Your Labor Cost Is a Scheduling Problem
Answer first: labor cost is not primarily a wage problem, and it is almost never solved by cutting hours across the board. It is a distribution problem — the right number of hours placed in the wrong parts of the week. Most operations I assess are simultaneously overstaffed on their slowest shifts and understaffed at peak, which produces the worst possible outcome: high labor cost and poor service, at the same time, from the same schedule.
The National Restaurant Association reports that salaries and wages including benefits reached a median of 36.5% of sales for full-service restaurants in 2024, against roughly 33% in its earlier benchmark years (source). Labor is structurally more expensive than it was. That makes the distribution question more consequential, not less.
What is a normal labor cost percentage for a restaurant?
Use the NRA median as a reference point, not a target: 36.5% of sales for full-service, 31.7% for limited-service in 2024, compared with historical averages near 33% and 28% respectively.
Two cautions before that number gets used to make a decision.
First, the definition has to be complete. Total labor means hourly wages, salaried management, payroll taxes, benefits and workers’ compensation. Operations that quote a flattering labor percentage have usually excluded salaried management, which is the least variable and often the largest single piece. The same rule governs prime cost, where labor is half the number.
Second, the right percentage depends on the model. A high-volume, simplified-production concept and a fine dining room with a full brigade cannot share a target. What they can share is a discipline: labor measured weekly, against a forecast, by shift and by department — not monthly, in aggregate, after the fact.
A single monthly labor percentage is a historical record. It tells you what happened. It does not tell you which Tuesday caused it.
Where does labor cost actually leak?
Rarely at peak. Almost always at the edges.
The pre-shift hour. Cooks and servers clocked in before there is work matched to them. Fifteen minutes per person per shift is small on a timecard. Across a crew of twelve, two shifts a day, it is six hours of wages every day with nothing assigned to them, and it is invisible in every report except the one that compares clock-in times to prep-list start times.
The shoulder. The hour between lunch and dinner, and the hour after the last table sits. Staffing that does not taper is the single most common leak, and the most tolerated, because the room still looks busy to the person standing in it.
Over-scheduled prep. Prep hours assigned by habit rather than by par level. When par levels are undefined, prep expands to fill the shift, and the surplus becomes waste twice — once in labor, once in product.
Unmanaged overtime. Overtime that is discovered at payroll rather than managed on Thursday. By the time it appears in the report, it has already been earned.
Rework. The most expensive labor in any kitchen is the labor spent making something a second time. It is never coded as labor cost; it appears as food cost, and so it never gets fixed by the person looking at the labor line.
How do you build a schedule against a forecast?
A schedule written from last week’s schedule will reproduce last week’s errors indefinitely. A schedule written from a forecast will not.
The method is unglamorous and it works:
Forecast covers by day-part, not by day. Use trailing four-week actuals adjusted for known events — occupancy on the books, weather, holidays, anything in the building.
Convert covers to production hours using your own measured output, not a benchmark. How many covers does one competent cook on that station actually produce per hour in your room?
Place the hours against the curve, with staggered starts and staggered cuts. Nobody who does not need to be there at 4:00 should start at 4:00.
Fix the floor. Every operation has a minimum crew below which service fails regardless of covers. Name it, and stop scheduling below it — cutting into the floor is how a labor saving becomes a revenue loss.
Compare forecast to actual every week, by shift. The variance is the whole point of the exercise. Two or three cycles of honest variance review will sharpen a forecast more than any software.
Staggered starts alone recover meaningful hours in operations that have never used them, and they cost nothing but attention.
What is the right way to measure productivity?
Labor percentage is a lagging outcome. It moves when sales move, which makes it a poor instrument for managing a shift.
Three measures are more useful because they can be read shift by shift:
Sales per labor hour (SPLH) — total sales divided by total hours worked. Track it by shift and by department. It exposes the shoulder shifts immediately.
Covers per labor hour — the same idea, stripped of check average, so a price increase cannot flatter it.
Hours per cover, by station — the most operationally precise of the three, and the one that tells a chef where to intervene rather than that they should.
Track those weekly, by shift, and the schedule starts to correct itself, because the conversation moves from “we need to cut labor” to “Tuesday dinner is running at half the SPLH of Thursday, and here is why.”
How do you cut labor cost without cutting service?
By removing work, not people. There are four honest levers, and they are all upstream of the schedule.
1. Simplify the menu. Every item is prep time, station load, training load and waste exposure. Menu reduction is the largest labor lever most operations have and the one they reach for last. It is also free.
2. Move work off the line. Anything that can be produced in a batch, earlier, by fewer people, should be. À la carte hours are the most expensive hours in the building; production hours are the cheapest.
3. Standardize the tasks. A cook working from a written station set-up with quantities does it once. A cook working from memory does it approximately, and somebody redoes it. Systems are a labor instrument before they are a quality instrument — which is why kitchen SOPs that actually get used show up on the labor line first.
4. Reduce turnover. Recruiting, onboarding and the productivity gap of a new hire are all labor cost, and none of it appears in the labor line as such. A cook who stays two years is cheaper per hour than a cook who stays four months at the same wage — the difference is real and it is never on the report.
What is not on that list: cutting a shift below the floor, or trimming hours uniformly across a schedule. Uniform cuts remove the same percentage from the shifts that needed the hours and the shifts that did not, which is how an operation ends up with a better labor percentage and a worse business.
The structural point
Labor is the only major cost in a restaurant that is scheduled in advance, by a person, every week. That makes it the most controllable line on the P&L and the most frequently mismanaged one.
The operations that control it are not the ones paying less. They are the ones that decided their schedule is a forecast rather than a habit — and that reviewed the variance every week until the forecast got good.
Cut the work first. The hours follow.
Frequently asked questions
What percentage should labor be in a restaurant? The National Restaurant Association’s 2024 medians are 36.5% of sales for full-service and 31.7% for limited-service, against historical averages near 33% and 28%. Treat these as reference points. The model, the wage market and the service style all move the correct figure, and a number without a definition of what is included in it is not comparable to anything.
Should labor be measured weekly or monthly? Weekly, by shift. Monthly labor tells you what already happened and averages away the variance you needed to see. Weekly labor is short enough to act on while the week can still be changed.
Is labor cost or food cost the better place to start? Whichever one is further from its own recent trend. Both belong inside prime cost, and moving one without watching the other usually just relocates the problem — cheaper product almost always costs more labor.
Does scheduling software solve this? It makes a good process faster and a bad process faster. Software cannot supply the forecast discipline, the measured output rates, or the decision about where the staffing floor sits. Those are operator decisions, and they are what actually moves the number.
Paul Forciniti is a restaurant and hotel operations consultant with 25+ years in professional kitchens — line cook in Buenos Aires, culinary training in France, executive chef of a five-star international hotel, owner-operator, and author of Your Food Is Not the Problem and From Concept to Cover.
Want to see where your hours are actually going? Every engagement starts with an operational assessment.
Comments