How to Cut Restaurant Overtime Without Losing Coverage

Overtime is a scheduling outcome, not a payroll line

By the time overtime reaches you it looks like a payroll problem, which is why it usually gets treated like one. Someone circles the number, tells the management team to bring it down, and for two weeks it does. Then it drifts back, because nothing about the schedule that produced it changed.

Every overtime hour was created days earlier, at the moment someone wrote a rota with no room in it, or failed to train a second person on a station, or set a prep list against a guess rather than a forecast. The payroll report is where the decision surfaces, not where it happens. Any attempt to fix overtime inside payroll is an attempt to fix it in the wrong place and at the wrong time.

The money is worth the attention. A team running 40 overtime hours a week at an $18 average base rate and a 1.5× multiplier pays roughly $22,000 a year in premium alone, before any burden reduction, and that premium buys nothing at all — the hours would have been worked either way. You can put your own figures into the overtime cost calculator to see where your site lands.

Find out where the hours actually go

The instinct on seeing a high overtime number is to act immediately: cut shifts, send people home earlier, tell managers to stop approving hours. This reliably fails, because without knowing which shifts and which causes produced the overtime, effort goes to whatever is most visible rather than what is most expensive.

Spend two weeks tagging instead. Every overtime hour gets one line: which shift, which person, and one cause from a short fixed list — absence cover, late finish, understaffed roster, prep overrun, or event. Five options, no more. Longer lists produce worse data because managers stop filling them in during service.

What you find is almost always lopsided. Most sites discover that half their overtime comes from two shifts a week, or from covering one person's absences, or from a closing routine that has quietly grown twenty minutes longer than the roster allows. That concentration is good news: it means one or two targeted changes recover most of the money, rather than a general campaign against overtime that irritates everyone and saves little.

Do the tagging before the fixing. Two weeks feels slow when there is a number on a page demanding action, but acting on a guess costs a quarter and teaches you nothing.

The four causes worth fixing

Rotas written with no slack

If everyone is scheduled to their full contracted hours before the week begins, you have no capacity left to absorb anything. One person calling in sick means the hours have to be bought at a premium, because there is no straight time available to buy them with. Deliberately leaving a few contracted hours unrostered at the start of the week gives you somewhere to put an absence without paying extra for it.

Skill concentration

When only two people can run the grill, only two people can cover the grill, and one of them works every gap the other leaves. Skill concentration is the quietest cause of overtime and the most expensive to leave alone, because it also means one resignation puts the station at risk. Cross-training against your specific bottlenecks — not general development, but named people on named stations — is a labour cost control disguised as a training initiative.

Late finishes nobody rostered

Fifteen minutes past the scheduled end, five nights a week, across four people, is five hours of overtime that never appeared on any schedule and never got a decision made about it. Staggering start and finish times by thirty minutes usually removes most of this, because the cause is normally that everyone begins and ends together, leaving the close under-covered and the mid-shift over-covered.

Prep against habit rather than forecast

Kitchens that prep the same quantities regardless of the week ahead run long on quiet weeks and short on busy ones. Both cost money, but the busy-week shortfall gets covered with overtime while the quiet-week overrun turns into waste. Tying prep quantities to a forecast fixes two problems with one change — and if waste is also an issue, the food waste cost calculator will show you the other half of the bill.

When overtime is the cheaper option

Not all overtime should be eliminated, and treating zero as the target leads to worse decisions than tolerating a sensible amount. The comparison that matters is between the overtime rate and the fully loaded cost of adding a person.

The wage side is simple: at a 1.5× multiplier on an $18 base, any cover you can hire below $27 an hour is cheaper per hour. What that comparison leaves out is everything a new hire costs before they are useful — recruitment time, training hours that produce nothing sellable, onboarding admin, and the real possibility that they leave within a quarter and you pay all of it again. The employee turnover cost calculator puts a number on that risk, and it is usually larger than people assume.

