As buying decisions move into AI chat, marketers need a metric for this new arena. Share of Model (SoM) — sometimes called AI share of voice — measures how often an AI engine recommends your brand relative to competitors for the prompts that matter. It is the AI-era answer to share of voice, and it is fast becoming a core KPI for GEO.
Defining Share of Model
At its simplest, SoM is the percentage of relevant AI answers in which your brand appears, compared with the total mentions across you and your competitors. If ten buyer prompts produce twenty brand mentions and four of them are yours, your SoM is twenty percent. It captures presence in the conversation, not just whether you rank somewhere.
Why SoM matters
SoM translates the vague question "are we visible in AI?" into a comparable number. It shows not only whether you appear, but how you stack up against rivals in the exact moments customers ask for recommendations. Tracked over time, it reveals whether your GEO efforts are gaining or losing ground.
How to measure it
Start with the prompts your customers actually use, spanning different intents — early discovery, direct purchase, and alternatives to a competitor. Run them across the major engines, and because outputs vary, run them repeatedly. Record every brand mentioned, then compute your share. The key is a consistent, repeatable methodology so changes reflect reality, not randomness.
Measuring by intent
A single overall number hides useful detail. You might dominate discovery prompts yet vanish when users ask for alternatives to a rival. Breaking SoM down by intent shows exactly where you are strong and where you are exposed, so you can target optimization precisely.
From measurement to action
SoM is most valuable when it feeds a loop: measure, optimize the weak areas, and re-measure. Geomyze is built for this — it runs intent-based prompts across AI engines, computes your share of voice against competitors, and tracks it over time, so Share of Model becomes a metric you can actively grow.
