Overtime started as a bridge. A large order, a line down for a week, a few Saturdays to catch up. Eighteen months later it is in the standard cost, the crews plan their lives around it, and the Saturday shift has a name on the roster.
Nobody has asked what it is paying for.
Overtime is the most expensive capacity a plant buys and the least examined. It sits in a labor line that HR owns, it is approved a week at a time by people who are measured on output, and it rarely appears in the OEE report as a number of its own, because weekend hours are either blended into the weekly average or left out as “extra” time. This article gives you a two-week diagnostic that answers one question: how much of your overtime serves real demand, and how much of it is buying back hours your lines lost during the week, at a premium.
The visible cost is the premium rate: time-and-a-half, double time on Sundays, whatever the agreement says. Most plants stop there. The full cost of an overtime line-hour has four more parts.
Lower output per hour. Sixth and seventh days run measurably lower OEE. Fatigue, thinner crews and a line that was cold at 6 a.m. all cost cycle time. In the plants we have measured, an overtime hour often produces 70 to 80% of a weekday hour on the same line.
The maintenance window that no longer exists. Saturday used to be when the preventive maintenance got done. Now it is production, so the PM slips, and the unplanned stops during the week get worse. Overtime creates the losses it is covering.
Overhead on a thin shift. Utilities, compressed air, a supervisor, a quality technician and a forklift driver for two lines instead of six. The fixed cost per unit on a Saturday is higher than any weekday.
Start-up losses. A line that was down for 36 hours does not run at speed for the first hour. On a 16-hour overtime shift that is 6% of the shift gone before the first good pallet.
Add these together and a fully loaded overtime line-hour often costs 1.6 to 2 times a scheduled line-hour, measured in cost per good unit rather than cost per hour. The figure is illustrative; the point is that whatever your premium rate is, the real multiple is higher.
There are two reasons a plant runs overtime, and they look identical on the roster.
The first is that demand genuinely exceeds what the lines can produce in the scheduled week, even at a good OEE. Overtime is then a legitimate way to serve a peak, and the real question is whether a fourth crew or a new line is the permanent answer. The hidden capacity method and the new-line cost comparison are for that case.
The second is that the lines lose a large share of the scheduled week to micro-stops, changeovers over standard, speed loss and unplanned stops, and the overtime shift exists to produce the output those hours should have produced. The plant has already paid for the scheduled hour once. It is now paying a premium to buy the same output a second time.
Every plant is a mix of the two. The diagnostic tells you the proportion. In our experience, plants that have been running structural overtime for more than a year are usually surprised by how much of it falls into the second category.
Pick the line that runs the most overtime. Collect four numbers over two normal weeks.
Then one ratio:
Overtime replacement ratio = recoverable hours lost on scheduled time ÷ overtime hours worked
Count only recoverable losses. Every line loses some time it will never get back, so a ratio built on all losses comes out above 1 almost everywhere, even on a line running at 85% OEE. At 1 or above, the scheduled week could produce everything the overtime produces once the recoverable losses are removed. Between 0.3 and 1, a large share of the overtime is loss-driven. Below 0.3, overtime is mostly serving real demand.
The number finance will care about is the loss-driven share of the overtime bill: the ratio (capped at 1) multiplied by overtime hours, multiplied by the fully loaded cost per overtime hour. That is what the plant spends each week to produce output it has already paid for once.
A packing line is scheduled 120 hours a week and runs at 58% OEE. Lost hours on scheduled time: 42% of 120, about 50 line-hours a week. The plant runs a 16-hour Saturday overtime shift on that line to keep up.
Most of those 50 hours cannot be recovered without capital. The hidden capacity method on the same line found 12 OEE points that can: about 14 line-hours a week. Replacement ratio: 14 ÷ 16, about 0.9. And because a Saturday hour produces only 70 to 80% of a weekday hour, the 16 Saturday hours deliver the output of about 12 weekday hours. The recoverable weekday hours already cover all of it.
The cost side, illustrative: a crew of six at $28 an hour with a 1.5× premium is $252 an hour in direct labor, about $4,000 for the shift. Add supervision, utilities and overhead on a thin shift, and the Saturday costs around $5,000 to 6,000 per line. Across six lines and 50 weeks, that is $1.5 to 1.8 million a year, and because the recoverable hours cover the Saturday’s whole output, almost all of it is loss-driven.
Now the comparison. Recovering those 14 hours puts the Saturday’s output back into the scheduled week, at weekday cost and with the maintenance window back. The remaining losses are either structural or accepted, and any overtime that remains after recovery is the real demand gap, now with a measured size.
One FMCG group running Fabrico across its plants increased output by 14%. The full story is in the Ficosota customer case.
Measure weekend OEE separately from weekday OEE. Most plants have never seen the two numbers side by side, and the gap is usually the moment the problem becomes visible in one figure.
