AI Visibility Forecasting
Volatility as a condition worth tracking
How much answers move between identical requests is itself a measurement, and it changes what any single result is worth.
What volatility means here
Ask the same question repeatedly and the answer varies. Volatility is how much. It is a property of a category and a moment rather than of a business, and it is rarely reported by anything.
Why it matters more than it sounds
It sets the noise floor. In a high-volatility category, a change in a single reading means nothing. In a low-volatility one, the same change is a real signal. Without knowing which, a business cannot tell improvement from variance.
It indicates how settled a field is. High volatility usually means no option is clearly established, which is an opportunity: a settled answer is expensive to displace and an unsettled one is available.
It changes how often measuring is useful. Volatile categories need more samples per reading and fewer readings, which is the opposite of what most people do.
What raises it
Many near-equivalent options. Thin independent evidence across the whole field, so small differences reorder the result. Recent change in the underlying systems. And a question that is ambiguous enough to be interpreted differently between runs.
What it is not
Not a measure of whether a business is doing well. A business can be consistently named in a volatile category, which is a strong position, and inconsistently named in a stable one, which is a weak one. Volatility is context for a result rather than a result.
How we report it
As a category-level condition alongside any selection reading, with the sample size that produced it. A selection result reported without the volatility of its field is missing the information required to interpret it, and reporting a single sample as a fact in a volatile category is the most common way this whole measurement gets misused.