The Overtime Line Is the Smallest Bill You Pay

The Overtime Line Is the Smallest Bill You Pay

This winter we ran the numbers on a three-line packaging operation at a Midwest meat processor. Sensors, fourteen days of data, ChatGPT-assisted proxy OEE. Every line is rated for twenty parts per minute. The peak day proved it. The daily average came in at five or six parts per minute. The plant was running ninety minutes of overtime per day to meet demand. The supervisor told us, in the same breath, that they were short-staffed.

They are not short-staffed. They are running at thirty-nine percent OEE on line one and around fifty-two percent on lines two and three. Availability is the gap, not equipment capability. Lines start late. Line one comes up at eight when the plan is seven. Line two at seven when the plan is six-thirty. That is thirty to sixty minutes of capacity erased before anyone is tired, before any of the day's chaos has even started. The plant is then paying overtime at the back end to recover the capacity it gave away at the front.

What the Overtime Line Doesn't Show

Plants book overtime as a line item. The wage premium. Time-and-a-half on whatever number of hours the shift ran past schedule. That number is the smallest of three bills the plant is paying for the same hour.

The first hidden bill is the marginal hour itself. Not the wage premium, the structural premium. At one client, the contract requires a four-hour minimum payout any time a worker is called in for unscheduled overtime. A one-hour overrun is billed as four. A two-hour overrun is billed as four. That is not a fifty percent wage premium. That is, on a typical short-call event, a three hundred to four hundred percent marginal cost compared to the underlying average hour. The accounting line says overtime; the cash line says something very different.

The second hidden bill is fatigue against quality. The same workers who started at the published time are still on the line ninety minutes past scheduled stop. Cognitive load, manual dexterity, and attention all degrade past the eight-hour mark. The plants we have modeled rarely book the quality cost of that ninth hour against the overtime that bought it. The defect lands on Tuesday. The overtime hit a payroll account on Monday. The two never meet in the same model.

The third hidden bill is fatigue against safety. Same mechanism, sharper consequence. A West Coast frozen-meals producer we worked with runs two five-hour shifts back to back: five AM to one-thirty PM, then three PM to eleven-thirty PM. Their frozen-meals line carries twenty-eight associates per shift. Their fully-automated line carries fourteen. The frozen-meals line is not just twice the wage exposure of the automated line. It is twice the fatigue exposure, twice the cumulative manual handling, and twice the safety-incident surface area per unit of output. The structural cost of staffing twenty-eight people on a line that could run with fourteen is not the fourteen extra paychecks. It is the fourteen extra people the plant is putting at risk every shift to compensate for an automation gap that was never modeled.

Model the Hour Before You Buy It

The right way to price the marginal overtime hour is as a four-component portfolio.

One: the wage premium itself. Time-and-a-half if you are lucky, the four-hour-minimum structural premium if you are unlucky.

Two: the next-shift productivity tax. Shift handoffs leak information. Run-rate decisions made under fatigue carry into the next shift's first hour. The first hour of every shift is already the worst hour in the OEE record at most plants we work with. Overtime makes it worse. The plant pays for that hour twice.

Three: the quality contingency. Some defect rate is being baked into the late hours. If the plant does not measure it, it is paying it anyway. Pull two weeks of customer complaints and align the timestamps. The clusters are not random.

Four: the safety contingency. Same mechanism. Pull two years of incidents and align them against shift hour. The data is usually present and almost never modeled.

When the wage premium is the only bill on the page, overtime looks like the flexible-capacity buy. Run the four-component model and overtime stops looking like flex and starts looking like the cost of a structural availability problem. At the three-line plant above, ninety minutes of daily overtime was buying back capacity that was already paid for, already on the floor, and already lost to startup discipline. Eliminating the overtime would not have saved time-and-a-half on ninety minutes. It would have unlocked headcount reduction from ten to six on the same lines, at the same throughput, with less fatigue and less quality leakage.

What the Plant Is Really Buying

The leader at that client, in a separate conversation about the same data, said the quiet part out loud. If he were honest, he said, he would admit he has been sitting on three hundred thousand dollars of savings for four years. He was right, and he was also wrong about the number. The savings are not in shaved labor cost. They are in the three bills the plant has been silently paying every week for four years to keep the overtime line item small. Quality contingency. Safety contingency. Non-linear marginal hour.

Most plants do not have a labor problem. They have a structural-availability problem. Overtime is the symptom the structural problem is paying every shift to stay invisible.