What is ai share of voice
Share of voice has existed in communications for a long time: it measures a brand’s weight in a conversation, relative to its competitors. Applied to generative AI, the idea stays the same. You ask ChatGPT, Perplexity, or Gemini a set of questions representative of an industry, and count how many times each brand shows up in the answers.
The principle is simple. The calculation is less so, because you need to define a consistent query scope before producing any percentage at all.
Why this number matters more than a ranking
A ranking tells you who comes first. Share of voice tells you how much space each brand actually occupies in the conversation. A brand can be cited first on one query and absent on ten others. A ranking looked at alone shows none of that.
For a marketing leadership team, that’s the difference between “we’re visible” and “we’re visible on 8% of the queries that matter, versus 34% for our main competitor.” The second number holds up in front of a leadership committee. The first one doesn’t.
AI share of voice is never read on its own. A score of 20% can be excellent in a market with ten players, or disastrous in a market with three. It always needs to be compared against a specific set of competitors, over the same batch of queries.
How to calculate your ai share of voice
The method comes down to three steps.
- Build a batch of queries representative of your industry, phrased the way a customer actually would
- Submit these queries to the major models and log every brand mention, yours and your competitors’
- Divide the number of mentions of your brand by the total number of mentions logged across all brands
| Brand | Mentions logged | Share of voice |
|---|---|---|
| Brand A (you) | 18 | 36% |
| Competitor B | 22 | 44% |
| Competitor C | 10 | 20% |
This simplified table hides a real difficulty: AI answers change from one day to the next, from one model to another, sometimes even from one session to the next for the same question. A one-off measurement isn’t worth much. It’s tracking over time that produces a reliable number, not a single snapshot.
Concrete cases to understand the use
A sporting goods retailer might discover it dominates generic questions (“which running brand should I choose”) but disappears entirely on more specific ones (“best trail running shoe for mountain terrain”). Its overall share of voice hides that gap. Breaking it down by topic reveals it.
A B2B software company might notice its share of voice climbs on English-language queries but stays low in French, even though its main market is French-speaking. In that case, the problem isn’t the brand, it’s the nature of the sources the AIs draw on to answer in French within that industry.
These gaps only show up when you segment queries. An average share of voice, calculated over too broad a batch, flattens everything and hides exactly what most needs fixing first.
Pitfalls to avoid
The first pitfall is comparing your share of voice to a competitor that has nothing to do with your actual market. A premium brand doesn’t need to worry about its share of voice against a discount player, even if both technically sell the same broad product category.
The second pitfall is stopping at an overall number without digging into how competitors actually show up on each type of query. A decent average score can hide a total absence on the queries that drive the most purchase decisions.
The third pitfall is treating share of voice as a standalone number, when it only really makes sense as part of a broader set of visibility indicators tracked over time. A number with no context or trend line says nothing about a brand’s trajectory.
AI share of voice is not a vanity metric. It’s a steering number, provided you calculate it over a consistent query scope and track it over time rather than measuring it once for reassurance. The brands making progress here are the ones segmenting their share of voice by topic and by persona, not the ones settling for a single overall percentage reported once a quarter.
Questions related to this article
How do you calculate your AI share of voice?
By counting mentions of your brand across a batch of representative queries, then dividing by the total number of mentions logged, including competing brands.
Does AI share of voice replace ranking in answers?
No, it complements it. Ranking shows who appears first on a given query, share of voice shows a brand's overall weight across an entire industry.
Do you only need to measure share of voice once?
No. AI answers vary over time and across models, so a one-off measurement doesn't reflect a reliable trend.
