Both errors cost money. One of them gets counted

The error that gets counted

Overstaffing arrives on the labour report next week, at full loaded cost, attributable to a named manager. It is immediate, precise and personal.

The financial question in “Both errors cost money. One of them gets counted” requires more than a raw activity total. An organisation evaluating employee monitoring software for employee monitoring software can connect time and project records with labour analysis, provided pay rules, outcome measures and corrections remain visible alongside the dashboard.

Understaffing produces a customer who left without buying, a queue nobody will tolerate twice, work carried to the next shift, and one person who decided that week to start looking. None of it appears in any report and none of it has a manager's name on it.

For an independent reference related to “Both errors cost money. One of them gets counted”, consult the U.S. Bureau of Labor Statistics wage resources; it provides a useful external check on scheduling, working-time and workforce-planning assumptions.

Which means a manager choosing between the two errors is choosing between one that is counted and one that is not. The rational response is to understaff — and so operations err systematically in the direction they believe they are guarding against, with no evidence to the contrary because the evidence is absent by construction.

The remedy is unglamorous: count something on the invisible side. Walkouts, abandoned calls, queue length at peak, tasks carried over. A rough number on the side nobody measures beats a precise number on only one side, because it restores the comparison.

What an hour actually costs

Most managers plan against the wage rate. The loaded cost — rate plus statutory contributions, holiday, sick pay, pension, induction — is commonly a fifth to a half higher. Planning against the bare rate understates every hour you schedule, which quietly makes overstaffing look cheaper than it is.

Calculate it once: total employment cost for a period divided by total hours worked. Crude, correct in aggregate, and better than the headline rate.

And not all hours are worth the same. An hour at peak and an hour at the trough cost identically and return very differently, which means where you cut matters far more than how much. Told to reduce by ten per cent, most managers cut proportionally — taking the same share from the hours that earn most and the hours that earn least.

A schedule is a forecast

Every published rota is a bet: that demand will be roughly this shape, that these people will be available, that the work will take about as long as usual. Three predictions, compounding.

Calling it a plan is the problem. A plan is executed; a forecast is checked afterwards and improved. Operations that treat the rota as a plan ask whether it was followed rather than whether it was right.

Measuring your own accuracy takes eight weeks and two columns: expected demand before the week, actual after. The median difference is your typical error. And averaging the signed differences finds the direction — most forecasts are biased consistently, and correcting a known bias costs nothing.

That number has a practical consequence: a budget tighter than your forecast error will be missed for arithmetic reasons, and knowing it by how much is what turns a complaint into a case.

Headcount is not coverage

A rota shows names against hours. Whether the work can actually be done depends on what those people can do, and most scheduling treats them as interchangeable.

Nobody on shift can authorise the refund, open the safe, operate the machine, sign the release. The headcount is right and the operation stops — a failure that looks like nothing on any staffing report.

The fix is an hour's work: a grid with people down the side and capabilities across the top. Count how many hold each. One is a single point of failure; two who work the same days is also one.

And a quick test for the informal version: can every person take two consecutive weeks off without the operation degrading? Where the answer is no, you have found a single point — and incidentally learned whether people can actually use their leave.

Why the variance is almost never the schedule

Labour is over budget and the schedule gets blamed. Six causes sit ahead of it: did demand differ from forecast, was the work mix different, was there absence and how was it covered, was there overtime or agency, did anything operational change, and only then — was the schedule loose.

Checking them in order takes ten minutes. Starting with demand is the important part: a busy week over budget is a success being reported as a failure.

The schedule gets blamed anyway because it is the thing the manager controls and the only cause with an obvious remedy. Which means the available answer is applied to whatever the cause was, and the actual cause persists.

When it is not a staffing problem at all

A persistent coverage problem is frequently generated somewhere else: a process that takes too long, demand that is its own rework, work assigned to the busiest role because that is how it evolved, equipment that is slow, or a service that costs more to deliver than it returns.

Counting what proportion of your demand is repeat or rework takes a week and in many operations the share is large and removable.

It also changes what you can ask for. "I need more hours" is a request that gets refused. "This step adds forty minutes per shift and here is the fix" is a proposal that gets agreed, and it costs less.

The people side is in the same arithmetic

The same hour is a cost on one side of the ledger and somebody's rent on the other. An hours reduction announced as an efficiency measure is received as a pay cut, because it is one.

Publishing two weeks ahead instead of three days costs no money — it costs flexibility, and the forecast error section tells you exactly how much. Priced against a few extra leavers a year, longer notice usually wins.

And goodwill is finite. An operation that uses short-notice changes weekly finds that nobody answers the phone, which is precisely when the surge arrives.