How well you can infer a system's internal state and diagnose problems from its existing logs, metrics, and traces — without having to add new instrumentation on the spot.
Example: Diagnosing an incident from dashboards instead of guessing.
In practice: The three pillars — logs, metrics, and traces — each answer a different question: what happened, how much or how often, and where time was spent across a request.
Debated: Increasingly criticized as a vague vendor buzzword — critics argue 'observability' often just means 'we bought a logging, metrics, and tracing tool,' without the org actually getting any better at diagnosing incidents.