# Incident runbook

Give the person on call a short path from a reported symptom to a verified recovery.

## Recognise the incident

## [Service] recovery
Owner: [team]
Last rehearsal: [date]

Start this procedure when [observable symptom] lasts [duration]. Check [dashboard] for [signal]. Escalate to [role] immediately if [condition].

## Restore service

1. Record the time and observed impact in [incident log].
2. Use [approved reversible action]. Expected result: [signal].
3. If the signal stays outside [range], call [on-call role] and share [diagnostic context].
4. Announce recovery in [channel] after [verification checks] pass.

## Close the loop

- [ ] Capture the first and last affected timestamps.
- [ ] Attach the dashboard view used to confirm recovery.
- [ ] Assign an owner and due date to the follow-up.
- [ ] Replace any instruction that was hard to execute during the incident.
