Start from an operational question

Useful telemetry helps answer questions such as whether requests are failing, which dependency is slow, and whether a recent deployment changed the pattern.

Logs provide event detail, metrics show aggregate behavior over time, and traces connect work across service boundaries. Together they help narrow an investigation.

Keep signals actionable

A metric without a threshold or an owner may add noise instead of clarity. Alerts should point to a user-impacting condition and give responders enough context to begin diagnosis.

Review telemetry after incidents and remove signals that do not help explain or prevent a recurrence.