There is a good chance you paid overtime wrong last week. Possibly too much, possibly too little. Either way, it almost certainly was not your team’s fault.
It comes down to how overtime gets calculated when a worker sits on two different shifts at two different rates in the same week. Most agencies handle it the same way, and that way produces a number that is either slightly too high or slightly too low depending on when in the week the worker crossed hour 40.
Slightly too high costs you margin. Slightly too low is a compliance problem.
Here is the whole thing, worked out.
Start with what a shift differential actually is
A shift differential is extra pay for working a less desirable shift. Nights, weekends, overnight, holiday coverage. It sits on top of the base rate for the role.
A picker earns $20 an hour on days and $22 an hour on nights. That $2 gap is the differential. Simple enough on paper.
It stops being simple the moment one worker is on more than one shift, at more than one rate, in the same week. In light industrial staffing, that is not an edge case. It is Tuesday.
Warehouses, manufacturing plants, distribution centres and 3PLs run days, nights and weekends, usually with a premium attached to each. A single worker gets placed on two different shifts, sometimes across two different clients, sometimes spanning two pay periods, inside the same seven days. Multiply that across a few hundred workers on assignment and you have a back office spending its week sorting hours into buckets by hand.
That sorting is the part everyone sees. The overtime calculation sitting on top of it is the part that quietly costs money.
The overtime trap
Once a worker crosses 40 hours in a week and those hours sat at more than one rate, you have to decide what rate the overtime gets paid at.
Most agencies answer that question the intuitive way: pay it at whatever rate the worker happened to be on when they crossed hour 40. If they were on nights, overtime gets calculated at the night rate. If they were on days, it gets the day rate.
That method feels reasonable. It is also wrong, and the direction it is wrong in depends entirely on how the week happened to fall.
If overtime lands on the premium shift, you overpay. Perfectly legal. It just comes straight off your margin.
If overtime lands on the base shift, you underpay. In the US, that is not FLSA compliant.
Same worker. Same hours. Same rates. Two different outcomes, decided by scheduling luck.
What the FLSA actually requires
Under the Fair Labor Standards Act, when a worker earns more than one rate in a workweek, overtime is calculated on the blended average of all the rates they earned that week. Not the rate they were on at hour 41.
You take all straight-time earnings for the week, divide by all hours worked, and that gives you the regular rate. The overtime premium is calculated from there.
The method produces one figure. It does not move based on shift sequencing.
The math, worked out
Take a worker who logs 48 hours in a week. Thirty of those hours are on days at $20 an hour. Eighteen are on nights at $22 an hour. That puts them 8 hours into overtime.
Three ways to calculate the same paycheque.
Method one: overtime lands on nights ($22)
Regular 40 hours: (30 × $20) + (10 × $22) = $820 Overtime: 8 × 1.5 × $22 = $264 Total: $1,084.00
Method two: overtime lands on days ($20)
Regular 40 hours: (18 × $22) + (22 × $20) = $836 Overtime: 8 × 1.5 × $20 = $240 Total: $1,076.00
Method three: FLSA blended rate
Straight time on all 48 hours: (30 × $20) + (18 × $22) = $996 Blended regular rate: $996 ÷ 48 = $20.75/hr Overtime premium: 0.5 × $20.75 × 8 = $83 Total: $1,079.00
The correct figure is $1,079.00.
Method one overpays by $5. Method two underpays by $3. Neither of those is a rounding error your team introduced. Both are the arithmetic working exactly as designed, on a design that does not match the law.
And notice what the blended rate does: it produces $1,079.00 whether the worker’s overtime hours happened to fall on days, on nights, or split across both. The number does not care how the week was scheduled. That is the entire point of the method.
Five dollars, and why it is not five dollars
A $5 overpayment on one worker in one week is nothing. Nobody is losing sleep over five dollars.
The problem is that this is not one worker in one week. It is every worker who picked up mixed-rate hours, every week, across every client site you staff. It is systematic, and systematic is what turns small numbers into real ones.
