Your Visibility Score tells you whether AI models mention you. Share of Voice tells you who is winning. It is the single best metric for the competitive question, because it forces every mention into context: being named five times means something completely different in a market where the leader is named fifty times versus eight.
What is AI Share of Voice?
AI Share of Voice is the percentage of all brand mentions in AI answers within your market that belong to your brand. If the models answering your category prompts name brands 200 times in a month and 30 of those mentions are yours, your Share of Voice is 15 percent.
The formula:
AI Share of Voice = (your brand mentions \u00f7 total brand mentions across all tracked brands) \u00d7 100
How is Share of Voice different from mentions, citations and visibility?
Metric | What it counts | What it answers |
|---|---|---|
Mention | Your brand named in an answer | Am I present? |
Citation | Your site linked as a source | Does the model use my content? |
Visibility Score | Share of prompts where you appear | How broad is my presence? |
Share of Voice | Your mentions versus everyone else\u2019s | Am I winning the market conversation? |
Teams that track only their own mentions routinely misread their situation. Mentions can rise while Share of Voice falls, which means the market conversation is growing faster than your part of it. That combination looks like progress and is actually erosion.
How do you measure it in practice?
Define the competitor set. Five to ten brands that genuinely compete for the same prompts. Too broad and the denominator becomes noise; too narrow and you flatter yourself.
Track shared prompts. Run the same category prompts, "best [category]", "top tools for [use case]", across the models that matter, daily.
Count every brand mention. Yours and each competitor\u2019s, per model, per day.
Compute the share. With the formula above, and keep the daily series so you can see trend, not just level.
Manually this is genuinely painful, which is why it is usually the first metric teams automate. Trackbase computes Share of Voice daily against up to 10 competitors across 11 models, with the trend lines built in.
What is a good AI Share of Voice?
Benchmark against your market position, not an absolute number. A rough rule that holds up in practice: your AI Share of Voice should at least match your market share. Ahead of it, AI is a growth channel for you. Behind it, competitors are converting your customers at the research stage before you ever see them in your funnel.
How do you improve it?
Four moves, in the order we would make them:
Close the coverage gaps. Find prompts where competitors appear and you do not, and build the content that answers those exact questions.
Expand beyond your own site. Reviews, comparison articles and community presence are the sources models actually cite when comparing brands.
Fix the technical floor. AI crawlers must be able to read your site, and your structured data must say clearly what you do.
Watch competitor moves. A sudden Share of Voice jump for a competitor almost always traces to a specific new source. Finding it tells you exactly what to respond to.
FAQ
How many competitors should be in the calculation?
Five to ten. Enough to represent the market conversation, few enough that the metric stays interpretable. Revisit the set quarterly; AI answers sometimes surface competitors you were not watching.
Should Share of Voice be measured per model or overall?
Both. The overall number is the headline; the per-model split is the diagnosis. It is normal to hold 20 percent on Perplexity and 5 percent on Gemini, and the gap tells you where to work.
How often should I check it?
The data should refresh daily; your review can be weekly. Single-day swings are noise, weekly trends are signal.
Can Share of Voice be 100 percent?
Only in markets so narrow that models know one brand. If you measure 100 percent, your prompt set is probably too branded. Category prompts should be phrased the way a customer who never heard of you would ask.
Does a higher Share of Voice actually produce revenue?
It measures presence in answers that demonstrably shape shortlists. Connect it to your analytics: AI referral traffic tends to be small in volume and unusually high in intent, which is consistent with what we see across tracked sites.
