Published by Gist GEO · About our content
How it is worked out
Take a fixed set of questions that buyers in your category ask. Ask them, more than once. Count every brand that gets recommended. Your share of voice is your count divided by the total.
A simple example, with made-up numbers:
| Brand | Times recommended | Share of voice |
|---|---|---|
| Brand A | 80 | 40% |
| Brand B | 60 | 30% |
| Your brand | 30 | 15% |
| Everyone else | 30 | 15% |
| Total | 200 | 100% |
What it is good for
Share of voice turns a fuzzy question, "does AI like us?", into something you can track. It is most useful in two ways:
- Next to the leader. The gap between you and the top brand is easier to act on than your own number alone.
- Over time. The same questions, asked again next month, show whether anything you did made a difference.
What it does not tell you
It does not say why. A low share of voice could mean AI can't read your site, or that the websites it relies on don't mention you. For that you need the citations.
It also says nothing about accuracy. You can have a healthy share of voice and still be described wrongly.
Questions marketers ask
What is a good AI share of voice?
It depends on how crowded your category is. In a category with three serious brands, 15% is weak. In one with forty, it may put you near the top. Compare yourself to the category leader, not to a fixed number.
Why does my share of voice change between tools?
Because the question sets differ. Share of voice is only comparable when the same questions are asked the same way. Check which questions a tool uses before you compare numbers.