Apdex (Application Performance Index) is a standardized method for measuring and reporting user satisfaction with application response time. It produces a score between 0 (completely frustrated) and 1 (fully satisfied) based on a configurable threshold T representing the "satisfying" response time.
The Apdex formula classifies each request into three categories: Satisfied (response time ≤ T), Tolerating (T < response time ≤ 4T), and Frustrated (response time > 4T or error). The score is calculated as: Apdex = (Satisfied + Tolerating/2) / Total. For example, with T=500ms, a response time of 300ms is satisfied, 1500ms is tolerating, and 2500ms is frustrated.
Apdex is useful as a high-level summary metric that is easy to communicate to non-technical stakeholders: "Our Apdex is 0.95" is more intuitive than "Our P95 latency is 450ms." However, it should be used alongside percentile metrics for detailed analysis, as Apdex can mask the severity of tail latency issues by grouping them into the "frustrated" bucket regardless of how slow they actually are.
