Concepts
How Escalation Policies Work
Understand how MonitoringDog routes incident alerts through ordered steps, contacts, and repeat cycles.
An escalation policy is a named, ordered list of steps attached to a check. When an incident opens, the policy fires automatically — no manual intervention required.
How a Policy Fires
- Incident opens → Step 0 fires immediately.
- If the incident is not acknowledged, Step 1 fires after its
stepDelay. - This continues through all steps in order.
- If
repeatCount > 0, the entire sequence repeats up to that many times, withrepeatDelaybetween each repeat.
Escalation stops as soon as the incident is acknowledged.
Step Configuration
Each step in a policy defines:
| Field | Description |
|---|---|
stepDelay | Seconds to wait after the previous step before this one fires |
contacts | Who to notify: on_call (current on-call member), entire_team, or a specific member |
severity | Which severity level to use — controls which notification channels fire |
Policy-Level Settings
| Field | Description |
|---|---|
repeatCount | How many times to repeat the full step sequence if never acknowledged |
repeatDelay | Seconds to wait between repeats |
resolutionAlerts | Which channels fire when the incident is resolved (defaults to email, push notification, integrations) |
Example Flow
A policy with two steps and one repeat:
Incident opens
→ Step 0 fires (email only)
→ [30s delay]
→ Step 1 fires (SMS + phone call)
→ [5min repeatDelay]
→ Step 0 fires again (repeat 1 of 1)
→ [30s delay]
→ Step 1 fires again
→ Policy exhausted — no more escalationIf a team member acknowledges at any point, escalation halts.