Share of Voice Explained
What the share-of-voice numbers, bars, and bubble colors in Brand Radar actually mean — explained with worked examples, no formulas required.
"Share of voice" sounds like marketing jargon, but the idea is one sentence: when AI assistants talk about companies like yours, how much of that talk is about you?
A worked example
You track two competitors. In this week's AI Visibility check, across all the answers collected on the topic "site performance tools":
- Your brand was mentioned 6 times
- Competitor A was mentioned 3 times
- Competitor B was mentioned 1 time
Ten brand mentions total, six of them yours — your share of voice is 60%. You're winning the conversation on that topic. If next week you're at 40%, someone is taking mentions from you, and the Brand Radar views show who.
Reading the bars
The share-of-voice timeline shows one stacked bar per check. Each bar is the whole conversation for that check (100%), split into colored slices — your brand always in the same blue, each tracked competitor in its own color, and everyone you don't track lumped as "Other".
Two things to watch:
- Your slice growing or shrinking check over check — the trend matters more than any single bar.
- A fat "Other" slice — it means brands you don't track are getting mentioned a lot. Expand the topic to see who they are, and add real rivals to your competitor list.
Reading the bubbles
In the topic view, each bubble is a subject of conversation:
- Size = how much that subject comes up at all (everyone's mentions, not just yours). A big bubble is a big conversation.
- Color = the tone of your mentions there: green mostly positive, gray neutral or mixed, red mostly negative. The bar for green is deliberately high — a topic only turns green when positive mentions clearly outweigh negative ones, so green means genuinely good news.
The combination is what's useful: a big bubble where your share is small is the best opportunity on the page — lots of customer interest, little of it currently yours.
Honest limitations
- Small numbers wobble. 60% of ten mentions is meaningful; 60% of three is a coin flip. Check the underlying counts (click through to the answers) before celebrating or panicking.
- The questions shape the answer. Share of voice is measured on your question set. If you change the questions, the numbers shift for that reason alone — the chart marks the point where the question set changed so you don't misread it as a real gain or loss.
- Per-platform differences are real. A competitor can own one AI assistant while you lead on another; the competitor matrix in Brand Radar breaks this out.