What is Share of Voice (SOV)?
Share of voice in AI search is the proportion of tracked questions in a category where a brand is named or cited, measured against the total for all brands appearing. It is a relative measure: it moves when competitors gain or lose visibility, not only when the brand does.
The distinction from citation share matters. Citation share asks what proportion of a brand's own tracked questions cited it — an absolute measure of that brand. Share of voice asks what proportion of all brand appearances across those questions belonged to it, which is inherently comparative.
A brand can hold citation share steady while share of voice falls, because a competitor started being cited alongside it on the same answers. That is a real competitive loss an absolute metric will not show.
Both numbers are only as honest as the question set behind them. A set chosen to flatter — questions naming the brand, or niche phrasings nobody types — produces a high number that means nothing. The questions have to be ones a buyer would actually ask before they knew the brand existed.
The denominator also has to be stated. If a question returned no AI Overview at all, there was no answer to appear in, and counting it as an absence makes visibility look worse than it is while implying a gap no content could close.
Why Share of Voice matters
It is the only visibility metric that moves when a competitor gains ground, which is the change most worth reacting to.
Relative, not absolute
It can fall while a brand's own citations hold steady, because a competitor gained. That is the signal.
The question set is the measurement
Questions that name the brand answer themselves. A set built to flatter produces a meaningless number.
Exclude questions with no answer
A query that produced no AI Overview offered nothing to appear in. Counting it as an absence reports an unclosable gap.
Sample repeatedly before trusting a move
Generative answers vary between runs for the same question. A single run is an anecdote, not a trend.