Skill Mix Is the Real Constraint
Why headcount is the wrong number for most rotas, how to express a shift requirement as a composition, and what changes when you do.
What one shift requirement looked like once it was written as a composition
Two of the five bind, one bends for two days of training, and two are assumptions — one of which has been quietly adding a person to every early shift on paper who is not there in practice. This is one employer's own requirement, not a statement of what any standard requires.
Most rotas are built by counting people, and most services require a composition. Four people on a shift is a number; four people of whom two can run the line, one can sign off the work and one is still in training is a requirement, and only the second can be checked.
The skill constraint in “Skill Mix Is the Real Constraint” should remain visible separately from simple headcount. For organisations assessing review the platform information here in relation to absence management, work and attendance records can support the operational picture without replacing qualification records or professional judgement.
Where the two diverge, the rota fills the shift and the shift does not work, which is experienced as a bad day rather than as a scheduling error.
The assumptions behind “Skill Mix Is the Real Constraint” can also be checked against ASQ quality resources. Record what applies locally and what requires qualified advice.
The requirement written properly
For each shift type, what has to be true: which tasks must be coverable, which authorisations must be present, and what the minimum is for each.
That is usually three or four lines per shift and almost nobody has written them. The absence is what forces everybody to rely on the judgement of one person who knows who can do what.
Why that one person is the real constraint
In most teams there is somebody who can look at a draft rota and say "that Tuesday will not work". They are holding the skill matrix in their head, and they are the reason the rota functions.
They are also a single point of failure, and the rota's quality drops visibly whenever they are on leave. That is the clearest possible signal that the composition requirement should be written down.
Building the matrix
Rows are people, columns are tasks or authorisations, cells say one of three things: can do unsupervised, can do supervised, cannot do.
| Line start | Sign-off | Autoclave | Night lead | |
|---|---|---|---|---|
| Person A | unsupervised | unsupervised | no | unsupervised |
| Person B | unsupervised | no | unsupervised | supervised |
| Person C | supervised | no | no | no |
Three states, not two. The middle state is where most of the useful information lives: it is the list of people who are two shifts of supervision away from removing a constraint.
Keeping it honest
A matrix filled in by managers tends to be optimistic. A matrix filled in by the people themselves tends to be cautious. Both are useful and they should be visibly different columns rather than averaged into one.
Where the two disagree about the same person and task, that is a training conversation that would otherwise never have happened.
Reading the matrix for constraints
Two readings matter. By column: how many people can do each thing, and which columns have one name in them. By row: how many things each person can do, and who is carrying the most.
The first reading finds the fragility. The second finds the person about to leave, because carrying every column is a reliable predictor of somebody going somewhere else.
What the rota should check
Before publication, for every shift: does the composition requirement hold, with the people actually assigned, under the assumption that the newest person needs supervision.
That last clause is the one that gets dropped. A shift that works because a trainee is counted as a full pair of hands is not covered; it is covered on paper and short in fact.
The arithmetic of training
A constraint held by one person costs whatever the cover arrangements around it cost. Training a second person costs a known amount of time and money, once.
Those two numbers are comparable and they are almost never compared, which is why single-holder constraints persist in organisations that would obviously rather not have them.
The shift that nobody can staff
Every team has one, usually mid-week nights or a weekend late, and it is always a composition problem rather than a headcount problem.
Finding which column causes it takes ten minutes with the matrix, and the answer is normally one authorisation held by three people, all of whom work days.
What to publish to the team
The matrix, with the names, is a sensitive document and does not belong on a wall. The column totals are a different matter: how many people can do each thing, without saying who.
Publishing the totals makes the fragility visible to everybody, including the people who could volunteer to be trained, and that is where most of the movement comes from.
The column that is really a person
Where a capability column has one name in it, the rota is not constrained by a skill; it is constrained by an individual's availability, and every holiday request becomes a scheduling incident.
Marking those columns on the matrix — a line under any column with fewer than three names — turns a spreadsheet into a plan. The marked columns are the training budget, in priority order, for the next year.
What to put on the constraint register
For each shift type, the composition requirement in one line, and for each capability, how many people hold it and the thinnest point in the rota.
Two short tables. Together they explain more about why a rota is hard than any amount of discussion about patterns, and they turn "we are short-staffed" into a specific statement about which capability is short.