What Is Share of Model?
Short answer
Share of model measures how often AI systems mention or recommend your brand compared with competitors when answering relevant prompts. It's the AI-era counterpart to share of voice, tracking presence inside model answers rather than ad impressions or search rankings.
Share of Model explained
Share of model was coined by marketers Jack Smyth and Tom Roach as the successor metric to share of voice for a world where buyers ask AI assistants for recommendations instead of scanning ads or search results. The term has a trademark claimant, so you'll see it written as Share of Model™ in some vendor material; the underlying concept is used generically across the industry.
Measurement works by querying AI engines systematically. You define a prompt set that represents how your buyers actually ask (comparisons, recommendations, "best X for Y" questions), run it across engines like ChatGPT, Perplexity, Gemini, and Google's AI Overviews, and score how often each brand is mentioned, how it's characterized, and whether it's recommended.
Two caveats keep the metric honest. Answers vary between engines, between runs, and between small changes in prompt wording, so single-shot checks mislead; trends across a consistent prompt set over time are what carry signal. And there's no public ground truth for how often real users see any answer, which makes share of model a directional benchmark rather than an audited number.
The reason to track it now is simple: the answers are already being generated, with or without you in them. A McKinsey survey of Fortune 500 consumer-brand CMOs, published in its October 2025 report on AI search, found that only 16% of brands systematically track their AI-search performance. The gap between how often buyers ask and how few brands measure is where the early advantage sits.
In practice
When I baseline a client's AI presence for the first time, the number is almost always in the single digits. One client's initial baseline showed they appeared in just 7.2% of relevant AI answers in their category. That number isn't a failure. It's the starting line, and it turns AI visibility from a vague worry into something you can measure and move.
Common misconception
People often assume share of model can be bought like ad share of voice. There's no placement to purchase inside model answers. Share of model is earned through citations, entity signals, and content that AI systems can retrieve and trust.
Free email course: Smarter SEO for the Age of AI
Seven short lessons over three weeks: why clicks are falling even when rankings hold, and how to get AI systems to recommend your brand.
