What overtime is costing your plant right now
Three numbers run this whole page, and then you can put your own in.
The average manufacturing wage
$30.37
Production and non-supervisory workers, August 2026.
The same hour, in overtime
$45.56
Time and a half. Same operator, same machine, same output.
What you lose on every overtime hour
$15.19
The top-up. It buys you nothing at all.
One person on 8 overtime hours a week, which is double the national manufacturing average of 4.0, costs you $6,319 a year in premium alone. What changes with the size of your floor is not that number. It is how many people cross the line.
The jump at 200 is not a rounding error. Two hundred people is roughly where you can staff four crews on a line that never stops, and a week holds 168 hours, so four crews puts every single person at 42 hours before anything goes wrong. Below that, overtime is an event. At 200 it is arithmetic.
Read the small numbers against what you keep rather than what you turn over. A 20 person shop clearing a few hundred thousand a year is handing a large slice of its profit to hours that produced nothing extra. And all of this moves with your state, in two ways. A California or Washington floor pays well above $30.37 an hour, so the same hours cost more there. More importantly, a handful of states count overtime by the day as well as by the week, so on a twelve hour rotation the shift itself can create overtime before the week is anywhere near 40.
Before any of the six, do this one. It is free and we do not sell it.
Move the weekday your rotation starts on
Overtime is counted inside a seven day payroll box, and a 12 hour shift cannot cut itself in half at midnight on Saturday. The same rotation, run by the same crews, producing the same tonnage, hands one person 4 overtime hours a week if the cycle opens on the right weekday and 12 if it opens on the wrong one.
Write your cycle out day by day. Slice it into the seven day blocks payroll actually uses and total each block. Then try all seven weekdays as the opening day and total them again. Keep the pattern your crews already know. Change only the day it starts on.
Going from 12 overtime hours a week to 4 on a 200 person floor is worth
$1,263,808
a year. Nobody works a different shift to get it, and you do not need us to do it.
Here are the six. Four of them stop a shift from ever being assigned. One reads the week you are already living in. One tells you what the week costs before payroll does.
1.A weekly hour ceiling, so the shift is refused before it exists
Why this puts you in overtime
The reason you keep going into overtime is the thing you build the schedule in. A spreadsheet, a whiteboard, or an outdated scheduling app that was never built to stop you from going into overtime. None of them will stop you, and none of them are counting a person’s hours while you type or while you generate the schedule. You drag a name onto Saturday because the line needs covering, and nothing on the screen tells you that name is already sitting at 44 hours. You put them on the shift and you move on. You are not being careless. You are working in a tool that gives you no option. Then you meet it later, in the payroll report, once it is too late. The average US manufacturing worker earns $30.37 an hour, so overtime pays $45.56 and every one of those hours costs you an extra $15.19. One person on 8 overtime hours a week is $6,319 a year, and on a floor where the rotation puts everyone over the line, that is every single name on it.
Every scheduling tool on the market will tell you about overtime. Read what they actually promise and you find the same four verbs, and not one of them stops anything.
“Track”
Adds the hours up after they have been worked.
“Flag”
Puts a colour on a name in a dashboard.
“Alert”
Sends a message to a supervisor who is out on the floor.
“Report”
Prints the total on Monday, when it is a receipt.
All four happen after the hours exist. By the time the dashboard turns a name red, the shift was assigned, the person worked it, and you owe the money. Counting is not preventing.
XShift AI is the only scheduling software that refuses the shift instead of reporting it.
You write one sentence. Nobody goes over 40 hours in a week. From that moment the schedule cannot be built against it. A supervisor who drops a name onto a shift that would carry that person past 40 does not get a warning. The shift does not save. If you run four crews around the clock, set it a little above 42 instead, because the rotation itself already puts everyone there and a ceiling below your own baseline would refuse every shift on Monday. And if you are in a state that counts overtime by the day, set a cap on hours in a single shift as well, because a weekly number on its own will not see what the day is doing.
