Skip to main content

Incident Runbook Template

TemplateWritten by Burak YReviewed by Burak YLast reviewed 3 September 2026

Use this for any alert that pages a human, starting with the alerts that fire most often and the services whose failure has the widest impact.

Placeholders

Everything in [square brackets] is a decision your team has to make. Replace all of them before you circulate the document.

A runbook is not documentation. It is a decision aid for someone who was asleep eight minutes ago and may not own the service that woke them.

What this template covers

Eight sections: trigger, first five minutes, safety boundaries, diagnosis, mitigation options with rollback, escalation, communication, and resolution with follow-up.

The section people skip

Safety boundaries — what a responder must not do without approval. It is usually missing, and it is usually the part that prevents a manageable incident from becoming a data-loss incident. Write it before you write the diagnosis section.

Related

Template

Incident Runbook

A runbook should help a tired responder make the next safe decision, not document every possible detail.

1. Trigger

  • Alert name: [name]
  • Service: [service]
  • Severity: [severity]
  • What this alert means: [plain-language explanation]
  • Common false-positive conditions: [conditions]

2. First five minutes

  1. Confirm [signal] using [dashboard/query].
  2. Check [dependency/region/recent change].
  3. Determine customer/business impact using [source].
  4. If [condition], perform [safe mitigation].
  5. If impact meets [criteria], declare/escalate the incident.

3. Safety boundaries

Do not:

  • [unsafe/destructive action]
  • [action requiring approval]
  • [action without backup/verification]

Approval required from:

  • [role/team]

4. Diagnosis

Key dashboards:

  • [link/name]

Key logs/queries:

  • [link/query description]

Recent changes:

  • [deployment/config source]

Dependencies:

  • [dependencies]

5. Mitigation options

OptionUse whenRiskRollback
[option][condition][risk][rollback]

6. Escalation

Escalate when:

  • Impact is unknown after [duration]
  • Required access is unavailable
  • Mitigation has significant data/customer risk
  • The issue crosses team ownership
  • The responder is unsure or overloaded

Escalation target: [policy/team]

7. Communication

  • Incident channel: [location]
  • Status update owner: [role]
  • Update cadence: [duration]
  • Customer communication trigger: [criteria]

8. Resolution and follow-up

Resolution evidence:

  • [metrics/tests]

After resolution:

  • Remove temporary mitigation if safe
  • Re-enable alerts/suppression as appropriate
  • Record timeline and actions
  • Create follow-up items
  • Update this runbook

Owner: [owner] Last tested: [date] Last reviewed: [date]