"Share of Model" Is the New Metric Replacing "Ranking"

There's a term showing up in more marketing conversations every week, and if you haven't run into it yet, you will soon: Share of Model. It's quietly becoming the scoreboard for AI visibility, the way "ranking" was the scoreboard for classic SEO.

Here's the plain definition: Share of Model is the percentage of AI-generated answers in your category that mention or recommend your brand, measured against your competitors. Ask ChatGPT, Perplexity, Gemini, or Google's AI Overviews a representative question in your space — "best CRM for a mid-size retailer," "how do I fix X problem" — and count which brands show up in the answer. Run that across enough real questions, and the resulting percentage is your Share of Model. It's essentially share of voice, rebuilt for a channel where the "voice" is an AI-generated answer instead of a search results page.

The term has picked up other names too — share of citation, AI share of voice, share of answer — all pointing at the same underlying measurement. Worth knowing so you recognize it regardless of which label a given tool or article uses.

Why this matters more than it might sound: rankings tell you where a URL sits on a page. They don't tell you whether your brand actually shows up inside the answer a buyer reads. Those are increasingly different questions with different answers. A page can rank well and still never get named when someone asks an AI system the question that page answers.

One finding worth sitting with: the same brand can have wildly different Share of Model across different AI systems. One INSEAD study found a well-known detergent brand scored nearly 24% Share of Model on Meta's Llama, and under 1% on Gemini — same brand, same category, same time period. Visibility "in AI" isn't one number. It's model by model, and a brand can be strong in one and functionally invisible in another.

How to actually measure it, without needing a paid tool to start: build a set of 15 to 25 real questions your buyers would ask — branded, topical, comparison, and "best option" style questions. Run them across ChatGPT, Perplexity, Gemini, and Google's AI Overviews, using a signed-out or incognito session so your own history doesn't skew results. Log whether you're mentioned, how prominently, and how favorably. A spreadsheet is genuinely enough to start.

Rough orientation, not a target to chase blindly: under 15% mention rate is early-stage, 15-35% is developing, 35-60% is established, and above 60% marks a category leader — though these vary enormously by category and competitive density. The number that actually matters is the trend line over months, not any single reading.

The uncomfortable part for a lot of marketing teams: this is genuinely new measurement work, not a repackaged version of what you're already tracking. Most haven't started. That's either a gap to close or a window to move first in, depending entirely on how quickly you act on it.