What that looks like on a Tuesday morning: your maintenance lead is at 36 hours and has not cost you a penny of premium yet. A machine goes down and a supervisor puts him on a 12 hour cover shift, which would take him to 48. The shift comes back refused, on screen, with his running total and the number he would have hit. The supervisor is still standing there with a second name in front of them, and the plant has lost nothing. Those 8 overtime hours would have cost $122 on their own, on a Tuesday nobody would ever have looked at again.
The same check runs on every door into the schedule. A week the software generates for you. A shift a supervisor types in by hand. A hole filled at 4am after a call-off. One rule, written once, in your words.
What you get back
The same arithmetic, stopped. One person on 8 overtime hours is $6,319 a year. A 200 person floor where the rotation puts everyone over is
$1,263,808
a year, and $6,319,040 across five plants. On a 100 person floor with 40 people over it is $252,762, at 50 people it is $94,786, and at 20 it is $31,595.
2.A cap on days in a row, so nobody reaches the seventh
Why this puts you in overtime
Nobody is counting how many days in a row a name appears, because your tool counts shifts and not people. So the operator who says yes gets Monday, then Tuesday, then Thursday, then Saturday, and nothing joins those boxes up. On twelve hour shifts she crosses 40 hours partway through day four. Six days in one payroll week is 72 hours, and 32 of them are overtime. The average US manufacturing worker earns $30.37 an hour, so overtime pays $45.56 and every one of those hours costs you an extra $15.19. That is $486 out of one person in one week. Ten people doing that once a month is $58,330 a year, and across five plants, $291,648.
Ask any scheduling tool what it does about somebody working eleven days straight and you get one of two answers.
“Nothing”
Consecutive days are not counted at all. The calendar looks fine because each shift sits in its own box.
“A report”
You can pull a list of who worked the most days, after the month has closed.
Twelve on, one off, twelve on is how you lose a good operator. The body that never gets a clear day is the body that calls out next week, and that call-off is a twelve hour holdover somebody else has to absorb at premium.
XShift AI will not create the seventh shift in a row.
You set the most days anyone can work back to back. It is checked at the instant a shift is assigned, exactly like the hour ceiling, and it counts days worked anywhere in your business rather than days worked at one plant. A shift that would put someone on day seven is refused while the supervisor still has a rested name to pick instead.
Friday afternoon, a supervisor is filling next week and puts a packer on Monday because she is reliable and she said yes last time. She has already been given the six days before it. The shift is refused with the six days listed, and the supervisor moves it to somebody who has had a weekend.
It counts her Saturday at your other site as one of the six, so the plant that hands her the sixth day can see the five she already had.
What you get back
That same run, refused. 32 overtime hours at $15.19 is $486 out of one person in one week. Ten people doing it once a month is
$58,330
a year, and $291,648 across five plants, plus the operators you were going to lose to a rota that never let go.
3.A rest rule, so nobody comes back nine hours after they left
Why this puts you in overtime
A short turnaround is almost never a normal shift. It is a cover shift dropped on somebody who has already worked their week, which is exactly why the gap is short. An operator comes off nights at 7am Saturday and is back on at 7am Sunday, and both shifts land in the same payroll week as everything else she did. She was at 36 hours on Friday, so that one favour puts her at 48, and twelve of those hours are premium the moment she clocks in. At $15.19 an overtime hour on the $30.37 manufacturing average, that swap costs $182. Five a week on a 200 person floor is $47,393 a year, and $236,964 across five plants.
Rest between shifts is the gap almost every scheduling product leaves open, because it lives between two shifts rather than inside one.
“Two green boxes”
A 7pm finish and a 4am start are different days on the calendar, so nothing objects.
“A policy document”
The rule exists in a handbook nobody opens while covering a hole at 3pm on a Thursday.
A swap made in good faith can put an operator off at 7pm and back on at 4am. It is nine hours door to door, most of it not sleep, on a machine. And the tired shift is where the mistake and the next call-off come from.
XShift AI refuses the short turnaround before anybody is told they are working it.
