# Share of voice in AI answers

> Share of voice in AI answers is the number of times a brand is named across a set of AI answers, divided by the number of times any brand is named across the same answers.

Canonical page: https://llmaudit.app/learn/share-of-voice-in-ai-answers

Share of voice in AI answers is the number of times a brand is named across a set of AI answers, divided by the number of times any brand is named across the same answers. It is a relative measure: it says how much of the assistants' attention a brand gets compared with everyone else named for the same questions, not whether the brand is named often in absolute terms.

## How to compute it from counts

Start from the same set of answers for every brand. Count how many times each brand is named. Add those up to get the total of brand mentions. Divide one brand's count by the total. As an illustration: if across 12 answers brands are named 40 times in total and yours is named 8 times, your share of voice is 20 percent. The denominator has to come from the same answers as the numerator; mixing runs or questions makes the number meaningless.

## The single-run trap

Share of voice looks precise, so it is tempting to compute it from one run per question. In our repeatability study, half of the brands named across five runs of the same question appeared only once. A brand that shows up in one run and not the next four is not a stable share of anything; it is a roll of the dice that happened to land. Share of voice from one run is therefore mostly noise, and the rank order of brands below the first one or two changes from run to run.

## How LLM Audit reports it

The audit reports counts with their total, such as "recommended in 3 of 12 open questions", and the competitors named most often across the same answers, so a share can be computed from the same denominator. It says how many runs each count came from, and it does not put a share of voice in the headline from a single run. The repeatability numbers behind this are on the research page.

## Measured

**Half**: of the brands named across five runs of the same question appeared in only one of the five runs. Source: https://llmaudit.app/research/how-repeatable-are-ai-recommendations

## FAQ

### Is share of voice the same as AI visibility?

No. AI visibility is how often a brand is named across the buyer questions. Share of voice is that count relative to everyone else's. A brand can have low visibility and high share of voice in a category where assistants name few brands at all.

### What is a good share of voice?

There is no general number. It depends on how many brands the assistants name for that category and on the questions chosen. Compare against the same questions over time, not against a benchmark from another category.

### How many runs before it is stable?

More than one. Three runs per question show whether a brand is named always, sometimes or never; resolving small differences in share needs many more. See the page on how many runs.

## Related

- [How many runs](https://llmaudit.app/learn/how-many-runs.md)
- [AI visibility](https://llmaudit.app/learn/ai-visibility.md)
- [Mention vs citation](https://llmaudit.app/learn/mention-vs-citation.md)
- [How repeatable are AI recommendations](https://llmaudit.app/research/how-repeatable-are-ai-recommendations.md)
- [All definitions](https://llmaudit.app/learn.md)

---

Machine-readable index: https://llmaudit.app/llms.txt · Full context: https://llmaudit.app/llms-full.txt · All URLs: https://llmaudit.app/sitemap.xml · Developer and agent resources: https://llmaudit.app/developers.md