The practical test is predictability. If the same hours appear in the same shifts every week, that is a headcount gap you are financing at a premium, and hiring wins. If the hours scatter unpredictably across the schedule and the total varies week to week, overtime is doing exactly what it is for: buying flexibility you cannot roster in advance. Paying a premium for genuine flexibility is a reasonable trade. Paying it for a permanent gap is not.

Fixes in order of return

• Require advance approval. Overtime that has to be authorised before it is worked drops immediately, purely because it becomes a decision rather than a default. Cheapest change on the list.
• Roster to a forecast. Building the schedule against expected trade rather than last week's schedule removes the largest structural source of premium hours.
• Hold deliberate slack. A few unrostered contracted hours at the start of the week is where absences go instead of into overtime.
• Stagger starts and finishes. Removes most late-finish overtime at zero cost, because it redistributes hours rather than adding them.
• Cross-train against named bottlenecks. Slower to pay back, but it fixes overtime and single-point-of-failure risk at the same time.
• Add part-time cover for the repeating gap. The right answer once tagging shows the same shift short every week, but only after the cheaper changes have been made.
• Review monthly. Overtime returns the moment nobody is looking, so the number needs a permanent home in your monthly review alongside labour cost percentage.

Work down that list rather than jumping to the bottom. Hiring is the most visible action and the one most operators reach for first, but it is also the one with the longest payback and the most ways to go wrong. The first four changes cost nothing and usually recover more than half the premium.

The mistake to avoid

The fastest way to make overtime disappear is to stop letting people stay. Send everyone home at their rostered finish time, refuse all approvals, and the number falls next week.

It is also the change most likely to cost more than it saves. The hours were being worked for a reason: the close was not finished, the prep for tomorrow was not done, the last table had not left. Removing the hours without removing the work pushes it onto the following shift, which then runs long itself, or it simply does not get done and the guest notices. You have converted a payroll number into a service problem, and service problems show up in like-for-like sales months later where nobody connects them back to the rota change that caused them.

The goal is to buy the same hours at straight time instead of at a premium — not to buy fewer hours. Every fix above targets the rate, the timing or the cause. None of them ask the kitchen to do the same job in less time and hope nobody notices.

Frequently asked questions

What is a reasonable overtime target for a restaurant?

Under about 2% of total scheduled hours is normal absorption of absences and late finishes, and chasing it below that usually costs more in rigidity than it saves. Above 5% you are almost certainly covering a repeating gap at a premium. Your own trend matters more than the benchmark.

How quickly can overtime actually be reduced?

Approval controls and staggered finishes show up in the next payroll run. Forecast-based rostering takes a month or two to bed in. Cross-training and hiring take a quarter, because the benefit only arrives once the new capability is reliable enough to be scheduled against.

Is it cheaper to hire someone than to pay overtime?

It depends on predictability, not just on the hourly rates. Repeating, forecastable hours are cheaper to cover with a hire. Scattered, unpredictable hours are cheaper to cover with overtime, because a new hire carries recruitment, training and minimum-shift costs that a few irregular hours will never repay.

Does cutting overtime damage morale?

It can, and it is worth being direct about it. Some of your team rely on overtime as income and will feel the reduction as a pay cut. Where that is true, say so openly and look at whether contracted hours can be adjusted upward for the people who want them, which costs you less than the premium did and gives them a more stable income.

How should overtime caused by absence be handled?

Cover it, then treat the absence as the problem rather than the overtime. Persistent absence concentrated in one or two people is an HR conversation, and persistent absence spread across the team is usually a scheduling or workload signal. Either way, blaming the overtime line addresses the symptom.

Should managers approve their own overtime?

Not without visibility. The point of an approval step is not distrust — it is that someone has to consciously choose the spend rather than let it happen. A simple rule that overtime is authorised before it is worked, by someone other than the person working it, is the cheapest control available.

Does overtime affect prime cost?

Yes, through the labour half. Overtime raises total labour cost without raising sales, so labour percentage and prime cost percentage both move up together. It is one of the more common reasons a labour percentage drifts while headcount, menu prices and covers all stay flat.

Run the numbers

Use the free Overtime Cost Calculator to apply everything above to your own figures.