Share of Voice (AI)

Share of voice turns a vague worry ("does AI recommend us?") into a number with competitors attached. It also exposes market structure: in our five-metro study, some markets concentrate — one Phoenix roofer held 10 of 18 answers — while dental answers scattered across dozens of practices with no leader anywhere.

The denominator is the part that decides whether the number means anything, and it is where most vendor reporting quietly cheats. Share of what? Across which questions, asked how many times, on which assistants, on what date? A business named in 10 of 18 answers has a defensible 56%. A business named in 1 of 1 has 100% and has learned nothing. Any share figure without its sample attached is a marketing number.

It is also worth separating share from quality of mention, because the two come apart. Appearing fifth in a list of five is a mention and contributes to share, but it is not the same as being the business an assistant leads with and explains. Reading the raw answers reveals which one you are getting; a summary statistic never will.

The competitive read is usually more actionable than your own figure. Ranking every business in a market by share shows immediately whether you are chasing one entrenched incumbent, competing in a crowded middle, or operating somewhere no one has established anything — three situations that call for entirely different work. That structural question is what the openness score summarises in a single number.

Because assistants are non-deterministic, share is a sampled estimate, honest only with dates and sample sizes attached. Ours are published per market, with every underlying answer recorded and the exact prompts listed on the methodology page.