How to define a brand's share of voice in AI answers

Ask ChatGPT which running shoe brand it would recommend, or which air purifier to buy, and the answer won't list the whole market. It will name two, three, maybe five brands. Everyone else is invisible to that shopper at that exact moment.

That's what share of voice in AI answers is meant to capture: the proportion of generated answers that mention your brand, against the ones that mention your competitors. It's a metric borrowed from traditional marketing, but it doesn't get calculated the same way anymore.

What share of voice means in AI answers

Concretely, AI share of voice measures how often your brand shows up across answers from AI engines (ChatGPT, Gemini, Perplexity, Google AI Overviews, Google AI Mode) on a set of queries representative of your category, relative to how often competitors show up on those same queries. It isn't a brand awareness score, and it isn't a keyword ranking. It's a measure of relative presence inside a conversation, engine by engine.

Two types of queries matter, and they tell different stories. Branded queries ("What's [your brand] like?") confirm what you already know. Unbranded, need-based queries ("Which insulated water bottle actually lasts?") reveal your real competitive share of voice, in a context where nobody typed your name at all.

What a reliable measurement framework needs to cover

A share-of-voice number built from a handful of spot checks doesn't mean much: AI answers shift from one day to the next, from one engine to another, and from one phrasing of the same question to the next. For the number to be usable, a few conditions have to hold.

• Cover unbranded queries that reflect real questions asked in your category
• Track each engine separately: a strong showing on ChatGPT says nothing about your standing on Perplexity or Gemini
• Watch it change over time, not a single snapshot
• Separate a plain brand mention from an actual product recommendation
• Read citation frequency alongside position in the answer and the tone used

Why share of voice alone isn't enough

Being mentioned often doesn't guarantee being mentioned first, or described favorably. Two mattress brands can both show up in a large majority of answers to "best mattress for back pain," yet one opens the answer with a glowing tone while the other gets tacked on after three competitors. Identical share of voice, a very different commercial reality. The metric shows scale; it needs average position and sentiment next to it to show the full picture.

Building this tracking by hand, engine by engine, query after query, isn't realistic over time. That's exactly what Tadow's MAP pillar covers: measuring your brand's share of voice across the AI engines that matter for your market, over time, so you know where to act first.