Demand in a warehouse is rarely flat, and staffing is frequently designed as though it were.
Building the curve
Order arrival by hour, not by day, which is the granularity where the pattern lives.
Work released by hour, which may differ substantially from order arrival because of wave scheduling.
Work completed by hour.
Plot all three together. The gap between release and completion is the queue, and the shape of the release curve is frequently the facility's own doing rather than the customer's.
The patterns to look for
Daily shape: the morning ramp, the post-release surge, the despatch cut-off compression.
Weekly shape: Monday recovery from weekend orders, Friday despatch pressure.
Monthly and seasonal, which drive the peak planning.
Cut-off effects, where a despatch deadline compresses several hours of work into one.
Self-inflicted peaks, created by wave timing rather than by demand. These are the cheapest to fix and the most often overlooked.
Shift shaping
Match staffing to the curve rather than to a uniform shift.
Staggered starts, which flatten the ramp.
Overlapping shifts at the peak.
Part-time and flexible cover at the compression points.
Break scheduling staggered so capacity does not drop at a fixed hour.
Each of these is a rota negotiation rather than a capital cost, and together they usually recover more capacity than any equipment purchase.
Flattening the demand itself
Better than staffing to a peak.
Change the wave schedule, which is entirely within the facility's control and frequently set years ago.
Smaller, more frequent releases, which reduce congestion and shorten cycle time.
Move non-time-critical work to the trough: replenishment, cycle counting, returns.
Negotiate cut-offs with commercial colleagues, which is a conversation worth having with the data attached.
Level the inbound schedule with suppliers, where there is leverage.
Peak planning
Use several years of history, not last year.
Plan for the peak day, not the peak month, because the daily figure is what breaks.
Model the constraint at peak volume, which may be a different stage than at normal volume — dock capacity frequently becomes the limit only at peak.
Plan the temporary labour ramp including training time, which is routinely underestimated and produces a peak staffed with people who are not yet productive.
Plan the equipment, including charging capacity, which is a common peak failure in electric fleets.
The staffing conversation
Present the curve, not the average.
Show the hours where the operation is short and the hours where it is over.
Quantify the cost of both: overtime and waiting at the peak, idle cost at the trough.
Offer the options: reshape the shifts, flatten the demand, or add capacity.
A curve makes this a design discussion. An average makes it an argument about whether people are working hard enough.
What to watch afterwards
Whether the queue moved rather than reduced, which happens when staffing is added ahead of the constraint.
Whether cycle time improved, which is the customer-visible outcome.
Whether overtime fell, which is the financial one.
Whether the peak moved, because reshaping frequently relocates the compression rather than removing it.
Finding the self-inflicted peak
The cheapest capacity available in most facilities.
Plot work released by hour against orders received by hour.
Where the release curve is spikier than the arrival curve, the facility is creating its own peak.
Trace it to the wave schedule, the batch job, or the cut-off.
Smooth the release, which requires only a configuration change.
Measure congestion and cycle time before and after, both of which usually improve immediately.
Planning the peak ramp
Temporary labour is routinely planned by headcount and not by productivity.
Include training time, which is weeks rather than days to full productivity.
Model the ramp curve, not a step change: new staff produce less for a period.
Plan the trainers, who are your experienced people and who are therefore not picking while training.
Plan the equipment and charging, which is a common peak failure.
Start earlier than feels necessary, because a peak staffed with untrained people is worse than a peak understaffed with experienced ones.