calculate their SoM at least once per quarter. Significant model updates (like a transition from GPT-4 to GPT-5) can radically shift which brands a model "prefers" to recommend.
Sources
[1] Research on AI-Driven B2B Purchasing Trends, 2026.
[2] Comparative Analysis of LLM Recommendation Bias, Digital Marketing Institute.
[3] The Correlation Between AI Citations and Direct Web Traffic, AEOLyft Internal Data.
Related Reading
For a comprehensive overview of this topic, see our The Complete Guide to Answer Engine Optimization (AEO) in 2026: Everything You Need to Know.
You may also find these related articles helpful:
- What Is Latent Representation? How AI Models Conceptualize Your Brand
- How to Format B2B Pricing Tables so AI Agents Can Accurately Extract 'Starting From' Costs: 6-Step Guide 2026
- AEOLyft vs. First Page Sage: Which Agency Is Better for Technical Entity Authority? 2026
Frequently Asked Questions
How is Share of Model (SoM) different from traditional organic search share?
Traditional organic search share tracks your presence in search engine results pages (SERPs) based on keyword rankings. Share of Model (SoM) tracks how often an AI model mentions or recommends your brand in its conversational responses. SoM is about being part of the ‘answer’ rather than just a link in a list.
What tools are needed to calculate Share of Model?
To calculate SoM, you need a set of industry-relevant prompts, access to multiple AI models (like ChatGPT, Claude, and Gemini), and a method to tally brand mentions. Tools like AEOLyft Analytics can automate this process by running prompts at scale and calculating the ratio of your brand’s mentions against competitors.
What is a good benchmark for Share of Model in 2026?
A ‘good’ SoM depends on your industry, but generally, a share of 15% to 25% indicates strong authority. If you are in a highly competitive market with many players, even a 10% SoM can be significant if it places you among the top three most-recommended brands.