AI Share of Voice: What It Is and How to Calculate It

Short answer

AI share of voice is your brand's mentions divided by all brand mentions across a set of AI answers, as a percentage. Ask your buyer questions on each engine, list every brand each answer names, then divide your count by the total. If you're named 20 times and all brands are named 90 times, your share of voice is 22%. Track it next to visibility, the share of answers that name you, because the two measure different things.

Key takeaways

Share of voice divides your mentions by all brand mentions. Visibility divides the answers that name you by all answers. You need both.
Tools define it differently: Peec AI and Profound count mentions, Ahrefs weights by estimated prompt demand. Compare numbers only within one method.
The brands you put in the denominator set the number, so fix the competitor rule and write it down.
For a brand named in about 30% of answers, a rate measured on 20 answers can be off by about 20 points either way. On 100 answers the margin is about 9 points.

What is AI share of voice?

Share of voice is an old marketing measure applied to a new place. The Marketing Accountability Standards Board's Universal Marketing Dictionary defines it as the "relative communication presence that a specific brand or product achieves", usually measured as a percentage of category media spend, impressions or gross rating points. It also describes a broader version built on mentions, where "share of voice would represent the proportion of the category discussion that the brand achieves."

AI share of voice uses that second version. The category discussion is the set of AI answers to your buyers' questions, and every time an answer names a brand, that brand earns a mention.

How do you calculate it?

Share of voice = your brand mentions ÷ all brand mentions × 100

Fix your questions and engines. Use the same buyer questions on the same engines every time. Our guide to measuring AI visibility covers building the set.
Collect the answers. Ask each question on each engine, and ask more than once, because answers vary between runs.
List the brands in each answer. Write down every brand the answer recommends or presents as an option.
Tally the mentions. Count each brand once per answer, however many times the answer repeats its name. Profound counts each brand once per run, and Ahrefs counts a mention once per response.
Divide. Your count over the total count, times 100.

Here is a worked example with made-up numbers. You collect 50 answers, and four brands are named in them:

Brand	Answers naming it	Share of all mentions
Tallybook (you)	20	22%
Ledgerly	35	39%
Billwise	25	28%
Paylane	10	11%
Total	90	100%

Tallybook's share of voice is 20 ÷ 90, or 22%.

What is the difference between share of voice and visibility?

Visibility is the share of answers that name you: 20 of 50, or 40%, in the example. Share of voice is your share of all mentions: 20 of 90, or 22%. Visibility tells you how often a buyer hears your name. Share of voice tells you how you compare with everyone else who was named.

Because most answers name several brands, share of voice is usually lower than visibility. In Profound's own worked example in its documentation, the same runs give a visibility score of 0.50 and a share of voice of 0.23. A brand can hold steady visibility while its share of voice falls, if competitors start appearing in more answers.

Do all tools calculate it the same way?

No. Here is how each tool's own documentation describes it:

Tool	How it defines share of voice	Detail worth knowing
Peec AI	The percentage of your brand mentions in AI responses compared with all brands mentioned, including yours	Visibility is a separate score: the share of responses that mention your brand
Profound	Mentions of your brand divided by mentions of all brands, each brand counted once per run	Calculated for each AI model separately, then averaged with equal weight
Ahrefs Brand Radar	A brand's share of impressions compared with other tracked brands	Impressions sum the Google search volumes of prompts where the brand appears, and platforms are weighted
Semrush AI Visibility Toolkit	The percentage of mentions your brand receives in AI answers compared with competitors in your market	No formula published
Citepoint	Your mentions as a share of all brand mentions across the answers	Shown with per-question results

So a share of voice from one tool can't be compared with a number from another. Even the word "visibility" differs: Semrush's AI Visibility score is a 0 to 100 benchmark that combines topic coverage and how consistently a brand is mentioned. Pick one method and stay with it.

Which competitors go in the denominator?

Share of voice is relative, so the brands you count decide the number. Count only your five tracked competitors, and a sixth brand that AI keeps naming stays invisible. Count every brand any answer names, and your share shrinks each time a new brand appears. Ahrefs warns that its metric can change a lot depending on how the brand and competitor entities are set up.

Choose a rule, write it down and keep it. In Citepoint, a language model reads each answer and lists every product it recommends or presents as an option, and we match that list against your brand, its aliases and your competitors. If an engine gave no answer, for example when Google showed no AI Overview, the answer counts neither for nor against you.

How many answers do you need?

Fewer than you might fear for a direction, more than you might hope for a precise number. AI answers differ every run (see why AI answers change), so a rate measured on a few answers is mostly noise. The standard margin of error for a proportion is z × √(p(1 - p) ÷ n), where p is the observed rate, n is the number of answers and z is 1.96 for 95% confidence. For a brand named in 30% of answers:

Answers collected	Margin of error	The true rate probably sits between
20	about 20 points	10% and 50%
50	about 13 points	17% and 43%
100	about 9 points	21% and 39%
400	about 4.5 points	25.5% and 34.5%

This table applies to a rate like visibility, where each answer either names you or doesn't. Share of voice is a ratio of counts, so its uncertainty is harder to write down, but it narrows with more answers in the same way. The formula needs more than five answers that name you and more than five that don't, according to OpenStax. For small counts, use the Wilson interval, which the NIST handbook prefers because it works for almost any sample size and its lower limit can't fall below zero.

SparkToro's research, which ran prompts 60 to 100 times each, concluded that to know an AI's set of recommendations you "need to ask over and over again; usually at least 60-100X, then average these out."

How do you read the number?

Compare like with like. Set the last 7 days against the 7 before, as Citepoint's methodology does, so one noisy day doesn't look like a trend.
Split by engine. A brand can lead on Gemini and be missing on ChatGPT. One blended number hides both.
Split by question. The average hides the questions you own and the ones you lose.
Check what changed in the denominator. Your share can fall because a new competitor entered the answers, not because you lost mentions.
Pair it with citations and AI referrals. Share of voice shows how often you're named. It doesn't show whether your pages are read or whether visitors arrive.

What is a good share of voice?

There is no universal benchmark, because it depends on how many brands the answers name. If four brands split the mentions evenly, each holds 25%. A more useful reference is the shape of a category. In Citepoint's October 2026 index, each category's leading brand was named in a median of 19 of 20 buyer questions, and the fifth brand in 10. Leaders dominate the answers, and a newer brand usually starts near zero. The goal is movement: a higher share this month than last, on more engines and more questions.

Frequently asked questions

Is AI share of voice the same as AI visibility?

No. Visibility is the share of answers that name your brand. Share of voice is your mentions as a share of all brand mentions across those answers. Visibility shows how often buyers hear your name, and share of voice shows how you compare with competitors.

Should I count a brand once per answer or every time it's named?

Once per answer is the common choice. Profound counts each brand once per run and Ahrefs counts a mention once per response, however often the brand appears in it. Counting once stops a long answer from inflating one brand.

How often should I calculate it?

Collect answers daily for the engines your buyers use most and weekly for the rest, then read the result over rolling periods such as the last 7 days against the 7 before. A single day's number is too noisy to act on.

Can I calculate it in a spreadsheet?

Yes. Keep a fixed list of questions, one row per engine per run, and a column for each brand named. Count the mentions per brand and divide yours by the total. The hard part is asking enough times, not the arithmetic.

Does a higher share of voice mean more traffic?

Not automatically. Share of voice shows how often AI names you. Whether people click, search your name later or buy is a separate question, so track AI referrals and conversions alongside it.