How to Compete with Other Brands in AI Answers (Share of Voice)

Share of voice in AI answers is your brand's share of mentions versus competitors across AI-generated responses. Here's how to calculate it, separate position from SOV, and win contested answers.

Share of voice in AI-generated answers measures what percentage of brand mentions your company receives compared with all brand mentions across AI responses in your category. When a buyer asks an AI engine a category question — for example, "what's the best CRM platform?" — the engine names certain brands. Your share of voice is your total mentions divided by all brand mentions in those responses, expressed as a percentage.

For example: if ChatGPT, Perplexity and Google AI Overviews produce 100 brand mentions combined across your tracked category questions, and your brand appears 18 times, your share of voice is 18%. That metric shows whether you are winning AI visibility against competitors who are being cited instead — not whether you spent more on ads.

Your competition is defined per answer

Ask the same buyer question to ChatGPT, Perplexity and Google AI Overviews and you'll often get different brand lists. The brands and source domains an answer cites are your real competitors for that question — sometimes your classic rivals, often review sites, roundups or communities you didn't think of as competition. The first step is to stop guessing and read the actual cited set, per question and per engine.

How to calculate AI share of voice

The simple formula is:

(Your brand mentions ÷ total brand mentions across tracked prompts) × 100

Worked example: if you run 25 prompts and your brand is cited 20 times while all brands combined are cited 150 times, your share of voice is (20 ÷ 150) × 100 = 13.3%.

Use the same prompt library each time so the trend is comparable. A useful read is not just one blended number. Break it down by engine, prompt group (feature comparisons vs vendor discovery), competitor and position in the answer. That distribution shows which questions you own, which a rival owns, which are split, and where the answer is up for grabs.

Position is separate from share of voice

Share of voice is a simple count: your mentions ÷ total mentions. Position — whether you appear first, middle or later in an answer — should be recorded separately so you can see competitive placement. A brand can gain share of voice while still rarely leading the answer, or lead a few answers while remaining invisible on the broader prompt set.

Mention-based vs citation-based SOV

Mention-based share of voice counts how often your brand is named in AI answers compared with competitors. It is the broadest visibility metric. Citation-based share of voice counts how often your domain is linked or cited compared with other cited domains. Citation-based SOV is often the more actionable signal because it shows which sources the engine trusted enough to reference, while mention-based SOV tells you whether the brand made the shortlist.

Find the answers you can actually win

• Contested vs locked. A question whose cited set churns — many different brands have won it over time — is contestable. One where the same authoritative source always wins is locked; spend elsewhere.
• Where you've won before. Questions you were cited on previously are the clearest re-wins.
• Crowded but shallow. Answers naming many weak sources are easier to break into than ones naming one dominant authority.

Prioritising the contested, winnable answers — instead of attacking every gap equally — is how a smaller brand takes share from larger ones in AI answers.

Take a contested answer

For a winnable question, publish a clear, direct, well-structured answer with valid schema so the engine can quote you, and earn genuine mentions on the third-party sources it already cites for that question. Then re-ask on a cadence and watch whether your share of voice for that question actually rose. Which move shifts your standing has to be measured, not assumed — it varies by category and engine.

Measure share of voice over time

jujuGEO automatically probes the live engines (ChatGPT, Perplexity and Google AI Overviews, with Gemini coming soon) with buyer-intent questions in your category. For each query it records which brands are mentioned and their position, calculates your share of voice across tracked prompts, shows which competitors are cited when you are not, and identifies the exact source domains those engines trust. When you publish answer-ready content jujuGEO drafts, you can re-probe to measure whether your citation rate and share of voice improved. There is no universal "healthy" SOV target — it depends on how crowded the category answers are — so the useful comparison is your trend plus the specific competitors named instead of you.