Share of Model
Share of Model (SoM) is the proportion of brand mentions a company receives from one or more large language models (LLMs) relative to total brand mentions in the same category. It quantifies how frequently and favorably AI platforms recommend a brand when users ask category-relevant questions.
Share of Model (SoM) is the proportion of brand mentions a company receives from one or more large language models (LLMs) relative to total brand mentions in the same category. It quantifies how frequently and favorably AI platforms recommend a brand when users ask category-relevant questions.
Why It Matters
As of 2026, ChatGPT alone has 815 million monthly active users and holds 60.7% of the AI search market. AI responses typically mention only one to three brands rather than displaying ten blue links. If yours isn't among them, you're invisible to a rapidly growing audience. Industry benchmarks suggest category leaders need 35–40% SoM on key prompts to maintain top-of-list positioning.
Share of Model vs. Share of Voice
| Metric | Share of Voice (SoV) | Share of Search | Share of Model (SoM) |
|---|---|---|---|
| Measures | Ad and media exposure | Brand-related search query volume | Brand mentions in AI responses |
| Key question | "How loud is our brand?" | "How often do people search for us?" | "How often does AI recommend us?" |
| Data source | Ad platforms, media monitoring | Google Trends, Search Console | LLM response collection and analysis |
How to Measure
- Design queries: Select 20–50 high-intent questions representing your category.
- Test across models: Submit identical queries to ChatGPT, Claude, Gemini, and Perplexity. Set temperature to 0 for consistency.
- Tally mentions: Record which brands appear in each response, noting frequency, position, and sentiment.
- Calculate share:
(Your brand mentions ÷ Total category mentions) × 100. - Track quarterly: LLM training data and algorithms shift frequently—measure at least once per quarter.
Tools for automated tracking include Profound, Conductor, Semrush, and HubSpot's AEO Grader.
Cross-Model Variation
Brand visibility can differ dramatically across LLMs. Ariel commanded nearly 24% of mentions on Meta's Llama but less than 1% on Google's Gemini, while Chanteclair held 19% on Perplexity but disappeared entirely from Llama. Single-model measurement is insufficient—always track across multiple platforms.
Improving Your Share of Model
- Publish authoritative content: Deep, E-E-A-T-rich content positions your brand as a category authority in LLM training data.
- Optimize for citations: Include statistics, research findings, and expert quotes so AI systems reference your content as a source.
- Expand presence on trusted platforms: Wikipedia, academic papers, and industry reports carry high weight in LLM training.
- Provide llms.txt: Help AI crawlers efficiently parse your site content with a structured llms.txt file.
Sources: