What is a good share of voice?
Short answer: It depends on how many competitors you track, but as a rule of thumb anything above your fair share (100% divided by brands tracked) means you are winning.
A good share of voice is one above your "fair share": 100% divided by the number of brands in the comparison. If you track yourself plus four competitors, fair share is 20%, so 30% means the assistants favour you and 10% means they favour someone else.
Share of voice (SoV) in AI answers is your mentions divided by all mentions of you and your tracked competitors, across every prompt, engine and sample in a period. It is the generative equivalent of market share, and it replaces "rank number one" as the headline metric because an assistant's answer has no ranks, only a short list of names.
Three things change what "good" means for you:
- Market structure. In a category with a dominant national brand, a regional firm at 15% against that brand may be doing very well.
- Prompt mix. Branded prompts ("is Acme any good") will naturally give you a high SoV; unbranded ones ("best accountant in Hull") are the real test. Look at them separately.
- Engine. It is common to have 40% on Perplexity and 5% on ChatGPT for the same prompts, because each leans on different sources.
Rather than chasing a single number, watch three things: SoV trend over eight to twelve weeks, SoV on unbranded prompts only, and the gap between your best and worst engine. The last one tells you where the quickest win is.
Seoptist reports SoV per engine and overall against the competitors you choose (3 on Starter, 8 on Growth, 10 on Agency) and averages three samples per prompt so a single unusual answer does not swing the figure.