What does an organisation at maturity stage 3 (Observability) actually do differently?

Question

Accepted Answer

Metrics, logs and traces are integrated and correlated, so investigations look for the cause rather than the symptom — root-cause analysis, visible service dependencies and end-to-end transparency become routine.
The defining change is correlation. At stage 2 the three data types exist in three tools and three teams; at stage 3 they are joined — typically through a shared trace ID — so that moving from "the error rate rose" to "this service, this query, this exception" is a few clicks rather than a cross-team investigation.
What that unlocks:

Root-cause analysis instead of symptom management.
Service dependencies that are visible and derived from real traffic rather than from an architecture diagram someone drew two years ago.
End-to-end transparency for a request across every hop it takes.

The canonical example: a slow user request is followed API → service → database, and the concrete cause is identified at the end of that chain. Note that the organisation is still reacting — the investigation starts after something broke. Getting ahead of the problem is stage 4.