You set the least rest anyone gets between the end of one shift and the start of the next. The check runs when the shift is assigned, so the turnaround never reaches a published schedule. You are not unpicking a week you already sent out, and you are never asking an operator to come back early as a favour.
A weekend swap would put a line operator off nights at 7am Saturday and onto days at 7am Sunday. The second shift is refused with the gap shown as 24 hours against your 10 hour minimum satisfied, but the pairing before it was 9. The swap is rebuilt around a third name in the same minute.
The gap is measured door to door, from the minute one shift ends to the minute the next begins, not by which calendar day each one falls on.
What you get back
Each of those swaps is 12 premium hours at $15.19, so $182 a time. Five a week on a 200 person floor is
$47,393
a year, and $236,964 across five plants, plus a rest standard that holds without anybody having to remember it.
4.Call-off coverage that routes around overtime
Why this puts you in overtime
At 4am your scheduling tool is not involved. Your spreadsheet is not involved. Ever. There is a supervisor, a phone, and a list that lives in their head, and they ring the people who always answer. Nobody is checking anybody’s hours at that hour. You do not have time to open a spreadsheet at 4am and work out who is at 34 hours and who is at 44, and most plants could not tell you even if they had the time, because the hours are spread across two systems and a paper board. So you call, somebody takes the shift, and you are in overtime. A twelve hour cover on top of a full week is twelve hours of premium, every hour of it. At $15.19 an hour on the $30.37 manufacturing average that is $182 a call. Three a week is $28,436 a year out of one habit, and $142,178 if you run five plants.
Every product has something for call-offs. What matters is who it puts in front of the supervisor at 4am.
“Blast the shift out”
Everyone gets the message and the fastest thumb wins, which is always the same three people.
“An open shift board”
The people already deepest into overtime are the keenest to pick it up.
An operator calls out and the supervisor rings the three who always answer. Those three are already carrying most of your premium. Every emergency makes the concentration worse, and one morning all three of them say no at once.
XShift AI works down the people who can take it without going into overtime first.
When a call-off lands it checks the plant they work at, whether they hold the role the shift needs, the days and the hours they marked unavailable, and anything they are already on that day. Only when there is genuinely nobody left does it reach for somebody who would go over, and it tells you that is what it is doing. A stopped line always costs more than a premium hour, so coverage still wins when it has to.
It is 4:10am and a forklift operator calls out of the 6am shift. Two names come back who hold the ticket, are clear that day and finish the week at 36 and 34 hours. The supervisor takes the first one. Nobody crosses 40 and the line starts on time.
It works the same list at 4am as it does at 4pm, so the person who happens to answer the phone is no longer the person who decides.
What you get back
Every one of those calls is 12 premium hours at $15.19, so $182. Three a week, routed to somebody not already in overtime, is
$28,436
a year out of one habit, and $142,178 if you run five plants. Your best three operators also stop being the answer to everything.
5.The Overtime Agent, on the week you are already in
Why this puts you in overtime
Your schedule was correct on Friday afternoon, and nothing has looked at it since. Somebody swapped a Tuesday. A supervisor added a cover shift on Wednesday. Two people picked up at your other plant on Thursday. Every one of those lands on a week that was already full, so every one is premium from the first minute, and not one passed anybody who was counting, because the counting happens in payroll and payroll runs afterwards. Four of those a week, twelve hours each at $15.19 of premium, is $729 a week. That is $37,914 a year on one plant and $189,571 across five, on a schedule you already approved.
The schedule you approved on Friday is not the schedule you are working by Thursday. Here is what the rest of the market does about that.
“A live dashboard”
Shows you the overtime building. Names no replacement and takes no action.
“A threshold alert”
Tells a supervisor somebody crossed the line, after they crossed it.
Knowing is not the hard part. Finding a specific person who can take a specific shift, at that plant, with that role, who is free that day and will not tip over the line themselves, is the hard part, and it is the part nothing else does.
XShift AI names the operator who can take the shift off you, and shows both costs before you decide.
