Written by Roman ChornovolShare of voice is the proportion of total category visibility, across a defined channel or medium, that a given brand occupies relative to its competitors.
Also known as: share of voice, SoV, AI share of voice
What Share of Voice Measures
Share of voice (SoV) quantifies how much of the available attention in a category belongs to a particular brand. In traditional media it was calculated from ad spend or impressions. In organic search it became keyword-level presence. In social it became mention volume. The denominator always represents the whole conversation: every competitor, every placement, every instance where the category surfaces. The numerator is the brand’s slice of that whole.
The metric is useful precisely because it’s relative. A brand can increase raw mentions while losing share if competitors grow faster. That competitive lens is what separates SoV from simpler reach or impression counts.
Why Share of Voice Gets Complicated in AI Search
AI search engines such as ChatGPT, Gemini, Perplexity, Google AI Overviews, and Copilot don’t return a list of ten blue links where counting appearances is straightforward. They produce synthesized, prose answers that recommend, explain, and contextualize. A brand can be named in an answer without being recommended. It can be recommended without being named. Two answers to the same prompt can name the same brand while telling completely opposite stories about it.
This means presence-based SoV, counting how often a brand appears across AI answers, captures a real but shallow signal. Mentions are a downstream symptom. What actually drives the recommendation is the narrative the model has adopted about the category: which frame it uses, whose definitions anchor the answer, and which sources it draws on when constructing its response. AI models are downstream of human-market consensus. They learn a story before they tell it.
Counting appearances without reading the frame is like auditing ad spend without knowing whether the ads persuaded anyone.
Narrative Share: The Upstream Metric
Mavel treats share of voice as a starting point, not a destination. The metric that matters is narrative share: whose story the model is actually telling about your category when it answers. A brand with modest mention-level SoV but strong narrative share owns the frame. Its definitions, its problem framing, and its criteria for evaluation are the ones the model reproduces. A brand with high mention-level SoV but weak narrative share is present without influence: visibility without guidance.
The practical implication is that optimizing for mentions without diagnosing the underlying narrative means chasing a symptom. Representation can be corrected only when you understand which frame is winning and why. Share of voice tells you the score. Narrative share tells you who wrote the rules of the game.