1) Name your rivals
Add the competitors buyers actually compare you with. The wizard suggests a starting list.
LLM share of voice
Your share of the answer, measured.
Share of voice was a media metric, then a search metric. Now the answer box is an AI assistant, and the question is what fraction of its recommendations are you. CitePrism computes share of voice from scheduled samples across seven engines, with your competitors tracked side by side.
- Share of voice per engine and overall
- Competitors tracked in the same runs
- First-mention rate and average rank
- Trends over weeks, not one-off snapshots
A fair fight, by construction
Every brand in your workspace is measured by the same prompts, the same schedule and the same detection logic. When the chart says a competitor overtook you, it is not an artifact.
Same questions for everyone
You and your competitors are scored against an identical prompt set, so comparisons hold.
Position matters
First name in the answer beats fifth. Rank and first-mention rate are tracked, not just presence.
Seven engines, one view
ChatGPT, Claude, Gemini, Perplexity, Google AI Overviews, Google AI Mode and Microsoft Copilot disagree more than you would expect. See where you win.
How it works
Live in minutes, honest forever.
2) Sample on schedule
Daily runs across the assistants build enough samples for percentages you can trust.
3) Read the race
Share-of-voice charts show who is gaining, engine by engine, with raw answers one click away.
FAQ
How is LLM share of voice calculated?
Of all brand mentions across sampled answers to your prompt set, share of voice is the fraction that are yours. CitePrism computes it per engine and overall, alongside visibility rate, average rank and first-mention rate, and every one of those definitions is documented on this site.
How many competitors can I track?
5 on Free, 25 on Pro. Each is detected in the same sampled answers, so adding a competitor does not multiply your costs. Most workspaces track three to eight.
Why does my share of voice differ between engines?
Because the engines genuinely differ: different training data, different retrieval, different citation habits. That disagreement is information. It tells you where to focus content and coverage work first.