Escalation Policy Design: Three-Tier Pattern

Escalation policies have to be predictable and unambiguous. Three tiers are the minimum that handles real situations.

Why escalation matters

Single-tier paging fails the moment the on-call's phone is silenced or the engineer is asleep. Escalation is the safety net that turns a missed page into a delayed page rather than an outage.

• Without escalation. A missed primary page is the outage; nobody else hears about it until customers do.
• With escalation. Missed pages reach a human within minutes; the system fails forward, not silent.
• Trust failure. Phones die, batteries fail, alarms get muted; the policy assumes none of these are reliable.
• Audit proof. Compliance frameworks expect documented escalation; the trail is part of the evidence.

Three-tier pattern

• Tier 1: primary on-call. Pages immediately.
• Tier 2: secondary. Pages after N minutes no-ack.
• Tier 3: manager / broader team. Pages after secondary no-ack.

Per-team timing

One escalation cadence for every severity wastes higher tiers on minor issues. Per-severity timing keeps the policy useful and the secondary on-call sustainable.

• SEV1. T1 immediate, T2 at +5 minutes no-ack, T3 manager at +15 minutes.
• SEV2. T1 immediate, T2 at +15 minutes, T3 at +60 minutes; lower urgency, longer windows.
• SEV3 / SEV4. Single tier, business-hours only; no off-hours escalation.
• Severity drives cadence. One policy for all severities loses signal and overuses higher tiers.

No-ack defaults

The escalation must fire without human cooperation. At 3am, engineering judgement is unreliable; the policy compensates by escalating automatically.

• Auto-escalate. Page escalates after the no-ack window without anyone clicking anything.
• Engineer judgement removed. At 3am, the on-call should not have to think to escalate; the system does it.
• Manual override. On-call can suppress escalation if they have it under control; default is auto, override is explicit.
• Test quarterly. Synthetic page in the staging policy verifies escalation actually fires; do not assume.

Antipatterns

• One-tier escalation. Lost pages.
• Manual escalation. Forgotten in the moment.
• Same timing for all severities. Wastes higher tiers on minor.

What to do this week

Three moves. (1) Apply this practice to your next on-call rotation. (2) Survey the team after one cycle. (3) Iterate based on feedback; the discipline is the cadence.