SoV adapts the classic marketing measure to LLM answers: your mentions divided by the total mentions of the whole competitor set, across a prompt set. It is intuitive and widely used as a relative-standing read.
Rhinegold's refinement: SoV is only as honest as its competitor set. An incomplete set systematically inflates a brand's SoV, and the bias grows as the market adds players. Naïve counting also over-counts brands with several aliases — span-aware detection corrects it. Read SoV as a directional share, not an absolute; when exclusivity is the real question, Brand Recommendation Share is the competition-weighted refinement.
SoV gives a quick read of relative standing across a competitor set and is well suited to tracking that standing over time.
It is sensitive to competitor-set completeness — an incomplete set inflates the number — and to alias over-counting. And it says nothing about exclusivity or about whether the brand was cited as a source.
Against Mention Rate, which is absolute presence, not a relative share. Against Brand Recommendation Share, which weights each answer by how many providers were named. And against Citation Rate, which is about authority, not share of mentions.
Neither the counting rules nor the competitor-set definition are standardized across tools, so cross-tool SoV figures are not comparable.