The Four Kinds of Constraint
The four kinds of constraint that fix a rota in place, how to tell them apart in practice, and why treating all four as the first kind is the normal case.
Nine constraints on one ward rota, sorted by what actually holds each
Three of the nine bind outright, three bend at a price somebody could state, two are assumptions nobody has tested and one is left over from a structure that no longer exists. The rota had been built as though all nine were the first kind. This is one employer's own list, not a statement of what any rule requires.
A rota is the arrangement left over once everything that could not move has been subtracted, and the subtraction happens before anybody opens a calendar. What gets subtracted is a list of constraints, and almost nowhere is that list written down.
The constraint in “The Four Kinds of Constraint” becomes useful only when its source, cost and effect on coverage are visible. A team considering see how the scheduling workflow is organised for does microsoft teams track your activity can add timely work and attendance context, with transparent settings, limited access and a clear correction route.
When it is written down, it sorts into four kinds that behave completely differently. The first two are real. The third is a guess that has been repeated until it sounds like a rule. The fourth was real once.
Additional context for “The Four Kinds of Constraint” is available from Chartered Institute of Ergonomics resources. Use external guidance to prompt a specific validation check rather than replace local review.
The one that binds
Something that cannot be moved by this organisation at all: a rest period required by law, a qualification the person either holds or does not, a staffing standard the regulator or the insurer requires in writing.
The test is simple and it is documentary. Somebody can produce the document, point at the clause, and say what happens if it is breached. If nobody can do that in ten minutes, the constraint is probably not this kind.
The one that bends at a price
Something that could move, and the price of moving it can be stated on one line: two days of training, a second room, a van, an allowance, a conversation with one person.
These are the valuable ones, because they are the only constraints a manager can actually act on. They are also the ones most often filed under the first kind, which is how an organisation ends up believing its rota is fully determined.
The one that is assumed
Somebody believes it, nobody can source it. A start time everybody treats as fixed. A person who "does not do nights" for a reason nobody present can explain. A ratio quoted in meetings that appears in no document.
Assumed constraints are not lies and they are not usually wrong. They are unverified, and the cost of verifying one is a single question asked of the right person.
The one that is stale
A constraint that was real and is no longer: a pattern built for a four-team structure when there are now five teams, a cover arrangement for a service that closed, a restriction attached to equipment that was replaced.
Stale constraints are the easiest to remove and the hardest to notice, because nothing in the rota announces that its reason has expired. Nobody is wrong, and the rota is wrong anyway.
How to sort one in practice
Four questions, in order, and most constraints are sorted in under a minute:
- What exactly is the rule, stated as a sentence somebody could disagree with?
- Who imposed it, and can that be checked with a document or a name?
- What happens if it moves by one shift?
- What would it cost to find out?
The second question does most of the work. A constraint whose author cannot be identified is almost never the first kind, whatever everybody believes about it.
Why the sort matters more than the rota
Rebuilding a rota inside an unexamined constraint set produces a different arrangement of the same compromises. That is why so much scheduling effort changes so little: the work goes into rearranging people inside a box whose walls were never measured.
Sorting the constraints usually finds two or three that bend, and those two or three decide what the rota can be. Everything else is detail.
What the sort usually finds
In the organisations that have done this, the proportions are surprisingly consistent: roughly a third genuinely bind, a third bend at a price nobody had ever been asked to state, and a third are assumed or stale.
That last third is the whole argument for doing the exercise. It is also the part that costs nothing to remove, which is why it is worth an afternoon before any conversation about systems or headcount.
The two lists worth keeping separate
Write the binding constraints on one list and everything else on another. The first list is short, boring and genuinely fixed; it is the shape of the problem and it should be agreed once a year.
The second list is the working document. Every line on it has a price, and the whole of rota improvement consists of picking lines off it in order of what they cost against what they would release.
Where to keep the list
With the rota, not in a planning document somebody wrote once. One page, one line per constraint, four columns: the rule, its source, which kind it is, and the date somebody last checked.
A list kept anywhere else goes stale faster than the constraints it describes, and a stale list of stale constraints is worse than no list, because it carries the authority of having been written down.