Run an illustrative model. Take an agency with 300 workers on assignment, where roughly 40 percent of them land in multi-rate weeks. That is 120 affected workers. Assume an average weekly overpayment of about $4 each.
That is roughly $25,000 a year walking out the door in overpayments alone.
Now add the labour. If each affected worker takes about 10 minutes of back-office time per week in preparation, checking and corrections, at a loaded payroll cost of around $35 an hour, you are looking at roughly 1,000 hours a year and about $36,000 in paid time.
Combined, about $61,000 annually, on an illustrative 300-worker model. Your numbers will differ. Plug in your own headcount, your own multi-rate percentage, your own loaded cost. The structure of the calculation holds regardless of what you put into it.
Two things worth sitting with about those figures.
The $25,000 is gross margin. In a take-rate business, that is the most expensive dollar you have. Replacing one dollar of leaked margin takes several dollars of new billings, which means new orders, new fills, new client acquisition. It is a genuinely bad trade.
The $36,000 is paid time, but it is also capacity. Those thousand hours are hours your back office did not spend on redeployment outreach, onboarding, or client service. The work that actually grows an agency.
The third cost, which is not on the P&L
When overtime lands on the base shift and you underpay, that is not a margin question. That is an exposure question.
Underpayment under the FLSA carries back-pay liability across the lookback period, potential liquidated damages, and legal cost. And because the underlying cause is a calculation method rather than a one-off mistake, the exposure is not a single worker in a single week. A claim or an audit scales across every affected worker and every affected week.
That does not show up anywhere on your financials. It sits as risk until it does not.
Canadian agencies are not off the hook here either. Overtime rules vary by province, and the mechanics of the calculation differ from the FLSA. But the underlying problem is identical: one worker, multiple rates, one week, and a calculation that has to allocate hours correctly before it can produce a defensible number.
Why “be more careful” does not fix this
The instinct, when a payroll error surfaces, is to add a check. Another review step. A second set of eyes on Thursday.
That works for mistakes. It does not work for methods.
If your process pays overtime at whichever rate the worker was on at hour 40, then a careful person following that process correctly still produces the wrong number. Reviewing it more closely just confirms the wrong number faster. The error is upstream of anyone’s attention to detail, which is why it survives every process improvement your team has tried.
It is also why the problem gets worse as you grow rather than better. More workers means more mixed-rate weeks. More client sites means more differentials to track. More volume means more hours of manual allocation before you even reach the overtime question. The leak scales with you, and it consumes exactly the back-office capacity you need in order to scale.
How Jombone handles it
This is the part of the process we built to disappear.
Jombone’s shift differential engine recognises every shift and differential attached to a worker, including when those shifts span separate pay periods or sit with different clients. Hours are allocated to the correct shift and rate automatically, with no weekly setup and no re-sorting of timesheets.
When a worker crosses 40 hours, blended-rate overtime is applied automatically. One correct figure, produced the same way every week, regardless of how the shifts happened to fall. For Canadian agencies, the correct provincial overtime rules are applied as part of a one-time setup.
What that removes is the re-work loop. No cascading corrections, no re-running payroll, no off-cycle payments, no compliance guesswork on Friday afternoon.
Setup happens once. After that, the math is just right.
Agencies running automated time capture and pre-payroll validation on the platform operate at over 97 percent timesheet accuracy and under 2 percent payroll exceptions. Shift differentials are one of the specific reasons those numbers hold at volume.
What to check this week
You do not need a platform to find out whether this is costing you. You need about twenty minutes.
Pull last week’s payroll. Find the workers who logged hours at more than one rate and crossed 40 hours. For each one, calculate the blended rate: total straight-time earnings divided by total hours. Compare the overtime you actually paid against 1.5 times that blended figure.
Then count how many workers that applies to in a typical week, and multiply.
Most agency owners have never run that comparison. The ones who do tend to find the same thing: it is not a big number per worker, and it is a much bigger number than they expected in total.
See it handled automatically.
Book a 30 minute walkthrough and bring your most complicated multi-rate week. We will show you what it looks like when the platform does the math.