The 24/7 Shrinkage Problem Nobody’s Tracking

The Assumption Nobody Questioned

If you ask most workforce leaders how their shrinkage figure was set, the honest answer is: it was measured during business hours, averaged across the year, and applied everywhere.

That made sense when the contact center ran Monday through Friday, nine to five. It stops making sense the moment you add overnight shifts, weekend coverage, and the skeletal staffing patterns that utilities operations require around the clock.

Shrinkage is not uniform. It never was. But the model treats it that way. And in a 24/7 operation, that single assumption quietly inflates labor cost on every overnight and weekend interval you schedule.

The day-shift calibration wasn’t wrong. Applying it to every other shift without adjustment is where the cost enters.

This is not a technology problem. WFM platforms will run whatever assumptions you put into them. The issue is that the assumption was set once, it seemed reasonable, and there has been no structural reason to question it until someone looked at the numbers shift by shift.

Most operations haven’t looked.

Why Off-Hours Shrinkage Runs Higher

There are three mechanics that consistently drive overnight and weekend shrinkage above the modeled rate. None of them are surprises to anyone who has managed a night shift. What is surprising is that almost none of them show up in the shrinkage figure.

1. Absenteeism by Shift

Unplanned absence rates are measurably higher on overnight and weekend shifts than on standard weekday business hours. This is well-documented and operationally intuitive. Coverage is thinner, personal conflicts are harder to avoid, and the social contract around attendance on a 2 AM shift is different than a 10 AM shift.

A blended shrinkage figure absorbs this variance into a single number. The day-shift average pulls the overnight reality down to something that looks manageable on paper. On the schedule, you see it as chronic understaffing on the shifts where it is hardest to flex in coverage.

What this looks like on the overnight schedule:

Headcount looks adequate on paper. The interval fills in. But three agents go unplanned absent on a Tuesday night, and the remaining team covers at occupancy levels that would trigger an alert on a Monday morning. The overtime gets approved to patch it. It happens again the following Tuesday. Nobody connects it to the shrinkage assumption.

2. Supervisory Coverage and Adherence Gaps

Day-shift operations run with fuller supervisory teams, more real-time management presence, and active intraday monitoring. Coaching happens. Schedule adherence is tracked and corrected. Agents who drift off-queue get pulled back quickly.

Overnight and weekend operations typically run with reduced supervision. Not because it is the wrong call, but because that is how most operations are staffed. The consequence is that planned shrinkage activities run long, adherence monitoring is less active, and the practical shrinkage rate climbs above the model assumption without anyone formally registering it as a variance.

This is not a management failure. It is a structural reality that the shrinkage model does not account for.

3. Planned Activity Loading

Training, coaching sessions, one-on-ones, and system maintenance windows tend to get scheduled on overnight and weekend shifts. The logic is straightforward: there is more schedule flexibility when volume is lower, and pulling agents for planned activities does less damage to service levels.

The problem is that those activities are often not captured in the shrinkage estimate for those intervals. The day-shift shrinkage model reflects day-shift training loads. Overnight shifts carry additional planned activity that was scheduled precisely because there was “room.” That room gets consumed. And then overtime covers the gap the model did not account for.

A rule of thumb worth testing:

Add up all planned training, coaching, and maintenance windows that were scheduled on overnight or weekend shifts in the last quarter. Calculate what percentage of paid agent hours those activities represent on those specific shifts. Compare that number to the shrinkage percentage in your model. If the model is lower, the gap is structural.

Why Your Reports Don’t Show It

If this gap is real and recurring, why doesn’t standard reporting surface it?

The honest answer is that most shrinkage reporting produces a blended figure. Total shrinkage across all intervals, all shifts, all days. The number looks reasonable because the day-shift data, which reflects the model’s assumptions reasonably well, dominates the average.

The overnight gap is there. It is just averaged out.

The cost does not hide because the data is not there. It hides because the data has never been broken out by shift window.

A second reason is how overtime gets coded. When overnight shrinkage runs hot and service levels slip, the response is immediate and operational: add a body, extend a shift, call someone in. The cost shows up as overtime. It gets approved. It gets budgeted. Over time it becomes an expected line item for nights and weekends.

Nobody asks whether the overtime is covering a staffing shortfall that the model should have caught. The overtime becomes the answer instead of the signal.

There is also a framing problem. When your blended shrinkage looks fine and your overnight service levels hold, the conclusion is that the model is working. What is actually happening is that the overtime is making the service levels work. The model’s error is being funded by labor cost that never gets traced back to the assumption.

What the Gap Costs

The math is straightforward once you break it out by shift.

Consider a 200-agent utilities contact center where 30% of FTEs cover overnight and weekend shifts. That is 60 agents across those windows. The current shrinkage model applies a uniform 15% shrinkage assumption. Based on actual absenteeism data, planned activity loading, and adherence patterns, overnight and weekend shrinkage is running closer to 22%.

That 7-point gap means the operation is chronically scheduling as though 51 agents are available on those shifts when the real available count is closer to 47. The shortfall of four available agents per interval, across overnight and weekend coverage, requires systematic overtime or service level compromise to patch.

These are illustrative figures. The actual number on any given operation will depend on shift distribution, actual absenteeism data, and how planned activities are currently tracked. But the direction is consistent. Operations that have broken out shrinkage by shift window have found gaps in this range. The ones that have not are carrying the cost without knowing it.

One additional factor compounds the annual figure: the staffing decisions made to compensate. Supervisors who manage overnight shifts often request additional headcount to account for the chronic understaffing. When those requests get approved, the modeled gap becomes a headcount problem that finance sees as a capacity expense rather than a calibration error. The fix and the cause are recorded as different line items.

The Question Worth Asking

Not: “Do we have a shrinkage problem?”

The right question is: does your shrinkage model treat all shifts the same?

If the answer is yes, you do not know whether that assumption holds. You know your blended number. You do not know what the overnight and weekend numbers are, and you do not know how much of your off-hours overtime is paying for a calibration that was set on different data.

The good news is that this is a solvable problem. It does not require new software or a platform change. It requires looking at actual shrinkage by shift window, comparing it to the modeled figure, and adjusting the schedules that flow from the model. Operations that have done this work have closed the gap within a single planning cycle.

The overnight schedule is not broken. It is running on an assumption that was never designed for it.

The harder part is finding the time to look. Operations leaders managing 24/7 environments are running on tight bandwidth. The analysis required to isolate shift-level shrinkage variance is not technically complex, but it requires stepping outside the operational flow long enough to ask the question.

That is the work. And it tends to pay for itself quickly.

Note: Figures in are illustrative estimates based on commonly observed patterns in 24/7 utilities contact center operations. Individual results vary based on operation size, shift distribution, and current model calibration.