You click run and the Overtime Agent reads the week in progress. It lists everyone heading past your threshold, and for the shift causing it, it names a specific replacement. Before that name goes forward it checks the plant, the role, the days marked off, the hours blocked, anything they are already on that day, whether they are salaried and kept out of overtime counting, and whether taking the shift would push them over themselves. Then it puts the two costs side by side and stops. It never moves anybody. A manager approves it or ignores it.
Wednesday, and a quality tech is on course for 56 hours because of two pickups nobody planned. The Agent shows the Saturday shift causing it, names a second tech who finishes at 32 hours, and prints what each version of the week costs. The manager approves it in one click and Saturday is somebody else.
It reads hours that person picked up at your other plants, which is where the overtime nobody can find usually lives.
What you get back
Four changes a week, twelve hours each at $15.19, is $729 a week. Taken back out before the week closes, that is
$37,914
a year on one plant and $189,571 across five, recovered inside the week rather than explained after it.
6.The labour cost, before payroll makes it real
Why this puts you in overtime
You approve a schedule with no price on it. The only system in your building that knows what a week costs is the one that pays for it, and it runs after the week you are looking at is already over. So the overtime was sitting in the schedule when you signed it. On a 200 person floor at 8 overtime hours a person, that is 1,600 overtime hours a week, and at $15.19 of premium each, $24,304 for the week. Nobody hid that number from you. It did not exist yet, in any screen you could open.
Most plants meet their overtime bill in a report, which is the one moment in the whole process when nothing can be done about it.
“Last month, by department”
Accurate, tidy, and about a week that is already paid for.
“A payroll export”
The first true number anybody sees, long after the decision that caused it.
You find out on Monday what Tuesday decided. Every lever you had was in the week before the one you are reading about.
XShift AI prices the week in front of you, before anybody has worked it.
Workforce Insights takes each person’s scheduled hours, splits anything past 40 out as overtime, pays that part at time and a half, and totals the floor. Salaried people you have marked as overtime exempt stay out of the count. You are reading the cost of a schedule you can still change.
It is Wednesday and next week is built. The screen says the floor costs $24,304 in premium alone. You move two shifts, run it again, and watch the number come down before a single person has been told what they are working.
Salaried people you have marked as overtime exempt stay out of the count, so the number is the one your operators actually generate.
What you get back
1,600 overtime hours a week at $15.19 of premium each, on the screen in front of you, for a week nobody has worked yet
$24,304
and you are reading it on Wednesday, while it is still a schedule you can change, instead of on the Monday after it became a receipt.
If you run more than one plant, there is a seventh leak
An operator picks up a Saturday at your second site. Each plant looks at its own board and sees a perfectly ordinary week for that person. Payroll adds both together into one weekly total and pays the premium. Nobody can find it afterwards, because the evidence does not exist on either schedule.
XShift AI counts a person against every hour they work anywhere in your business.
Not their hours at one site. Every hour, at every plant, inside the same payroll week. The ceiling, the consecutive day cap, the rest rule and the Overtime Agent all read that one total, so a name that is clear at your Ohio plant and at 44 hours across the group comes back refused at both.
What you get back
Five plants of 200, all running 8 overtime hours a week, is
$6,319,040
a year in premium, and the cross-plant share of it is the part no single site can even see today.
Frequently asked questions
How do you prevent overtime in manufacturing?
Six ways. Move the weekday your rotation cycle starts on, because overtime is counted inside a seven day payroll week and the same pattern opened on a different day produces four overtime hours a person or twelve. Set a weekly hour ceiling that is checked at the moment a shift is assigned, so a shift crossing it is never created. Cap how many days anyone works in a row. Set a minimum rest period between the end of one shift and the start of the next. Cover call-offs by working down the people who can take the shift without going over first. And scan the week already in progress, so the swaps and pickups that land after the schedule is published get caught before the week closes. XShift AI provides the capabilities to do all of this.
Why does a plant that runs around the clock pay overtime when nobody works extra?
