On-Call Rotation Fairness

Even distribution.

Key metrics

Rotation fairness only works if it’s measured. Four metrics capture different facets of load; together they expose the imbalance that single-metric views hide.

• Pages per engineer per quarter. Absolute page count; the gross-imbalance signal; coarse but unambiguous.
• Off-hours pages per engineer. Nights-and-weekends share; the metric the page count alone hides; the burnout driver.
• Total page-minutes per engineer. Duration-weighted load; matches actual workload; one 4-hour incident is not the same as one 5-minute page.
• Per-engineer ack latency. Median ack time; catches delegation-by-default where one engineer is always second to ack.

Targets

Without targets the metrics are just data. With targets they become a contract; the team holds itself accountable to the spread, not just the average.

• Within 20% across engineers. Spread target per rotation; sustained variance beyond is structural imbalance.
• Off-hours less than 30% of total. Per-engineer off-hours cap; consistent skew warrants investigation, not acceptance.
• Page-minutes drive prioritisation. Page-minute target per rotation; matches actual load instead of count alone.
• Per-quarter review the targets. Targets evolve with team and service mix; the quarterly review keeps them honest.

Addressing imbalance

Imbalance is structural; rotating harder will not fix it. Address the underlying cause, compensate the outliers, reshuffle the pairings; rotation policy is a tool, not a magic wand.

• Reshuffle rotation. Different shift pattern; pair the heavily-loaded engineer with a light week; rebalance the spread.
• Compensate outliers. Time off, stipend, public recognition; the engineer who carried the bad week deserves the bad-week response.
• Fix underlying causes. Alert tuning, staffing increase, handoff fix; the imbalance has a root cause; address it.
• Per-engineer comp time. Per-incident recovery time policy; the engineer who responded at 3am gets the next day off.

Quarterly review

Quarterly review keeps the rotation honest. Without it, fairness drifts; the team loses trust; the policy collapses; the rotation becomes the reason engineers leave.

• Per-engineer metrics. Per-quarter per-engineer metric review; catches outliers before they burn out.
• Long-term trends. Multi-quarter trend; catches structural imbalance the single-quarter view misses.
• Action items with owners. Per-quarter action items with named owners; supports root-cause fixes, not just observation.
• Document the policy. Per-team fairness policy committed to the handbook; supports operational reviews and onboarding.