Share of voice (AI)
How often your business shows up in AI engine answers for the questions in your category, relative to competitors, measured across a set of prompts at a point in time.
You probably already checked whether an engine names you for one question. Share of voice is the version that asks the harder thing: across all the questions a buyer might ask, how often is it you, and how often is it them.
What share of voice is
Share of voice in AI search measures how present you are in engine answers for your category, relative to competitors. You take a set of prompts a real buyer would ask, run them across engines, and count who gets named.
In old search, ten links meant ten chances to appear. In an AI answer, an engine may name two or three businesses and stop. Presence is scarcer, so the share that exists is worth measuring.
How it is measured
You define the prompts that matter, the engines you care about, and a moment to measure. Then you count: in how many answers do you appear, and how does that compare to the names beside you.
It is a snapshot, not a constant. Run the same prompts a month later and the picture can shift, because the inputs underneath, your pages and the engines, both move.
The old way and the new way
The old way read share of voice off ranked results: tally positions across keywords, average them out. The shelf was long and visibility was graded.
The new way reads it off generated answers, where the shelf is short and the engine has already narrowed the field. Being absent from a three-name answer is a different problem than being on page two.
The damaging admission
A share-of-voice number is a measurement, not a lever you pull directly. The engines decide what to say, and they change. We can show you the snapshot and how it moves, and we can help your pages read more clearly, but we cannot promise the number climbs.
And if you compete in a tiny niche with almost no AI search demand, tracking this weekly is overkill. It earns its keep when buyers in your category really are asking engines and getting answers without you in them.
How to check yours
Pick the handful of questions a buyer asks before choosing someone like you. Ask each across the engines and note who gets named. That rough count is your starting picture.
Run a free scan on any URL to see how engines read you today, or read how AI engines decide what to recommend.