How to Calculate AI Share of Voice, Visibility and Average Position

Start with the raw material: answers and mentions

Every AI visibility metric is built from the same two things: a set of answers (one AI response to one question, from one model, at one time) and the brand mentions found in each answer. Decide these before you compute anything:

• Which questions. Real buyer questions for your category, frozen so they don’t drift between periods.
• Which models. Report each model separately before you combine them; they disagree a lot.
• What counts as a mention. Your brand name as a whole word, its known variants, and not a longer name that contains it (“Ramp” inside “Ramp Network” is not Ramp).

1. Visibility rate

visibility = answers that mention you ÷ total answers × 100

The share of answers in which you appear at all. It is the most robust metric because it doesn’t depend on how many other brands an answer lists. An answer that names you three times still counts once.

2. Share of voice

share of voice = your mentions ÷ mentions of all brands × 100

Your slice of every brand-naming across the answer set. It goes down when answers start naming more competitors, even if your own visibility didn’t change — which is the point: it measures how crowded the answer is around you. If no brand was named at all, share of voice is undefined, not 0%.

3. Average position

average position = sum of your ranks ÷ number of answers that ranked you

When an answer is a list, your rank is your place in it. Average only over the answers that ranked you: counting a miss as “position 11” would turn this into a noisier copy of visibility. Report it next to visibility, never alone — position 1.0 from one answer in twenty is not a strong result.

A worked example

An illustrative example with made-up brands: four answers to “best expense management software for startups”.

• Visibility: YourBrand appears in answers 1 and 3 → 2 ÷ 4 = 50%.
• Share of voice: 13 brand namings in total, 2 of them YourBrand → 2 ÷ 13 = 15.4%. Acme, named in all four answers, has 4 ÷ 13 = 30.8%.
• Average position: ranks 3 and 1 → (3 + 1) ÷ 2 = 2.0.

Same data, three different stories: you appear in half the answers, own a sixth of the conversation, and rank well when you do appear.

Edge cases that make tools disagree

• Mentions vs citations. A mention is your name in the answer text. A citation is your URL in its sources. They are different metrics; label which one you report.
• Prose vs list mentions. “Brands like YourBrand are popular” is a mention without a rank. Count it for visibility and share of voice, leave it out of average position.
• Repeated samples. If you ask each question several times, count per answer, not per question, so the denominators match.
• Combining models. Averaging ChatGPT and Gemini hides the most useful finding — that one of them ignores you. Show per-model figures first.
• Small samples. Every one of these numbers is an estimate. See how many answers you need before you read a change as real.

The definitions we use

These are the same definitions behind the numbers in Mentioned: visibility is mentioned answers over answers, share of voice is your namings over all brand namings and is left blank when no brand was named, and average position is averaged only over the answers that ranked you. Our glossary defines the surrounding terms, and how it works shows how each answer is scored.

Cite this

You are welcome to quote and chart these figures with a link back. Suggested citation: Mentioned, “How to Calculate AI Share of Voice, Visibility and Average Position,” October 25, 2026, https://mentioned.at/learn/ai-share-of-voice-formula. Data and text are published under CC BY 4.0.

Run a free check