Because covering every hour of the week takes more hours than a standard work week gives you. A line that never stops has to be staffed around the clock, and however you split that between crews, each person’s share lands above the 40 hour line unless you hire enough people that everybody works part time. So the overtime is written into the pattern before anyone does anything unusual. Nobody worked extra and nobody made a mistake. It was settled the day the crew count was chosen, and no amount of managing people on the floor changes it.
How much does manufacturing overtime cost per employee per year?
Price it off the top-up, not the whole hour. Production and non-supervisory workers in US manufacturing averaged $30.37 an hour in August 2026. Time and a half makes an overtime hour $45.56, so $15.19 of it buys no extra output at all. Four overtime hours a week comes to $3,160 a year per operator. Eight comes to $6,319. Twelve comes to $9,479. Multiply by the number of people on your floor who actually cross 40, not by your headcount. Five people over the line in a 20 person shop is $31,595 a year.
What is the average overtime in manufacturing?
Production and non-supervisory employees in US manufacturing averaged 4.0 overtime hours a week in August 2026. That figure includes every day shift plant that shuts at five on Friday, so it runs low for anyone comparing a continuous operation against it. A four crew plant covering all 168 hours starts at 42 hours a week before anything unusual happens, which is already two hours over the line. A plant sitting at 12 overtime hours a week is running at three times the national figure.
Can scheduling software refuse a shift instead of just reporting it?
XShift AI is currently the only scheduling software on the market that works that way. You write the ceiling in one sentence, and from that moment a shift that would carry somebody past it comes back turned away, with the arithmetic on screen, while a second name is still standing there. Everything else tracks, flags, alerts or reports, and all four of those happen once the hours already exist, which means the shift was assigned and the person worked it. Refusing means the check runs while the supervisor is still choosing a name, on a generated week, a hand-typed shift and a hole filled at 4am alike.
What causes overtime in manufacturing?
Five things, and only one of them is anybody working harder. First, the coverage pattern itself: a line that runs around the clock needs more hours staffed than a standard work week gives you per person, so the schedule is over the line before anything goes wrong. Second, where your cycle sits against your payroll week, which can triple the same rotation’s overtime depending on which weekday it opens on. Third, holdovers, when the oncoming crew is a body down and whoever is standing there keeps working. Fourth, shifts picked up at your other plants, which are invisible on both boards and land in one payroll total. Fifth, the three or four operators who always answer the phone at 4am and end up carrying most of the premium on the floor.
What does XShift AI actually do to prevent overtime?
Six things, and four of them happen before a shift exists. A weekly hour ceiling you write as one sentence, checked at the instant a name is dropped onto a shift, so a shift that crosses your limit never saves. A cap on days worked in a row, so nobody quietly reaches a seventh straight twelve hour day and crosses 40 partway through day four. A minimum rest rule between shifts, so the 7am finish and 7am start that turns a cover shift into twelve premium hours is refused before anybody is told about it. Call-off coverage that works down the people who can take the shift without going into overtime first, checking the plant, the role, the days and hours they marked unavailable and anything they are already on that day, and only reaching for somebody who would go over when there is genuinely nobody left. Then the Overtime Agent, which reads the week already in progress, lists everyone heading past your threshold, and for the shift causing it names a specific operator who could take it instead, with the cost of both options side by side for a manager to approve or ignore. Finally Workforce Insights, which prices the schedule before it is worked by splitting each person’s hours past 40 out as overtime and paying that part at time and a half. All six read a person’s total hours across every plant you run rather than one site, so a pickup at your second plant is caught rather than hidden.
How do you calculate overtime cost for a manufacturing plant?
Price the premium, not the whole hour. Take your average hourly wage and halve it, because time and a half means the extra half is the part that buys you no additional output. On the US manufacturing average of $30.37 an hour, an overtime hour pays $45.56 and the premium is $15.19. Multiply that premium by the overtime hours one person runs in a week, then by 52, then by the number of people on your floor who actually cross 40 rather than by your headcount. Eight overtime hours a week is $6,319 a year per person. Five people over the line in a 20 person shop is $31,595. Fifteen in a 50 person plant is $94,786. Forty in a 100 person plant is $252,762. All 200 on a four crew continuous floor is $1,263,808, and five plants of 200 is $6,319,040. One thing that formula does not cover: federal law counts hours past 40 in a week, but several states, including California, Alaska, Colorado, Nevada, Oregon and Kentucky, also count hours past a set number in a single day. That matters on a twelve hour rotation, because those shifts can generate overtime before the week reaches 40 at all. California goes furthest and is the only state with double time. Check your own state before you trust any weekly figure.
