What is share of voice?
Share of voice (SoV) is a metric that expresses a brand’s share of the total presence of all brands in a market within one channel, as a percentage. Share of voice originally measured a brand’s share of the advertising pressure in its industry; today the metric is also calculated for search results, social media and AI answers. In AI answers, share of voice counts how often a language model names a brand relative to its competitors, measured across a fixed list of questions.

Term profile
| Attribute | Detail |
|---|---|
| Part of speech | Noun, marketing term |
| Abbreviation | SoV or SOV |
| Pronunciation | ʃɛər əv vɔɪs |
| Also known as | Share of advertising pressure |
| Related terms | Share of market, share of search, excess share of voice, brand mentions |
| Field | Marketing measurement, SEO, GEO |
How do you calculate share of voice?
Share of voice is your own presence divided by the presence of all brands in the set you are looking at, multiplied by 100. The channel defines what counts as presence: In advertising, it is media spend or ad impressions, in social media it is mentions, in search it is estimated clicks, and in AI answers it is the number of times a brand is named.
Formula: Share of voice in percent = presence of your brand ÷ presence of all brands in the set × 100
A worked example for AI answers: A measurement asks 40 questions of 5 AI systems and receives 200 answers. Those 200 answers contain 500 brand mentions in total, and 60 of them go to your brand. Your share of voice is therefore 12 percent.
The denominator decides the result. Share of voice is always a share of a set that someone has defined. That set can be specific competitors, specific keywords or specific questions. Shrinking the set raises your own share although nothing has moved in the market. A share of voice value can therefore only be read together with its set, and two values are only comparable when the set stays the same.
What types of share of voice are there?
There are 4 types of share of voice, one per channel: advertising share of voice, social share of voice, SEO share of voice and AI share of voice. The formula stays the same, only the counted quantity changes.
| Type | What is counted | Typical data source |
|---|---|---|
| Advertising share of voice | A brand’s media spend or ad impressions relative to its industry | Media planning, in Google Ads the share of eligible impressions |
| Social share of voice | Mentions of the brand in social media, forums and the press | Social listening and media monitoring tools |
| SEO share of voice | Estimated organic clicks across a keyword set | Rank tracking: search volume times the expected click-through rate of the position |
| AI share of voice | Mentions of the brand in answers from ChatGPT, Gemini, Perplexity and Google | Prompt tracking across several AI systems |
For advertising share of voice, Google Ads reports the value directly as the share of eligible impressions. SEO share of voice weights every position by search volume and the expected click-through rate (CTR), so that position 1 for a popular keyword counts more than position 8 for a niche term. Social share of voice is based on brand mentions, meaning mentions of the brand with or without a link.
A closely related metric is share of search. It describes the share of searches for one brand among all brand searches in a category. The British effectiveness researcher Les Binet presented the metric at the IPA’s EffWorks conference in 2020, because it gives an early reading of a brand’s market share.
How do you measure share of voice in AI answers?
Share of voice in AI answers is measured in 5 steps: Define the questions, define the brands, query all AI systems, measure repeatedly and count the mentions. This metric shows how often a brand is named in AI search when people ask for a provider.
- Define the questions: The prompt set contains selection questions (“Which agency for X is good?”), comparison questions (“A or B?”) and brand questions. Every prompt keeps its exact wording, because a single different word produces a different answer. Choosing the right questions is the part that decides what the measurement is worth.
- Define the brands: The competitor set names the brands whose mentions are counted, each with all its spellings. Your own brand appears in answers as a short name, with its legal form or as a domain.
- Query all AI systems: ChatGPT, Gemini, Perplexity, Google AI Overviews and Google AI Mode answer the same question with different brands. A measurement based on a single system describes only that system.
- Measure repeatedly: Language models answer the same question differently from one run to the next. Share of voice becomes reliable only as an average across many measurement days.
- Count the mentions: Simple share of voice counts every mention equally. It gets more precise when the position within the answer and the number of your own pages cited as sources are factored in.
