AI share of voice: your slice of the answer.
AI share of voice measures how often AI assistants name your brand compared with your competitors when people ask questions in your category. Instead of asking “did we appear”, it asks “what slice of the answers is ours”, across ChatGPT, Claude, Perplexity and Google’s AI.
The short version
- AI share of voice is your brand’s slice of AI answers, measured against competitors, not in isolation.
- It reframes visibility as a race: the goal is a bigger share than rivals, on the questions that matter.
- You calculate it by running a fixed prompt set and counting who gets named, and how prominently.
- The lever that moves it is out-showing rivals where AI reads, especially Reddit comparison threads.
- subding catches the competitor and category threads live so a person can take part in time.
What AI share of voice means
Borrow the idea from traditional media, where share of voice meant your slice of the advertising in a category, then point it at AI answers. AI share of voice asks: when people put category questions to an assistant, of all the brands it names, how big a fraction is you.
It is a comparative measure by design. A simple tracker might tell you that you appeared in an answer and feel encouraged, while three competitors appeared in the same one and got chosen first. Share of voice refuses to let you feel good about that. It counts everyone.
Appearing is table stakes. The question that matters is how much of the answer belongs to you rather than the field.
Why a share metric beats a yes or no
A present-or-absent check flatters you. It can hold steady while your position quietly erodes, because it never notices that rivals are being named more often or more warmly. A share metric catches that drift, because your slice shrinks the moment theirs grows.
It also aims your effort. When you can see that a competitor owns the answer to a specific, valuable question, you know exactly where to concentrate, rather than spreading energy across questions you already win. It turns a vague worry into a target.
How to calculate it
The maths is deliberately simple; the discipline is in doing it consistently.
- Fix the questions. A stable set of category prompts you rerun each period, so movement means something.
- Count every brand. For each answer, log all the brands named, not only your own.
- Weight by prominence. Optionally give a first, warm mention more weight than a passing one at the end.
- Divide. Your mentions, weighted or not, over total brand mentions across the set, gives your share.
The mechanics of running and logging the prompts are the same as in AEO tracking, and the monitoring surface for it is AI citation monitoring.
What actually moves your share
Share is comparative, so moving it means out-performing rivals in the sources the models read. For recommendation and comparison questions, that ground is community discussion, and Reddit sits right in the middle of it. When a thread weighs you against a competitor, the outcome can echo through AI answers for weeks.
trimming our stack this quarter and torn between two options everyone mentions. which SEO tool would you keep if you could only afford one?
→ A fair, first-hand answer helps
A thread like that is where share is won or lost, and turning up to it late means turning up after the comparison is settled. Catching the competitor and category threads while they are live is the practical lever, which is what Reddit competitor tracking and a Reddit monitoring tool are for. The reply stays a person’s job, made honestly.
How to start
You can stand up a first measurement this week and refine it later.
- 1
Name your competitors
Decide which brands count as your field, so you know whose mentions to count against yours.
- 2
Build a prompt set
Write the category questions where you want to be chosen, and keep the list fixed between runs.
- 3
Run and tally
Put the prompts to the assistants, record every brand in every answer, and work out your share.
- 4
Watch the trend
Repeat on a schedule and track whether your slice is growing or shrinking against the field.
The honest limits
Share of voice is a useful lens, not a precise instrument, and it pays to hold it loosely.
What to keep in mind
- It is an estimate. Answers vary between runs, so your share is a trend line, not a certified figure.
- Framing beats frequency. Lots of lukewarm mentions can be worth less than a few strong ones, so read the tone too.
- Your prompt set shapes it. Choose questions that reflect real buyer intent, or you will measure a race nobody is running.
- subding measures nothing here. It watches Reddit and flags the thread when your brand or a rival comes up. It does not score AI answers, and it never posts on Reddit for you.
Used sensibly, share of voice keeps you honest about where you really stand against the field, and points you at the threads where you can change it.
Questions, answered straight
What is AI share of voice?
AI share of voice is the proportion of AI answers in your category that name your brand, measured against the competitors named alongside you. It treats visibility as a share of a finite pie rather than a simple present-or-absent check.
How is it different from an AEO tracker?
An AEO tracker tends to ask whether you appear at all. Share of voice takes the same runs and turns them into a comparison: of everyone named for these questions, what fraction is you. It is the competitive read on the same data.
How do you calculate AI share of voice?
Fix a set of category questions, run them across the assistants regularly, and for each answer record every brand named. Your share is how often you are named, ideally weighted by prominence, divided by all brand mentions across the set.
What moves AI share of voice?
Being chosen over rivals where the models read. For recommendation questions that means the Reddit and community threads where people compare options; earning the nod there, honestly and repeatedly, is what shifts your share over time.
Is a higher AI share of voice always better?
Higher is generally good, but context matters. Being named a lot in dismissive terms is worse than fewer, warmer mentions, so read the framing alongside the count rather than chasing raw frequency.
Can I guarantee I will gain share?
No. You influence what the models say by being the better, more-present option; you do not control it, and the answers shift on their own. Aim to improve the odds and the trend, not to command a number.
Grow your slice where it’s decided
Share of voice moves when you out-show rivals in the threads AI reads. Catch the comparison while it’s live and a real answer from you can tilt it. subding is what tells you the thread has opened.
See what every plan includes on the pricing page.