Escalation Policies

Define who gets notified, when, and what happens if they don't respond.

---

How Escalation Works

When an incident triggers, NotifyHero walks through your escalation policy level by level:

1. Level 1 — notify immediately (e.g., on-call engineer)
2. Wait for acknowledgment
3. If no response after the timeout → Level 2 (e.g., team lead)
4. Continue until someone acknowledges or all levels are exhausted

Level 1: On-Call Schedule "Backend Primary" → wait 5 min
Level 2: On-Call Schedule "Backend Backup" → wait 10 min
Level 3: User "vp-engineering@your-company.com" → wait 15 min
Level 4: Repeat from Level 1

---

Creating an Escalation Policy

Go to Settings → Escalation Policies → New Policy.

Each level requires:

• Target — a schedule, a specific user, or a team
• Timeout — minutes to wait before escalating (0 = notify simultaneously with next level)

Targets

Target Type	Description
Schedule	Notify whoever is currently on-call
User	Notify a specific person
Team	Notify all members of a team

---

Multi-Level Escalation

Add as many levels as needed. Common patterns:

Standard (3-level)

L1: Primary on-call → 5 min
L2: Backup on-call → 10 min
L3: Engineering manager → 15 min

Aggressive (for critical services)

L1: Primary on-call → 2 min
L2: Primary + Backup on-call → 3 min
L3: Entire team → 5 min
L4: VP Engineering → 10 min

Flat (small team)

L1: Entire team → 0 min (everyone notified immediately)

---

Repeat Behavior

At the bottom of the policy, choose what happens after all levels are exhausted:

• Repeat from Level 1 — loops back (recommended for critical services)
• Stop escalating — no further notifications
• Repeat N times — loops a set number of times, then stops

Tip: For production services, always enable repeat. An unacknowledged critical incident should never go silent.

---

Urgency Rules

Set urgency based on severity or time of day:

By Severity

• Critical / Error → High urgency (phone + SMS)
• Warning / Info → Low urgency (email + Slack)

By Time

• Business hours (09:00–18:00) → Low urgency
• Off-hours → High urgency

Combined

• Critical during off-hours → High urgency, aggressive escalation
• Warning at 2 PM → Low urgency, standard escalation

---

Fallback Behavior

If an incident goes unacknowledged through all escalation levels and all repeats:

1. The incident is marked as Escalation Exhausted
2. Optionally notify a fallback target (e.g., a Slack channel or email alias)
3. The incident remains open and visible on the dashboard

Configure fallback at Escalation Policy → Advanced → Fallback Target.

---

Assigning to Services

Each service has exactly one escalation policy. Assign it at Service → Settings → Escalation Policy.

Multiple services can share the same policy. For example, all services owned by the Backend team can use the "Backend Escalation" policy.

---

Best Practices

• Keep timeouts short for critical services — 2–5 minutes per level
• Use schedules, not users — individual users go on vacation; schedules don't
• Test your policies — trigger a test incident and verify the escalation chain works
• Review quarterly — teams change, so should your policies