Alvaro Peña de Luna 
Share of model is the share of brand mentions in an AI model's answers about your category that belong to your brand, compared with your competitors. It is share of voice applied to a single AI model, such as ChatGPT or Perplexity, instead of to a media channel.
The name borrows from share of market and share of search, and the question behind it is simple: when people ask this model about your category, how much of the answer is about you?
How share of model works
You pick a set of prompts that a buyer in your category might ask, such as “best project management software for agencies”, “alternatives to [competitor]” or “which CRM works for a small sales team”, and run them on a model. Each answer names some brands. Share of model adds up the mentions of every brand you track and works out what part of that total is yours.
Each model retrieves different sources, has a different knowledge cutoff and phrases answers its own way, so your share can differ a lot from one model to another. That is the point of the metric: you read it per model, not averaged across all of them.
Share of model formula
Share of model = your brand's mentions / mentions of all tracked brands × 100
Worked example. You run 50 prompts on ChatGPT. Across the answers, your brand and four competitors are named 200 times in total, and 30 of those mentions are yours: your share of model on ChatGPT is 30 / 200 × 100 = 15%. The same 50 prompts on Perplexity produce 160 mentions, 32 of them yours: 20%. You now know where you are weaker and can check which competitors take your place there.
A raw count treats every mention as equal. A more useful version weights each mention by how the brand is presented, so that a recommendation counts for more than a name buried in a long list. The share of voice entry explains how that weighting works.
Share of model vs share of voice and mention rate
- Share of voice is the broader concept, used for ads, press, social, search and AI. Share of model is its AI-specific, per-model version.
- Mention rate is how often you appear at all (answers that name you / answers analysed × 100). It ignores competitors. You can have a high mention rate and a low share of model if every answer also names five rivals.
Read them together: mention rate tells you whether you are in the conversation, share of model tells you how much of it you own.
Why share of model matters
When a buyer asks an AI assistant for options, the answer usually names a handful of brands and stops. There is no page two. If a model keeps giving most of that space to two competitors, they are building preference with your audience before anyone visits a website. Tracking share of model per model shows where that happens and whether your work on content, PR and third-party coverage is changing it. Our guide to AI share of voice in GEO covers the tactics.
Training data vs live answers
Some people use share of model to mean what a model “knows” from its training data, measured by asking it with web access turned off. Others mean what users see in the product, where many assistants search the web before they answer. The two can differ: a newer brand may be absent from the training data yet appear in answers built from search results. Decide which one you measure and keep it consistent.
Common mistakes
- Averaging across models. A strong share on one engine can hide a weak one on another.
- Moving the competitor set. The denominator is the brands you track. Change the set and every share changes with it, so fix it before you compare periods.
- Using branded prompts. Prompts that name your brand inflate your share. Category prompts show whether the model picks you on its own.
- Reading one run as final. The same prompt can get a different answer tomorrow. Repeat the checks and read the trend.
How Mencoro measures share of model
Mencoro does not use the label share of model, but its Share of voice is calculated per day, engine and country, which is what share of model describes. Each mention is weighted by type × tone × qualifier: a positive, direct recommendation weighs 1.0, a neutral, direct listing 0.28, and a link to your domain counts as a direct, neutral reference (0.14).
Only you and your tracked competitors count. Untracked competitors appear in each answer but count in no metric until you track them, and when no tracked competitor appears, Share of voice stays empty rather than showing 100%. Mencoro asks ChatGPT with web search on, so it measures the answers people get in the product, alongside Perplexity, Google AI Overview and Google AI Mode. Read Share of voice next to Coverage, which says how often you are named at all.
The methodology page has every formula, the competitor analysis feature shows the comparison engine by engine, and you can try the weights on your own numbers in the share of voice calculator.
Related glossary terms
- Share of voice: your share of the conversation against competitors, across any channel.
- Mention rate: the share of answers that name your brand.
- AI brand mentions: the times an AI answer names your brand, and how.
- Large language model (LLM): the kind of model behind the answers you measure.
