How do you measure competitor share of voice in AI answers?

What is share of voice in AI answers?

It is the percentage of sampled AI answers that name a given brand, across the questions and engines you track.

Classic share of voice counts ads, press or social posts. In AI answers the unit is a single response to a single question from a single engine. If you track 10 questions in six engines, you have 60 answers a week; a competitor named in 24 of them has a 40% share for that week.

Because several brands usually appear in one answer, shares do not add up to 100%. That is fine. What you compare is each brand's share against yours on the same questions.

How do you calculate competitor mention rates across LLMs?

Count, for each brand, how many answers name it, then divide by all answers; repeat per engine.

Per-engine rates matter because engines disagree. A competitor can dominate one assistant's answers and barely appear in another's, which usually means different sources are feeding each engine.

• Use the same questions and engines every week.
• Count a brand once per answer, however many times it is repeated.
• Track 'named without you' separately: it is the displacement number.
• Look at four or more weeks before drawing conclusions.

How does MarketHQ show competitor share of voice?

The AI Visibility page ranks the brands AI names most often in your space and spells out where competitors appear without you.

On paid plans MarketHQ asks your tracked questions in ChatGPT, Claude, Gemini, Perplexity, Grok and DeepSeek every week. The AI Visibility page states how many sampled answers named you, then shows a 'Who AI recommends in your space' list: the brands named most often in this week's answers, with the competitors you track marked. Brands you do not track yet can appear there too when MarketHQ recognizes them, which is a quick way to find a rival you missed.

For the tracked competitors that appear most often, MarketHQ writes a plain line saying, for example, that Competitor A was named in 6 answers this week, including 2 where you weren't. Each answer row shows the engine, the question, whether you were named and which competitors were named, with the full text one click away.

The same data is available to your own tools: paid plans include a REST API and an MCP server with an AI-visibility call that returns your share against tracked competitors.

What should you do when a competitor's share is higher than yours?

Find the questions where they win without you, and work on those questions one at a time.

A competitor's overall lead is usually concentrated in a few questions. Open those answers and read how the competitor is described and which pages, if any, the answer cites. Then decide whether the fix is a clearer page of your own, a listing on a comparison page the engines cite, or a presence in the community threads where the question comes up.

MarketHQ's weekly gap analysis does this sorting for you and turns lost questions into an action plan.

Which engines should you include?

The ones your buyers use; for software buyers that usually means ChatGPT, Claude, Gemini and Perplexity at least.

MarketHQ covers ChatGPT, Claude, Gemini, Perplexity, Grok and DeepSeek. It does not cover Google AI Overviews, Google AI Mode or Microsoft Copilot, so a share-of-voice figure from MarketHQ describes those six engines only.

What does a share-of-voice finding look like?

Illustrative example: one competitor wins every 'open source' question.

Illustrative example: a log-management company tracks 20 questions and three competitors. Its share is close to the leader's overall, but on the five questions that include 'open source' one competitor is named in nearly every answer and the company in none. The company's open-source edition exists but is buried on a pricing page. It publishes a dedicated page and watches those five questions over the next month.

What are the limits of AI share of voice?

It depends entirely on your question set, and it samples answers rather than counting every conversation.

Choose questions that favour you and your share will look great while buyers see something else. Pick questions with sales and support, include ones you currently lose, and keep them stable. Shares also move a few points week to week without any real change, so react to trends rather than single readings.