Being mentioned in an AI answer matters less than how you compare to everyone else mentioned alongside you. Share-of-voice turns visibility from an absolute fact into a relative one — it tells you not just whether you appear, but how much of the conversation you actually own.
- Share-of-voice is your mentions divided by the mentions of you plus your tracked competitors, across a prompt set and engines.
- A high visibility rate means little on its own — what matters is your position relative to rivals on the same questions.
- Measure it per-engine and in aggregate, on a fixed prompt set, and watch it over time rather than as a one-off reading.
- A single number hides the story; the value is in the trend line and the per-prompt breakdown that show where to act.
What share-of-voice means here
Share-of-voice is a simple ratio with a sharp implication. It is the number of times your brand appears in AI answers divided by the total appearances of you and the competitors you've chosen to track, measured across the same set of prompts and the same engines. If you and three rivals are the named players in your category, and across a probe run your brand surfaces in 20 of every 100 mentions, your share-of-voice is 20% — regardless of how often the engines mention anyone at all.
That framing is deliberate. A raw visibility rate answers "how often am I mentioned?" Share-of-voice answers "how much of the mentioned conversation is mine?" The first is about you in isolation; the second is about you against the field. Because AI answers tend to name a short list of options rather than a single winner, the more useful question is almost always the relative one: when an engine hands a buyer a shortlist, how often is your name on it compared to the alternatives?
One thing to keep honest from the start: share-of-voice is defined entirely by the competitor set you pick. It is not a universal score handed down by the engines — it is a measurement you scope. Change the rivals you track and the number moves, which is why the comparison set deserves as much care as the metric itself.
Why absolute visibility isn't enough
Suppose you appear in 60% of the answers for your core prompts. On paper that sounds strong. But if a single rival appears in 85% of those same answers — usually named first, usually described more fully — your 60% is a second-place finish, not a win. Buyers reading those answers are forming a ranked impression, and you're losing it. Absolute visibility can't see that; it only counts your own appearances and stays silent about everyone else's.
The reverse is just as true. A 30% visibility rate looks weak in isolation, but if every competitor sits below 15%, you're the dominant voice in your space. The absolute number understated your position; the relative one corrects it. This is the core reason share-of-voice exists: visibility is only meaningful in context, and the context is who else got mentioned alongside you.
Relative position is also the signal that actually moves with the market. Your absolute visibility can hold steady while a rival quietly climbs past you — same number for you, worse standing in reality. Tracking share-of-voice means you notice that shift, because a competitor gaining ground shows up as your share eroding even when your own count never changed.
How to measure it
Measuring share-of-voice honestly takes a little discipline, because the number is only as trustworthy as the inputs behind it. The mechanics are the same probing approach you'd use for any AI-visibility metric — run a fixed set of prompts across engines on a schedule — with one addition: you also record every competitor mention, not just your own. The inputs that matter:
- A defined competitor set. Decide which rivals count before you measure; the share-of-voice number is meaningless without a fixed, named field to divide against.
- The same prompt set every run. Use one representative, unchanging set of prompts so the comparison stays apples-to-apples across time.
- Per-engine and aggregate views. Compute the ratio separately for each engine and as a combined figure, since the same prompt can favour different brands on ChatGPT, Claude, and Perplexity.
- A cadence over time. Re-run on a schedule and store the history, because a single reading is a snapshot and the value is in the movement.
This is the kind of measurement is built to run: it probes your prompt set across ChatGPT, Claude, and Perplexity on a schedule, records who gets mentioned in each answer, and computes share-of-voice per engine and in aggregate over time — using official engine APIs rather than scraping. However you assemble it, the goal is a stable, repeatable ratio you can trust to compare against itself week over week.
Read trends, not snapshots
A single share-of-voice figure is the least useful way to look at the metric. Answers vary by phrasing, by engine, and from one run to the next, so any one reading carries noise. Worse, one number flattens everything into a single point and hides the two things you most need to see: which direction you're moving, and which specific prompts are driving it.
The trend line is where the story lives. A share that's climbing means you're winning ground on rivals; one that's slipping is an early warning that someone else is being named more often, often before it shows up anywhere else. Watching the line over weeks separates a real shift from the normal flicker of a single probe run.
The per-prompt breakdown is the other half. An aggregate of 40% might be a healthy spread across every question, or it might be 80% on prompts that don't matter and near-zero on the ones that do. Only the breakdown tells you which. Read together, the trend and the breakdown turn share-of-voice from a vanity figure into a map of exactly where you stand and where you're exposed.
Moving your share
Improving share-of-voice is a competitive exercise, not a content checklist, because every point you gain comes from a prompt where a rival currently wins. The work is to find those prompts and understand why the other name shows up instead of yours.
- Find the prompts rivals dominate. Use the per-prompt breakdown to isolate the questions where a competitor's share is high and yours is low — those are your highest-leverage targets.
- Inspect the sources behind those answers. Look at which domains the engines cited when the rival was named. Those are the sources shaping the answer, and the places your absence is being decided.
- Act on the gap. Become the clearer, better-corroborated answer to those specific questions, and earn accurate presence on the sources those answers already trust.
- Re-measure against the same field. Re-run the prompt set and check whether your share moved relative to the same competitors, then repeat on the next set of gaps.
There's no trick that beats being a better answer than the rival who currently holds the prompt. Share-of-voice just makes the contest visible — it points you at the exact questions, competitors, and sources where the gap lives, so the work has somewhere honest to aim.
Frequently asked questions
How do I choose which competitors to track?
Pick the brands your buyers actually weigh you against — the names that show up on the same shortlists, not every company in the market. A tight, honest set of real rivals gives a meaningful share-of-voice; padding it with weak or irrelevant names inflates your share and tells you nothing useful. Revisit the set occasionally as your competitive landscape shifts.
Should I look per-engine or aggregate?
Both, because they answer different questions. The aggregate gives you an overall sense of standing, while per-engine views reveal where you're strong or weak — the same prompt can favour different brands on ChatGPT, Claude, and Perplexity. If you only watch the aggregate, you can miss being nearly invisible on one engine while leading on another.
Is 100% share-of-voice realistic?
Almost never, and chasing it usually isn't the right goal. AI answers tend to name several options, so a healthy category has share spread across a few credible players. A 100% reading is more often a sign of a too-narrow prompt set or competitor list than of true dominance. Aim to lead your real field on the prompts that matter, not to erase everyone else.
How is this different from SEO share-of-voice?
The idea is the same — your presence relative to competitors — but the surface differs. SEO share-of-voice is measured across ranking positions and impressions on search results pages. AI share-of-voice is measured inside generated answers: how often you're named versus rivals when an engine writes a response, across a prompt set rather than a keyword list. The unit shifts from a ranked position to a mention in synthesized prose.
How often does share-of-voice change?
It moves for two reasons: genuine shifts as you or competitors change, and the natural variance of AI answers between runs. That's exactly why you measure on a cadence and read the trend rather than a single point. Run-to-run wobble is normal; a sustained move across several runs is the signal worth acting on.
Know your share-of-voice.
Probe ChatGPT, Claude, and Perplexity with the prompts your buyers actually ask, and see how much of the conversation is yours versus your rivals — per engine, in aggregate, and over time. No vanity number; just where you stand and where to act.
Start tracking