Share of voice in AI answers is how often, and how favourably, assistants name your brand when people ask about your space. A tool asks a model a set of questions, records which brands it names and tracks that over time. Most of the AI SEO market is built on this.
It is a real signal, and worth watching once you are established, because it tells you whether the models have picked you up. But be clear what it is: a lagging, aggregate measure. It tells you the score, not why, and it moves slowly.
The levers that move it are upstream, at the page level. Whether a crawler can read your content, whether your pages answer the real questions, whether your facts are trusted and cited. Chasing the score without fixing the pages is chasing a shadow.
Kenovar works at that page level on purpose. Fix what an engine can read, answer and trust, and share of voice follows. Tracking the score is a later layer, and one we treat as a byproduct of the work, not the work itself.