AI Mention Finder

AI share of voice calculator

Turn your own counts ("named in 5 of 12 answers") into share of voice, and see the range each number could plausibly sit in. Import totals straight from checks saved in this browser.

Total AI answers you collected with one fixed prompt set.

BrandAnswers naming it

Result

Your brand31% share · named in 42%

Plausible mention rate: 19%–68% (95% interval)

Competitor A50% share · named in 67%

Plausible mention rate: 39%–86% (95% interval)

Competitor B19% share · named in 25%

Plausible mention rate: 9%–53% (95% interval)

Share = this brand's mentions ÷ all mentions of the brands listed. Mention rate = answers naming it ÷ answers sampled. Both describe your sample, not the market. If two brands' ranges overlap a lot, you can't yet say one is ahead.

Two numbers, not one

Share of voice answers "of all the mentions among these brands, what fraction went to us?" It always adds up to 100% across the brands you list, so adding or removing a competitor changes everyone's share.

Mention rate answers "in what fraction of answers did we appear at all?" It doesn't depend on who else you list, which makes it the steadier number to track month to month.

Why the ranges matter

The bars show a 95% Wilson score interval for each mention rate, a standard way to express uncertainty for a proportion from a small sample. When ranges overlap heavily, the honest reading is "too close to call", no matter what the point estimates say. The AI visibility guide has an interactive version that shows how the range shrinks as you collect more answers.

What this number can't tell you

These figures describe the prompts and models you sampled. They are not a share of real conversations: no outside tool can see what people actually type into ChatGPT or how often. Keep the prompt set fixed, note the date and models, and compare like with like.

Need counts to put in? Run the mention check, which saves every run here in your browser.

Questions

What is AI share of voice?
AI share of voice is the portion of brand mentions in AI-generated answers that go to your brand, compared with a set of competitors, across a defined set of prompts. It is a sample-based metric: it depends entirely on which prompts and models you chose.
How do I calculate AI share of voice?
Collect answers to a fixed set of prompts, count how many answers name each brand, then divide each brand's count by the total count for all brands. Also report each brand's mention rate (answers naming it divided by answers sampled) with an error range, so readers can see how certain the figure is.
Why does my AI share of voice change so much?
AI answers vary between runs, and small samples exaggerate that variation. With 12 answers, a brand named in 5 has a plausible true rate anywhere from roughly 19% to 68%. Collect more answers before reading much into a change.