Step 5 yields 3 variants of AI share of voice: The mention share counts how often the brand appears, the position-weighted share looks at whether it stands at the top or at the end of a list, and the source share measures how often the model links one of the brand’s pages as evidence. The source share often explains the mention share: Models prefer to name brands whose pages they use as sources.
Which errors distort share of voice in AI answers?
4 errors distort share of voice in AI answers especially often: Generic terms used as brand names, name variants counted separately, a single AI system as the measurement base and imprecise prompt wording. We ran into all 4 errors in our own measurement in September 2026, across 64 prompts in 5 AI systems.
- Generic term as a brand name: If a competitor brand carries a title such as “SEO” or “GEO agency” in the measurement tool, every answer that contains this word counts as a mention of that brand. The competitor’s share of voice rises artificially, and everyone else’s falls. A brand therefore always belongs in the set under its proper name or its domain.
- Name variants counted separately: “taismo”, “taismo GmbH” and “taismo.de” are three different brands for a measurement tool until someone merges them as aliases of the same entity. The brand’s share of voice then splits into three small values.
- Only one AI system queried: The same selection question about a GEO agency returned position 2 for taismo in Google AI Mode on September 19, 2026, and no mention at all in Perplexity. Anyone who measures only one of the two systems ends up with one of two opposite results.
- Imprecise prompt wording: The wording of a prompt is the unit of measurement. A German question typed with transliterated umlauts, or a query run in a logged-in account with its own history, measures something other than the question a prospect actually asks.
On top of that comes a limit of the method: Share of voice in AI answers cannot be collected retroactively. A model answer exists only if it was stored at the moment the question was asked. The measurement series starts on the first measurement day.
What is the difference between share of voice and a visibility index?
Share of voice is a relative share within a competitive set you choose yourself; a visibility index is an absolute value that an SEO tool calculates across its own fixed keyword set. The share of voice values of all brands in one set add up to 100 percent. A visibility index has no such upper limit.
| Attribute | Share of voice | Visibility index |
|---|---|---|
| Type of value | Share in percent | Index number |
| Reference base | All brands in the chosen set | The tool provider’s keyword set |
| Who defines the set | You | The tool provider |
| Channels | Advertising, social media, search, AI answers | Organic Google search |
The two metrics answer different questions. The visibility index shows how a domain’s presence in Google develops over time. Share of voice shows how much of the competition for exactly the searches or questions that matter to your business goes to your brand. If the visibility index rises while share of voice falls, the competitors in your own market are growing faster.
What does a high share of voice tell you?
A high share of voice shows that a brand is more present than average in its market, and it is considered an early indicator of market share. In 1990, the advertising researcher John Philip Jones described in the Harvard Business Review that brands with a share of voice above their market share tend to grow. Marketing research calls this surplus excess share of voice (ESOV). Les Binet and Peter Field derived a rule of thumb from the effectiveness database of the British advertising association IPA: 10 percentage points of ESOV go along with roughly 0.5 percentage points of market share growth per year.
In AI answers, share of voice carries an additional meaning. For selection questions, language models usually name 3 to 8 providers. A brand outside that list is missing at the moment a prospect builds a shortlist. AI share of voice therefore measures how often a brand makes it onto that list. A GEO audit collects this value across all AI systems and shows which sources the models cite for their lists of providers.
Frequently asked questions about share of voice
What is a good share of voice?
A good share of voice is above your own market share. A fixed percentage is of little use as a benchmark, because share of voice depends on the chosen competitor set. If share of voice is above market share, that points to growth; if it is below, the brand tends to lose ground.
Can you measure share of voice in AI answers retroactively?
You can measure share of voice in AI answers only from the day the measurement starts. A model answer exists only once someone asks the question and stores the answer. For search results, however, share of voice can be recalculated from an existing ranking history.
How often should you measure share of voice?
For AI answers, a daily measurement makes sense, because language models answer the same question in different ways. Then evaluate the average over several weeks and never a single measurement. For advertising and SEO share of voice, a monthly rhythm is enough.