The simple rule
Start at the lowest level you haven’t mastered, then move up.
Common starting points
| Your situation | Start here |
|---|---|
| New to observability | Level 1 — Integrations |
| Have Prometheus but no tracing | Level 2 — Application Observability, Knowledge Graph, RCA Workbench |
| Have tracing but struggle to debug slow requests | Level 3 — Continuous profiling |
| Need custom metrics for business KPIs | Level 4 — OpenTelemetry custom instrumentation |
Decision tree
Script
We’ve covered a lot of ground. Let’s wrap up with the practical question: where should you actually start?
Here’s the simple rule: start at the lowest level you haven’t mastered, then move up. Don’t skip ahead.
Ask yourself these questions. Can you see infrastructure metrics and logs today?
If not, you’re at Level 1. Start there.
If yes, can you identify which service is causing problems when something breaks?
If not, focus on Level 2.
If yes, can you trace why specific requests are slow?
If not, focus on Level 3.
If yes, you’re ready for Level 4: custom instrumentation.
Here are some common starting points. If you’re new to observability entirely, start with Level 1 and the integrations course.
If you have Prometheus but no tracing, focus on Level 2 and Application Observability, Knowledge Graph, and the RCA Workbench.
If you have tracing but struggle to debug slow requests, start with Level 3 and continuous profiling.
If you need custom metrics, choose Level 4 and OpenTelemetry.
The decision tree on this slide makes it visual. Follow the questions to find your starting point.