Share of Model Voice (SOMV) is a directional ratio calculated over a defined set of usable captured responses. It can compare how often the target brand and named competitors appear within that exact prompt, provider, model where known, geography where available, time, and capture cohort. It is not population market share, proprietary rank, customer demand, or proof that content caused a mention.
Current EchoScan evidence is analyst-captured
EchoScan can store analyst-captured responses and score compatible brand-versus-competitor evidence. Automated paid multi-engine capture is not launch-available. LLM Presence is a separate workflow for configured providers and prompts; it requires credentials, entitlement, schedules, capacity, and completed jobs.
Denominators and missing evidence matter
Only usable responses belong in the denominator. Missing, failed, empty, or incompatible observations must stay disclosed instead of being converted to zero. Comparisons should preserve the same cohort definition and show the sample size.
Change is not automatic attribution
A higher ratio in a later compatible cohort is an observed difference. It does not by itself prove that a publication, build, campaign, or other action caused the change.
Run a free diagnostic for structural site evidence; it does not produce a provider-mention baseline.