Definition

Answer volatility is how much AI answers change across time, prompts, surfaces, or repeated runs.

Expanded definition

Volatility is expected in AI answer systems. Teams should look for durable patterns rather than treating every single answer as a stable result. As a metric, answer volatility should be read as a diagnostic signal rather than a standalone KPI. The next step is to connect the number to a specific answer pattern, source issue, buyer risk, or content decision.

Why it matters

Without accounting for volatility, teams may overreact to one bad answer or overstate one good answer.

Example

The brand appears in a recommendation answer on Monday but not on Wednesday for a similar prompt.

Common mistake

Making content decisions from one prompt run without repeat checks.

Diagnostic question

Is this answer pattern repeated enough to justify action, or should it be monitored first?