On-call without burnout: rotations that hold

On-call burns people not through incidents but through noise: pages that needed nobody, rotations with no backup, alerts without runbooks. A sustainable rotation is a design with three parts: fairness, actionability and payback.

Fair, actionable, paying back

Fair rotations. Short enough shifts, named backup, real handover notes, and no-hero coverage gaps. The schedule respects nights, weekends and holidays explicitly.

Actionable pages only. A page carries the symptom, the runbook link and the first command. FYI messages are tickets, not pages. Each new alert needs a runbook before it may page.

Every page buys a fix. Threshold tuning, automation or elimination follows every page within days. The page count trends down over quarters, or the design failed.

Worked example: a fictional noisy team

The context below is fictional. Fictional team ParcelTrack (fictional) gets forty night pages a month, mostly disk warnings and flaky checks nobody owns.

The redesign pages only on user impact with runbooks, converts the rest to tickets, adds backup and handover, and ties every page to an action item. Three months later: single-digit pages, most with fixes attached, volunteers instead of draftees.

Checklist: a rotation that holds

1. Night pages are user-impacting and actionable only.
2. Backup and handover named every shift.
3. Every alert links a tested runbook.
4. Page count trends down quarter over quarter.

Related reading

• Runbooks first: Runbooks the next person can run.
• The first-hour frame: Incident response.