Metrics are the cheapest, fastest signal Datadog has, right up until custom metric cardinality turns them into the most expensive line on the invoice. Understanding how metric collection works, and how custom metrics are billed, is the difference between a metrics estate that scales and one that gets audited in a panic at renewal.
Start with how Datadog collects metrics: agent checks, integrations, DogStatsD and the API. Then decide what to track. Five metrics cover most of what a small team needs to know about system health, and starting there beats instrumenting everything.
Custom metrics are where the power and the cost both live. Every unique combination of metric name and tag values is a billable series, which is why a single high-cardinality tag can multiply spend overnight.
When metrics misbehave, the causes are predictable: metrics that stop reporting, custom metrics that never appear, and API submissions that fail silently. Each has a short checklist, and working through it beats guessing.