What Is AI Share of Voice and How Do You Measure It?

AI share of voice is the percentage of brand mentions across AI-generated answers in your category that belong to you rather than to competitors. To measure AI share of voice, run a fixed prompt set across the major engines, count every brand mention in the answers, and divide your mentions by the total.

What Share of Voice Tells You That Mention Rate Cannot

Mention rate tells you how often you appear. Share of voice tells you how much of the available attention you capture, which is what determines whether an AI-assisted buyer hears your name or a rival's. Two companies can both appear in half the answers in their category, yet one dominates every list it appears in while the other is always the last name mentioned. Share of voice, especially when weighted, exposes that difference. Mention rate and its siblings are covered in our guide to AI visibility metrics and KPIs.

The Basic Calculation, Step by Step

1. Fix a prompt set of 20 to 50 buying-intent questions for your category
2. Run every prompt on each engine you care about, ideally two or three times
3. Count every brand mention in every answer, yours and your competitors'
4. Divide your mentions by total mentions to get your share of voice
5. Repeat with the identical prompt set each month and chart the trend

As a worked example, if the answers across your prompt set contain 200 brand mentions and 30 of them are yours, your AI share of voice is 15 percent. Whether that is strong or weak depends entirely on how many serious competitors share the category with you.

Weight for Position and Recommendation Strength

Not all mentions are equal. A lead recommendation influences buyers far more than a passing reference, so many teams apply a simple weighting: three points for being the first or clearly preferred option, two points for a top-three mention, and one point for any other appearance. Weighted share of voice tracks actual buying influence more honestly than raw counts, and it rewards the improvements that matter.

A Worked Example: Weighted Share of Voice in Practice

Take two project management tools that each appear in 20 answers across a 50-prompt test. On raw counts they are tied. Apply the weighting and the tie disappears. Brand A leads the answer 12 times, sits in the top three 6 times, and trails twice: 3 points times 12, plus 2 times 6, plus 1 times 2, for a score of 50. Brand B leads twice, sits mid-list 8 times, and is a passing reference 10 times: 3 times 2, plus 2 times 8, plus 1 times 10, for a score of 32. Same mention count, very different influence on buyers. If you tracked only raw mentions, Brand B's team might believe it has parity when it is actually being presented as an afterthought in most of the answers where it appears.

Segment by Engine and by Topic

A single blended number hides the story. Calculate share of voice separately for each engine, because winning on Perplexity while being invisible in Google AI Overviews calls for different work. Segment by topic as well: you might dominate answers about one service line while being absent from another that drives more revenue. Segments tell you exactly where to invest next.

How to Grow Your Share

Share of voice is close to zero-sum, so growth means winning answers competitors currently own. Study the prompts where rivals appear and you do not: which sources the engines cite, what questions are really being asked, and what content those competitors have published that you have not. Then close the gap with clearer, more citable content and a stronger presence on the sources the engines already trust. If rivals keep winning answers you deserve, our guide on why AI recommends your competitors instead of you breaks down the usual causes, and what sources AI engines trust shows where to build your presence.

Frequently Asked Questions

What is a good AI share of voice?

It depends on how fragmented your category is. In a market with two or three serious players, anything below a roughly even split suggests you are underweight. The more useful benchmark is trend and rank: aim to be the most-mentioned brand for the prompts tied to your best revenue.

How many prompts do I need for a reliable number?

Enough to smooth out run-to-run variation, which for most businesses means at least a few dozen prompts, each run more than once per engine. Small prompt sets make the metric jumpy; a fixed, broad set run consistently makes it trustworthy.

Related reading

• How to Measure AI Visibility for Your Business
• Benchmarking Your AI Visibility Against Competitors
• Tools for Tracking AI Search Visibility
• How AI Engines Decide Which Businesses to Recommend

Claim Your Share of the AI Conversation

Every AI answer in your category is shelf space, and right now some set of brands is filling it. Measure your share, watch it monthly, and go win the answers that matter most. GrowBiz10x calculates AI share of voice across engines and competitors automatically, so you always know your slice of the conversation. Get your share of voice baseline with GrowBiz10x.