← Blog · 2026-07-02 · share of modelAI visibility metricAIVO score

Share of Model: The Metric That Replaces Search Rankings in the AI Era

You can't rank #1 in an AI answer — you're either recommended or invisible. Share of Model quantifies it: definition, measurement method, and benchmarks.

TL;DR — Share of Model (SoM) = the percentage of relevant buyer questions for which an AI engine recommends your brand. Measured across engines with neutral, brand-free prompts, it's the AI-era equivalent of search ranking + share of voice combined. Below 20% you're invisible; 40%+ makes you a default answer in your category.

Definition and measurement

Take 25 real buying questions in your category ("best gentle retinol for beginners?"). Ask each of China's six major engines. Count the answers that mention or recommend your brand. That percentage is your Share of Model. Two rules make it honest: questions must never contain your brand name (unprompted recall is what matters), and mentions are verified against your Chinese aliases — engines usually cite the 中文名.

Why it beats old metrics

• vs. search ranking: AI gives one answer; position #4 doesn't exist.
• vs. share of voice: SoV counts what you publish; SoM counts what the AI actually internalized.
• Position-weighted SoM adds nuance: being the first recommendation is worth more than a trailing mention.

Reading your score

0–15%: entity problem — engines barely know you exist. 15–35%: recognized but not preferred; competitors dominate the sources. 35–55%: contender — narrative shaping decides who wins. 55%+: you are the category default. Track monthly; single runs have natural variance.

FAQ

Why does my score change between runs? LLM outputs are stochastic and retrieval sources refresh. Trend over 3+ measurements is the signal.

Is one engine enough? No — each engine trusts different sources, and your buyers are spread across all of them.