AI Share of Voice Calculator
AI share of voice is your brand's share of all brand mentions in the AI answers you check: your mentions divided by every tracked brand's mentions. Enter your counts from ChatGPT, Google's AI and other assistants. The calculator adds a 95% margin of error, so you can see whether a gap to a competitor is real or noise.
Your results
Gap to the leader: May be noise
You're probably behind Competitor A, but check about 385 answers to confirm the gap within ±5 points.
AI share of voice
23%
95% range: 16%–31%
Mention rate
25%
95% range: 18%–33%
Citation rate
10%
95% range: 5.8%–17%
Based on 120 answers and 132 brand mentions in total.
Share of voice by brand
- Competitor A36% (29%–45%) · 48 answers
- Competitor B27% (20%–35%) · 36 answers
- You23% (16%–31%) · 30 answers
- Competitor C14% (8.8%–21%) · 18 answers
Bars show each 95% range on a 0–50% scale. Overlapping ranges mean the difference may be noise.
How many answers to check
Answers needed for a margin of error of ±m points (worst case, a 50% rate). You've checked 120.
- ±10 points97 answers (reached)
- ±5 points385 answers
- ±3 points1,068 answers
Adding distinct prompts helps more than adding repeat runs of the same prompt.
Your inputs are stored in the link itself, after the #. They aren't sent to our server.
How this is calculated
Everything is calculated in your browser from your own counts. A brand counts once per answer, however many times the answer names it.
- Answers checked (N) = prompts × runs per prompt × platforms. You can overwrite it.
- Mention rate = x ÷ N, where x is the number of answers that mention your brand.
- Citation rate = c ÷ N, where c is the number of answers that cite or link your site.
- Total brand mentions (T) = x + the answers mentioning each competitor (m₁ + m₂ + …).
- AI share of voice = x ÷ T: your mentions divided by all tracked brands' mentions (Semrush's definition).
- Leader share = the top competitor's mentions ÷ T. Gap to the leader = leader share − your share, in percentage points.
- First-mention rate = f ÷ x, where f is the number of answers that name your brand first.
The margin of error
Each rate gets a 95% range from the Wilson score interval. For a rate p̂ = k ÷ n, with z = 1.96:
(p̂ + z²/2n ± z·√(p̂(1 − p̂)/n + z²/4n²)) ÷ (1 + z²/n)
It's applied to the mention rate (k = x, n = N), the citation rate (k = c, n = N) and each brand's share of voice (k = its mentions, n = T). If your range overlaps the leader's, the gap may be noise, and the calculator says how many answers would settle it.
Answers needed for a ±m margin, using the worst case (a 50% rate):
n = ⌈1.96² × 0.25 ÷ m²⌉
That's 97 answers for ±10 points and 385 for ±5 points.
Worked example (the default inputs)
- N = 20 × 3 × 2 = 120 answers; T = 30 + 48 + 36 + 18 = 132 brand mentions.
- Mention rate = 30 ÷ 120 = 25.0% (95% range 18%–33%).
- Citation rate = 12 ÷ 120 = 10.0% (5.8%–17%).
- AI share of voice = 30 ÷ 132 = 22.7% (16%–31%).
- Leader (Competitor A) = 48 ÷ 132 = 36.4% (29%–45%). Gap = 13.6 points.
- The ranges overlap (31% vs 29%), so the result reads: “You're probably behind Competitor A, but check about 385 answers to confirm the gap within ±5 points.”
Assumptions & sources
This calculator uses no benchmark constants. Every number is your own count, and it shows no “good share of voice” tiers, because none could be verified. The sources below explain the method.
| Constant | Value | Source | As of |
|---|---|---|---|
| ConstantRepeated runs where ChatGPT or Google AI Overviews returned the same brand listThe study found visibility % across many runs more consistent than ranking position, which is why this calculator measures mention share, not rank. | Value< 1% of the time | SourceSearch Engine Journal on the SparkToro and Gumshoe AI recommendations studySparkToro and Gumshoe: hundreds of volunteers ran 2,961 prompts through ChatGPT, Claude and Google AI Overviews, 60–100 times per prompt per platformAlso: Original SparkToro post (Not verified: returned HTTP 403 to the fetch tool on 2026-09-29 [verify original]) | As of
|
| ConstantAI share of voice definition | ValueYour brand's mentions ÷ all tracked brands' mentions. | SourceSemrush, “How to measure AI share of voice” | As of
|
Why mention share rather than rank position. SparkToro and Gumshoe had hundreds of volunteers run 2,961 prompts through ChatGPT, Claude and Google AI Overviews, 60–100 times per prompt per platform, in November–December 2025. ChatGPT and Google AI Overviews returned the same brand list less than 1% of the time, and the study found visibility % across many runs to be more consistent than ranking position.
Why 3 runs per prompt by default.It's Vidern's methodological choice, not a published standard. The Search Engine Journal write-up says the right number of runs is still an open question.
Why the Wilson interval. z = 1.96 is the standard multiplier for a 95% interval. The Wilson score interval stays sensible for small counts and for rates close to 0% or 100%.
Limits of this estimate
- AI answers change from run to run, and differ by location, account and model version. This is a snapshot of the answers you checked.
- The margin of error assumes each answer is independent. Repeat runs of the same prompt are related, so the true uncertainty is larger. Adding more distinct prompts helps more than adding more runs.
- Count brand name variants consistently, e.g. “Acme” and “Acme Inc.”.
- A mention isn't a citation. Being named doesn't mean you get the link.
Frequently asked questions
What is AI share of voice?
AI share of voice is the share of all brand mentions in AI answers that belong to your brand: your mentions divided by the mentions of every brand you track. If the answers you checked mention you 30 times and three competitors 102 times in total, your share of voice is 30 ÷ 132, about 23%. It's the definition Semrush uses.
How many AI answers should I check?
Enough for the margin of error you need. In the worst case, about 97 answers give a margin of ±10 percentage points and about 385 give ±5 points. Repeat runs of the same prompt are related, so spread your answers across more distinct prompts rather than running a few prompts many times.
What's the difference between mention rate, citation rate and share of voice?
Mention rate is the share of answers that name your brand. Citation rate is the share that cite or link your site. Share of voice compares your mentions with your competitors' mentions. A brand can be named without being linked, so the three numbers answer different questions.
Why do AI assistants give different answers each time?
AI answers vary from run to run, and by location, account and model version. In a SparkToro and Gumshoe study reported by Search Engine Journal, ChatGPT and Google AI Overviews returned the same brand list less than 1% of the time. That's why this calculator measures how often you're mentioned across many answers, not your position in a single answer.
What is a good AI share of voice?
There's no verified benchmark, so this calculator doesn't grade your score. The useful comparison is with the competitors in your own prompt set: whether you're ahead of or behind the leader, and whether the gap is bigger than the margin of error.
Can I track AI mentions automatically?
Yes. Vidern's AI Visibility Checker checks whether ChatGPT, Gemini, Perplexity, Claude and Google AI Overviews mention your website for your keywords, so you don't have to count answers by hand.
Next step
Stop counting by hand: track mentions across ChatGPT, Gemini, Perplexity, Claude and AI Overviews.
Try the AI Visibility CheckerWant an expert review of your AI visibility? Or get a free GEO audit
By Malhar Shah, Founder & SEO Strategist at Vidern. Updated .