What happens when an operator picks up a shift at a second plant in the same week?
Payroll adds both plants into one weekly total for that person, while each plant looks at its own board and sees an ordinary week. The evidence does not exist on either schedule, so it only shows up in a tool that counts a person against every hour they work anywhere in the business, which is how XShift AI counts them. The ceiling, the consecutive day cap, the rest rule and the Overtime Agent all read that one company wide total, so a name that looks clear at your Ohio plant and sits at 44 hours across the group comes back refused at both.
Does preventing overtime mean running a line short?
No. A ceiling that emptied a machine would be worse than the problem it solved. The ceiling governs the schedule you build in advance, while you still have choices and time to use them. On the day itself, when somebody calls out, coverage wins, because a stopped line costs far more than a premium hour ever will. XShift AI works through the people who can take the shift without going into overtime first, checking the plant, the role, the days and hours they marked off and anything they are already on, and only falls back to somebody who would go over when there is genuinely nobody else. The line keeps running and the premium stops landing on the same three names.
How many hours can a factory worker work in a week?
There is no federal limit. The Department of Labor states it plainly on its overtime fact sheet: “There is no limit in the Act on the number of hours employees aged 16 and older may work in any workweek.” What the law does require is pay: covered employees must receive overtime for hours worked past 40 in a workweek at not less than time and one-half their regular rate. Some states set their own hour limits and some of those single out manufacturing, so check yours. The number your floor actually runs is the one you set, which is why a weekly ceiling is worth writing down and enforcing rather than leaving to whoever is covering a hole.
Where is a plant’s overtime usually concentrated?
Pull one full quarter of payroll and rank premium hours by name, highest first. Almost every plant finds the same shape: a long flat tail, and three or four operators carrying an enormous share, because they are the ones who answer the phone at 4am. That is not a 200 person problem, it is a four person one, which changes the job from taking hours away to putting them somewhere else.
Plant terms used on this page
- Overtime premium
- The top-up part of an overtime hour, not the whole hour. Time and a half makes it half the base rate, which is $15.19 on the average US manufacturing wage.
- Payroll week
- The seven day window your hours are sorted into before anyone works out what is owed. Any weekday can open it, and once chosen it stops moving.
- Holdover
- When an operator keeps working past the end of their own shift because the oncoming crew is a body down. On a 12 hour rotation that is 12 premium hours for whoever stayed.
- Continuous operations
- A line that never stops, so all 168 hours of the week have to be staffed. The reason a plant is in overtime before anybody does anything unusual.
- Weekly hour ceiling
- A limit on how many hours one person can be given in a week, checked when the shift is assigned rather than when the timesheet closes.
- Overtime exempt
- A marker on a person, usually a salaried superintendent or planner, that keeps them out of overtime counting entirely.
- Crew
- One of the groups a plant splits its people into so a line can be covered around the clock. Four crews is the usual answer on a twelve hour rotation.
- Cycle
- The full pattern of days on and days off before the rotation repeats. It can be any length, and it rarely divides evenly into seven day payroll weeks.
- Short turnaround
- When somebody finishes one shift and starts the next after very little rest. It is almost always a cover shift dropped on a person who has already worked their week.
- Daily overtime
- Overtime counted by hours worked in a single day rather than hours in the week. A handful of states use it, and it changes the arithmetic on a twelve hour shift.
Related reading
Stop paying for hours you never agreed to
Write your ceiling in one sentence and watch the next shift that breaks it get turned away. $29 a month plus $1 per active user, twenty-one days free, and you can set it up this afternoon.