The Incident Escalation Tree

When an incident escalates, who do you page next? The tree that defines.

Levels

The escalation tree defines who gets paged when. Each level has a clear commitment so escalation is fast and predictable.

Primary on-call. First responder per rotation. Acks within the team’s MTTA target.
Secondary on-call. Named backup per rotation. Engaged when primary does not ack within the time threshold.
On-call manager. Manager-tier engagement per team. Engaged for Sev-1 or sustained escalation across multiple levels.
Director. Executive-tier engagement per org. Engaged for major customer impact or business-critical incidents.

Escalation triggers are explicit, not judgement calls. Time-based and severity-based triggers both live in the paging tool, not in tribal memory.

Time-based. 15-minute no-response trigger per rotation. Next level pages automatically.
Severity-based. Sev-1 triggers immediate manager escalation. Skip the time-based wait when the severity warrants it.
Customer-impact trigger. Impact-threshold escalation catches incidents whose blast radius is under-recognised at first.
IC-requested escalation. The IC’s explicit “I need help” path. Judgement-driven escalation backed by the tool.

Escalation trees go stale fast as teams reorganise. Annual audit plus quarterly synthetic-page tests keep the routes accurate.

Update on team change. Each reorg, hire, or team split touches the tree. Out-of-date trees route to ex-employees.
Annual full audit. Walk every rotation, every escalation level, every named contact. Stale trees are operational liabilities.
Quarterly synthetic page. Simulated escalation verifies the tree actually reaches the right people. Catches drift before the real page misroutes.
Named tree owner. Maintaining team named per tree. “Everyone-and-no-one” ownership is how trees rot silently.