What is the Apdex Score?

Industry convention: 1.0 is perfect, 0.94+ excellent, 0.85-0.94 good, 0.70-0.84 fair, 0.50-0.69 poor, below 0.50 unacceptable. Scores depend entirely on the threshold you pick for T — a 500ms T on an API and a 2s T on a dashboard produce very different numbers for the same underlying latency.

T should represent the latency under which users feel the app is responsive. For web APIs, 200-500ms. For interactive dashboards, 1-2s. For report generation, 5-10s. Pick per endpoint class, not globally — a single T across all routes muddies the signal.

It's fallen out of favor compared to raw percentiles (p95, p99) because Apdex compresses nuance into one number. It's still useful as a single-digit executive summary ('site health = 0.91') and for historical comparison. Most modern APMs show both Apdex and percentiles side by side.

p95 tells you the latency for the 95th-percentile request — an absolute number in milliseconds. Apdex tells you the fraction of requests that were acceptable, weighted by a tolerance threshold. p95 gives you precision; Apdex gives you a comparable score across services. They answer different questions.

Apdex was invented in 2004 by the Apdex Alliance (NewRelic was a founding member) to give executives a simple 0-1 score for application quality. Before Apdex, APMs showed raw latency histograms — accurate but hard to compare across services or communicate to non-engineers.