A Saturday shift runs with a skeleton crew, often the people who volunteered rather than the people who are best on that line. There is no maintenance technician on site, so a fault that takes ten minutes on a Tuesday takes forty on a Saturday or ends the shift. There is no engineering support, no planner, and the material handler covers three lines. The line was cold, and the first hour is ramp-up. Fatigue is real on a sixth consecutive day.
The result in many plants is that the overtime hour produces 70 to 80% of a weekday hour. So the overtime hour is both more expensive and less productive, which means the replacement ratio above understates the problem: 16 Saturday hours are buying back fewer than 16 weekday hours’ worth of output.
There is a second-order effect. The Saturday that used to be the maintenance window is now production, so preventive maintenance slips, which raises unplanned stops during the week, which raises the lost hours, which justifies more Saturdays. Structural overtime feeds the losses it exists to cover.
Ratio of 1 or above: recover the losses, retire the structural overtime. The scheduled week has more than enough capacity once the top losses are addressed. Build the loss Pareto, sort it into recoverable and structural, fix the top three. Keep overtime as a flex tool for genuine peaks. Payback is measured in months, and the maintenance window comes back as a bonus.
Ratio between 0.3 and 1: recover first, then size what remains. A large share of the overtime is loss-driven and goes away with the recovery program. What is left is real demand, and it now has a measured size, which is the input the fourth-crew or new-line decision needs. Deciding on a crew or a line before recovery means sizing it to a demand gap that is partly your own losses.
Ratio below 0.3: the overtime is genuine demand. Permanent overtime is still the wrong way to serve it; it is the most expensive capacity on the menu. The choice is a permanent additional crew, which is usually cheaper per unit than standing overtime once the premium and the lower output are counted, or new equipment. The new-line cost comparison is the next article to read.
This is not an argument against overtime. Flexing a crew for a peak week is one of the cheapest, fastest levers a plant has, and a plant that never runs overtime is probably carrying too much fixed capacity. The argument is against structural overtime that nobody has measured: a Saturday shift that has been on the roster for a year, is in the standard cost, and has never been tested against the question of what it is replacing. That is a loss hiding inside a labor line, and it is often the largest unexamined cost in the plant.
Add one line to the monthly pack: overtime hours, with the loss-driven share next to it. Three columns are enough: overtime hours, replacement ratio, loss-driven cost.
This does two things. It moves overtime from an HR line that gets approved to an operations metric that gets questioned. And it gives the recovery program a budget it funds itself: when the loss-driven overtime cost on six lines is in the region of $1.5 million a year, a program costing a fraction of that does not need a business case so much as a start date.
How much does overtime really cost in manufacturing? More than the premium rate. Once lower output per hour, overhead on a thin shift, start-up losses and the lost maintenance window are counted, a fully loaded overtime line-hour often costs 1.6 to 2 times a scheduled line-hour per good unit produced.
Is overtime cheaper than adding a shift? For a short peak, yes. As a standing arrangement, usually no: a permanent additional crew runs at standard rates, at weekday OEE, with maintenance support, and is cheaper per unit than structural overtime. Before adding either, measure how much of the overtime is replacing hours lost on the scheduled week.
Why is weekend production less efficient? Skeleton crews, no maintenance or engineering support, fatigue on a sixth day and cold-line start-up losses. Many plants find a weekend hour produces 70 to 80% of a weekday hour on the same line. Measure weekend and weekday OEE separately to see your own number.
How do you reduce overtime without losing output? Measure the hours lost on scheduled time, rank the losses, and remove the top three. On lines running below 65% OEE, 8 to 12 points are often recoverable without capital, which is usually more than the overtime shift was producing.
What is a reasonable overtime level for a manufacturing plant? There is no fixed number, but a useful test is the overtime replacement ratio: recoverable hours lost on scheduled time divided by overtime hours. At 1 or above, the plant is paying a premium to buy back losses it could remove, and the overtime level is not a demand problem.
Two weeks of machine data on one line gives you the ratio. It is the cheapest piece of analysis the plant will do this year, and it decides whether the next conversation is about a recovery program, a fourth crew or a new line.
The fastest route is a fixed-scope pilot: one line, six weeks, an industrial sensor and hub installed with your team, a machine-verified loss profile, the three interventions that retire the Saturday shift, and an executive readout with the number in dollars. The fee is fixed and credited in full against a first-year subscription if you roll out. The pilot runs on Fabrico’s manufacturing performance platform (MES, OEE, CMMS & AI), which connects machine data, OEE and loss analysis, production scheduling, SKU-level output value and maintenance in one system, so the hours won back from the Saturday shift stay won.
Request a demo or estimate your own line with the OEE calculator.
Login
Blog
