Give people weeks where nobody will call them, and the on-call week becomes something they can survive.
What is wrong with calling people in a fixed order?
Everybody carries the phone and nobody is ever free of it.
Before the rotation, we’d four people running the primary infrastructure and a simple escalation order. Call the first person. If no answer, the second. Then the third. Then the fourth.
That arrangement looks fair on paper and it’s not. Person one takes almost every call and gets destroyed. Person four is almost never reached and learns nothing.
The worse problem is what it does to the other three. Any of them might be next, so all four keep their phones on them all the time. Four people carry the burden and only one of them gets the calls.
How should an on-call rotation work?
One person, one week, then the next person.
Nothing else to it. During your week you’re the one who gets called. During the other weeks you’re not.
Nobody found this appealing at first. My team hated it, because what they saw was a week where every call was theirs with no one to pass it to.
The week they were on had become harder. That part of their reaction was correct.
Why do people come around to a rotation?
Because they work out what happens during the other three weeks.
Off rotation, they didn’t have to keep their phones with them. They didn’t have to answer. They could go somewhere without coverage, have dinner without checking, and sleep without listening for anything.
That had never been available under the old arrangement, and none of them had noticed it was missing until they got it.
An on-call week is a cost. Three weeks of genuine freedom is what you buy with it, and the trade is a good one as long as the freedom is real.
What happens in a genuine emergency on an on-call rotation?
Anyone can be reached, and that path has to exist.
I put procedures in place so a real disaster could still reach anybody. Not everything is survivable by one person on duty, and a rotation that pretends otherwise will be broken quietly on the first bad night, which destroys it permanently.
The distinction I drew is the part that makes it work. Off-rotation people didn’t need to be on call and didn’t need to be in a position to answer.
They might get reached. They weren’t required to be reachable. That’s a different thing and it’s the whole difference between a rotation and a schedule that exists on paper.
When is it acceptable to break an on-call rotation?
When the situation requires more than one person, and almost never otherwise.
The escape hatch is the most fragile part of the whole arrangement. Used for real emergencies, it costs nothing and preserves the boundary. Used because the person on duty was unsure and wanted company, it teaches everyone that off weeks aren’t real.
Two or three casual uses and the rotation is decoration. People go back to carrying their phones, because they’ve learned that not being on call doesn’t mean anything.
Protecting that boundary is the manager’s job and it costs something. Occasionally the person on duty struggles alone with something a colleague could have solved faster, and that’s the price of everyone else having a real week off.
Why does schedule predictability matter more than pay?
Because nobody leaves over the night they got called.
They leave over two years of never being able to put the phone down. The particular Tuesday isn’t what wears people out. The absence of any period where the phone doesn’t matter is what does it.
That’s the entire argument about operations burnout without using the word. A team with genuine weeks off can sustain a hard on-call week indefinitely. A team where every week is a low-grade on-call week runs down slowly and nobody can point at when it started.
For a small company this is one of the few retention levers available that costs no money. A rotation with a real boundary is a scheduling decision and a discipline, and it changes how the job feels considerably more than a raise does.