The moment a small team has customers who depend on its service outside office hours, somebody has to be reachable. Without a plan, that somebody is always the same person, usually the most conscientious one, until they quit. A rotation spreads the load fairly and makes sure someone is always responsible. These are the questions teams ask most when setting one up.
When do we actually need an on-call rotation?
You need one when an outage outside working hours would cost real money, safety or trust, and waiting until morning is not acceptable. If a failure can wait nine hours, you probably need good alerts and a morning routine, not a rota. Be honest about which services are truly critical and cover only those. Four to five people is a comfortable minimum; with two or three, each person is on call so often that it stops being a rotation and becomes a lifestyle. In that case, narrow the scope, shorten the covered hours, or share the rotation with an adjacent team.
How long should one shift last?
One week is the most common choice, with the handover at the same time each week, often mid-week so a weekend is not split awkwardly. The alternatives:
- Daily shifts. Light on each person, but many handovers and a lot of context switching.
- Weekday and weekend split. Weekends are the hardest part, so some teams rotate them separately.
- Follow-the-sun. Only if you have people in different time zones, so nobody is woken at night.
Whatever you choose, publish the schedule at least a month ahead and let people swap shifts freely, as long as the swap is recorded.
Do we need a primary and a backup?
Yes, even on a small team. The primary receives every alert first. The backup is contacted if the primary does not acknowledge within a set time, say ten minutes, or asks for help. This protects against the ordinary reasons an alert is missed: a flat battery, a tunnel, a shower. Keep the backup role lighter than the primary, with no expectation of sitting at a laptop, only of picking up when called. Rotate both so everyone learns each role.
What makes a good handover?
A short, written handover at the end of every shift, ideally a ten-minute call plus notes. Cover:
- Incidents from the week and their status
- Known flaky alerts and risky changes coming up
- Anything half-fixed, with a link to the ticket
- Planned maintenance or deployments in the next shift
The outgoing person should stay reachable for questions for a few hours. A good handover is the cheapest way to avoid investigating the same incident from scratch twice.
How should on-call be compensated?
Being available is work, even when the phone never rings. Pick a model and write it down. Common options are a flat stipend per shift, extra pay for each hour actually worked on an incident, time off in lieu, or a combination. Time off matters most: if someone is woken at 3 a.m., they should not be expected at their desk at 9 a.m. Rules differ by country and employment contract, so check local labour law on rest periods and on-call pay, and agree the terms with the team rather than imposing them. Avoid treating on-call as a favour that heroes do for the company.
What should the escalation path look like?
Write it as a simple chain anyone can follow half-asleep: primary, then backup, then a named team lead or manager, and finally whoever owns the affected system or the decision to inform customers. For each step, state how long to wait before moving on. Add a one-page list of who owns what, plus rules for when to wake someone senior. Make it explicit that escalating is expected and never a sign of failure; people who fear judgement hold on to incidents for too long.
How do we keep alert noise down?
Alert hygiene is what makes a rotation sustainable. Page a human only when the alert is urgent, actionable and needs a person right now. Everything else becomes a ticket or a morning report. Practical habits:
- Delete or fix any alert that fired and needed no action
- Attach a short runbook to every alert saying what to check first
- Group duplicates so one problem produces one page
- Separate urgent pages from low-priority notifications
A useful goal is that most shifts are quiet, with no more than a couple of out-of-hours pages in an average one.
How do we prevent burnout?
Watch for the signs: broken sleep, resentment about the schedule, people dodging shifts. Protect people with a few firm rules: a minimum gap between shifts (never back-to-back weeks), recovery time after a bad night, exemptions for holidays, illness and new parents, and a way to step out for a period without penalty. New joiners should shadow a shift before taking one alone. And fix causes, not symptoms: if the same problem keeps paging, schedule engineering time to remove it.
How do we review the load?
Every month, look at the numbers: pages per shift, pages outside working hours, time to acknowledge, time to resolve and hours worked per person. Check fairness: is the load even, or does one person carry the nights? Then hold a short, blameless review of the noisiest alerts and the worst incidents, and agree one or two fixes. If you already plan work in FulcrumGrid Command Center, you can keep the rota, the follow-up tasks and these load figures beside your other projects and KPIs, though a shared calendar and a spreadsheet are enough to start.
Begin small, write the rules down and adjust after every month. A fair rotation is not one that never inconveniences anyone, but one where the burden is shared, visible and regularly improved.
Frequently asked questions
How many people do we need for an on-call rotation?
Four to five is a comfortable minimum. With fewer, shifts come around too often. Narrow the covered hours or services, or share a rotation with another team.
How long should an on-call shift be?
One week is the most common length, with a fixed handover time. Shorter shifts mean more handovers, longer ones increase fatigue. Publish the schedule at least a month ahead.
Should on-call staff be paid extra?
Availability is work, so most teams pay a stipend, extra pay for incident hours, time off in lieu, or a mix. Check local labour law and agree the terms openly with the team.
What is a reasonable number of pages per shift?
There is no universal figure, but most shifts should be quiet. If people are regularly woken more than once or twice a week, fix or remove noisy alerts before adding staff.
FulcrumGrid Command Center
Keep your rota and follow-ups in one place
FulcrumGrid Command Center helps you track projects, tasks and KPIs together, so on-call load and the fixes it triggers stay visible.