The formula, and the two versions of it that get confused
Labour cost percentage is labour cost divided by sales for the same period. Both halves have to cover the same days or the number means nothing, which is the single most common way the figure gets misreported: a payroll period that runs Monday to Sunday set against a sales week that runs Sunday to Saturday produces a percentage that is wrong by whatever the busiest day contributed.
The second confusion is what goes in the numerator. Scheduled hours multiplied by wages is not the cost of labour, it is the cost of wages. On top of it sit the employer share of payroll taxes, unemployment insurance at your own experience rate, workers compensation at your class code, and whatever benefits you provide. That total is what leaves the bank account, and the gap between it and the wage figure is large enough to move the percentage by several points. This page shows both, so that when someone quotes you a percentage you can ask which one it was.
The burden rate to use is yours, from your own payroll provider and your own policy notices. It varies by state, by industry classification and by your own claims history, and there is no general figure worth putting in a calculator. The employee cost calculator builds it up line by line for an individual person if you need to derive it rather than look it up.
Why sales per labour hour is the better management number
Labour percentage has a structural flaw as a scheduling tool: it changes when sales change even if the schedule did not. A quiet week produces a bad percentage from a schedule that was perfectly reasonable, and a busy week flatters a schedule that was overstaffed. Since you write the schedule before you know the sales, judging it afterwards by a ratio that moves with the denominator tells you as much about the weather as about your decisions.
Sales per labour hour has the opposite property. It divides sales by hours rather than cost by sales, so it measures how much business each scheduled hour produced. It compares directly across weeks of different size, across locations, and across days. Most operators who watch it seriously watch it by daypart rather than by week, because the useful question is which shifts are carrying hours they did not need, and a weekly figure averages a bad Monday into a good Saturday.
Covers per labour hour is the same idea in a different unit and is better where the average check varies a lot. Neither replaces the percentage, which is what the profit and loss will be read on. They are the numbers to schedule by; the percentage is the number to report on.
Fixed hours and variable hours are different problems
A schedule is not one pool of hours. Some of it is fixed: the manager, the opener, the person who has to be there whether four people come in or forty. The rest flexes with volume. When sales come in below forecast, only the variable part can be cut, and cutting it has service consequences well before it has cost consequences.
| Hour type | Moves with sales | What to do when the week is soft |
|---|---|---|
| Salaried management | No | Nothing this week — it is a fixed cost decision |
| Openers, closers, single-cover positions | Barely | Trim the tails of shifts, not the shift |
| Second and third line positions | Yes | The first place a cut is available |
| Floor staff above the minimum section count | Yes | Cut sections, which cuts hours in blocks |
| Prep hours | Yes, with a lag | Batch sizing rather than headcount — see the batch calculator |
This matters because the hours figure this calculator produces from a target percentage is a single number, and you cannot remove hours evenly across those rows. Take the allowed hours as the ceiling and then work out which rows can actually give them up.
Labour and food trade against each other
Every decision that lowers one raises the other. Buying prepped product raises food cost and removes prep hours. Making sauces in house lowers food cost and adds them back. A menu with thirty items needs more prep hours and produces more waste than a menu with twelve. Judging either percentage in isolation will push you toward whichever one you happen to be looking at, which is why the two get watched as a pair.
The pair total is worth tracking as one line, since it is the largest controllable block in the business and the split between its halves is a choice rather than a fact. When it is moving, the useful question is whether the two halves moved together or traded, because a trade is a decision working as intended and a joint rise is a problem. The food half by dish is on the menu pricing calculator, and the whole monthly picture including rent and everything else is on the cafe and restaurant profit calculator.
Everything about pay rules is outside this page
The calculator multiplies hours by the rate you entered and adds the burden percentage you entered. It applies no overtime rule, no tip credit, no tipped minimum wage, no service charge treatment, no split shift or reporting time provision, no meal or rest break premium, and no predictive scheduling requirement. Those exist, several of them apply to almost every food business, and they are set by a combination of federal law, your state, and in many cases your city, with the tipped wage rules in particular varying enormously between states and carrying conditions that are easy to get wrong.
None of it belongs in a general calculator, because a figure that is right in one state is a liability in another and the rules change. Your state labour department publishes what applies to you, and a payroll provider or an accountant who works with food businesses in your state is the practical route. Use this page for the shape of a schedule and its cost, and use them for whether the schedule is lawful and what the hours really have to be paid at. For event and catering work where staffing is quoted rather than scheduled, the catering quote calculator handles the same labour block as part of a price.
Questions people ask
What is a good labor cost percentage for a restaurant?
There is no figure that survives being quoted without context, and the ranges that circulate cover formats so different that the average is meaningless. A counter-service operation with a limited menu, a quick-service model with heavy equipment, a full-service dining room with table service and a bar, and a fine dining room with a large brigade all have structurally different labour models, and each is right for itself. Rent as a share of sales matters too, because labour and rent are competing for the same space under the gross margin. The useful approach is to build a monthly model with your own fixed costs and find the labour figure that leaves an acceptable profit, then treat that as the target. What matters afterwards is the direction it moves, not how it compares to somebody else.
Should the owner or manager salary be in the labour number?
It should be somewhere, and where you put it changes what the percentage means. Including salaried management gives you total labour, which is the honest cost of running the place and the right figure for a profit and loss. Excluding it gives you variable or hourly labour, which is more useful for judging a schedule week to week because it only contains what you can actually change. Most operators track both and are careful to say which one they are quoting. The one thing that is always wrong is leaving an owner who works a full week out of the calculation entirely, because it produces a labour percentage that only holds while that person keeps working unpaid.
How is scheduled labour different from actual labour?
The schedule is a plan and payroll is what happened. People clock in early, stay late, get cut early, pick up shifts, call out and get covered. The gap is routinely several percent of the schedule and it runs in one direction, upward, because the deviations that add hours are easier than the ones that remove them. This page costs the plan, which is the version you can still change. Comparing it against what payroll actually reports for the same week is the exercise worth doing, because a consistent gap tells you the schedule is not the binding document you thought it was, and that is usually a clock-in and cut-time problem rather than a scheduling one.
What does the break-even cover count actually mean here?
It is the number of covers whose contribution, after food and beverage cost, exactly pays for the loaded labour in the schedule. It answers a narrow question: does this schedule pay for itself. It is not a break-even for the business, because it carries none of the rent, utilities, insurance, fees or anything else. Those come out of what is left after labour, which the output shows as a separate line. If the forecast covers barely clear the labour break-even, the week loses money regardless of what the labour percentage looks like. A full break-even including fixed costs is a different calculation and is on the break-even sales calculator.
How do tips affect this calculation?
They do not appear in it at all, deliberately. Tips flow to staff and are not an employer labour cost in the way wages are, but the rules around them are the most jurisdictionally variable part of food service pay: whether a tip credit against minimum wage is available, at what rate, what conditions attach to it, how pooling and sharing may work, who may participate, and how service charges are treated are all matters where states differ sharply from each other and from the federal position, and where getting it wrong is expensive. This calculator asks you for the wage you actually pay per hour and does nothing else with it. Work out what that wage has to be with your state labour department and your accountant, then enter the result.