AI share of voice is the share of relevant AI answers that name your brand, measured against your competitors, across engines like ChatGPT, Gemini, Claude and Perplexity. It is the metric that turns a pile of volatile single readings into one comparable number you can track over time and benchmark against rivals. This guide covers what it is, the formula, and how to measure it step by step. If the wider concept is new, start with what AI visibility is.
Why share of voice beats a single check
A single check tells you almost nothing, because AI answers are volatile: the same query changes roughly 70 per cent of the time, and two identical queries return the same brand list under a 1-in-100 chance (SparkToro). Share of voice fixes this by aggregating many readings into a rate, so normal noise averages out and a real trend becomes visible. It also makes you comparable: instead of asking whether you appeared once, you ask what share of the answers named you versus the competitors that matter.
Share of AI citations
Most-cited domains in AI answers
The formula
AI share of voice is a simple ratio. Across a fixed set of buyer questions run across the major models, it is the number of answers that name your brand divided by the total number of relevant answers, expressed as a percentage. A competitive version divides your mentions by the combined mentions of you plus your named competitors, which tells you your slice of the whole conversation rather than a raw rate.
Step 1: Define your prompt set and competitor set
Start with 15 to 30 buyer questions that matter in your category, phrased the way a customer types them, mixing category recommendations, comparisons and problem-led queries. Then name the competitors you want to measure against, the brands you actually lose deals to, not every company in the space. Keep both sets fixed, because share of voice is only meaningful when the inputs stay the same between readings.
Step 2: Run the set across the engines and record mentions
Run every question across ChatGPT, Gemini, Claude, Perplexity and Google's AI Overviews, and for each answer record whether you are named, in what position, and which competitors appear. Run each question more than once, because answers vary between sessions, and treat the aggregate as the reading. Coverage differs by engine, so measure each separately before you combine them.
Step 3: Calculate your share
Count the answers that name you and divide by the total relevant answers. As an illustrative example, if you run 20 questions across four models three times each, that is 240 answers; if your brand is named in 72 of them, your raw share of voice is 30 per cent. For the competitive version, divide your 72 mentions by the total mentions across you and your competitors. Position matters too: many teams weight a first-place mention more heavily than a passing one, because the first name carries disproportionate weight with readers.
Want to see this in action?
See how every major AI model talks about your brand. Free to start.
| Metric | How to calculate it | What it tells you |
|---|---|---|
| Raw share of voice | Your mentions / total relevant answers | How often AI names you at all |
| Competitive share of voice | Your mentions / (you + competitors' mentions) | Your slice of the named conversation |
| Weighted share of voice | Mentions scored by position, then divided | How prominently, not just how often |
Step 4: Benchmark against competitors
A share-of-voice number is only useful in context. Run the same prompt set for your key competitors and compare. This is where the metric earns its keep: it shows whether you are gaining or losing ground in the answers that decide deals, and which competitors are pulling ahead. If a rival's share is climbing while yours is flat, that is a signal worth acting on.
Step 5: Track the trend, not the snapshot
Measure on a fixed cadence, weekly or daily, and read the direction of travel over weeks rather than reacting to any single reading. Report ranges, not single numbers, and expect wobble. A sustained change across multiple engines is the signal; a one-off swing on one model is noise. For the fuller monitoring process this sits inside, see how to monitor your AI brand mentions.
Where the numbers come from, and their limits
Two caveats keep share of voice honest. First, a mention is not the same as a citation: 62 per cent of AI citations never surface the brand behind them, so a model can lean on a source about you without naming you, which your share of voice will not capture. Second, share of voice measures presence, not sentiment; being named as the cautious option counts the same as being the confident recommendation, so read it alongside how you are described. To improve the number, the levers sit off your own site: AI visibility correlates most with third-party mentions and video (Ahrefs).
Doing it manually versus with a tool
You can calculate share of voice in a spreadsheet for a first baseline, and it is a genuinely useful exercise. It becomes impractical at scale, because you need to run many prompts across several models repeatedly, record positions and competitors, and chart the trend. That is what dedicated tools automate. For the options, see the 9 best AI brand monitoring tools and the 8 best AI visibility tools.
Start with a baseline
The fastest way to get your first reading is to run your brand through a free AI visibility check, see which models name you and which name your competitors, and use that as the baseline your share-of-voice tracking builds on.
