Restaurant Metrics and KPIs:
How to Calculate All 10, and How to Hit the 5 Your Schedule Sets

What are restaurant KPIs?

A restaurant KPI is a metric you have given a target and a day of the week to check it. KPI is short for key performance indicator, so restaurant key performance indicators are the few numbers you have promised to watch. A metric is any number your business produces. A KPI is one you have decided to act on, so every KPI for a restaurant is a metric, but most metrics never become a KPI.

A number you only see when the accountant sends it is a metric, and by then nobody can change the week that made it. Give it a target and a day to check it, and a bad week gets caught while next week can still be changed.

For example, the restaurant this guide uses is one full-service location with a $100,000 week of net sales. It pays its hourly staff about $20 an hour, close to the $20.20 the Bureau of Labor Statistics reported for non-supervisory restaurant workers in July 2026. Every section below runs on this same week.

The 10 most important restaurant metrics, at a glance

The most important restaurant metrics are the ten below, because together they cover what you sell, what the food costs, what the people cost and what is left over. The best restaurant metrics for a successful restaurant are the ones you can act on this week, and every one of these qualifies. Longer lists of 10 metrics for restaurants often spend slots on debt ratios and inventory turnover, which nobody changes during a shift.

Read all ten off the profit and loss and every one of them arrives after the month has closed. Five of the ten are fixed the day the schedule is built, so five of them can be held before a single shift is worked.

Five KPIs your sales and your kitchen set

Your register and your stock count set these five, and each one takes a single division. Your point of sale gives you the first three. Your stock count gives you the fourth.

Net sales

Net sales is what you actually took in after comps, discounts and voids, before sales tax and tips. It is the top half of almost every other number on this page, so get it right first.

Gross sales − comps, discounts and voids = $100,000

For example, the restaurant rang up $104,000 and gave away $4,000 in comps, discounts and voids. Those come straight off the top, so $104,000 minus $4,000 leaves a $100,000 week.

Average check

Average check is how much the typical table spends. Divide net sales by the number of checks you closed.

Net sales ÷ checks closed = $40

For example, $100,000 divided by 2,500 checks is a $40 average check. Watch it next to your guest count, because a busier week with a smaller check can still bring in less.

Table turnover rate

Table turnover rate is how many times each table is used during one service. Divide the parties you seated by the number of tables.

Parties seated ÷ tables = 2.5 turns

For example, 75 parties across 30 tables at Saturday dinner is 2.5 turns. A higher number means the same room earned more that night. Some operators take it one step further and divide sales by seats times hours open, which is called revenue per available seat hour.

How to calculate food cost percentage

Food cost percentage is the share of your sales spent on the food and drink you used. The accounting firm Warren Averett writes it as total food cost divided by total food sales, times 100. To find what you used, add your starting stock to what you bought, then take away your ending stock.

(Starting stock + purchases − ending stock) ÷ sales × 100 = 30%

For example, $10,000 of starting stock plus $31,000 of purchases, minus $11,000 left on the shelves, is $30,000 used. Against $100,000 of sales, that is 30 percent. The National Restaurant Association found food and non-alcohol drink costs ran to a median of 32.0 percent of sales for full-service restaurants in 2024, and 32.4 percent for limited service. At 30 percent, the example kitchen is running two points under the full-service median. Count stock only once a month, though, and one bad week of waste hides inside three good ones.

Net profit margin

Net profit margin is the share of sales you keep after every bill is paid. Divide net profit by sales and multiply by 100.

Net profit ÷ sales × 100 = 2.8%

The Association puts the full-service median at 2.8 percent of sales before tax, and limited service at 4.0 percent, in its 2025 operations data. For example, at that full-service median the restaurant keeps $2,800 of its $100,000 week, which is $100,000 times 0.028. At a 2.8 percent margin, you have to sell about $36 to keep $1.

Five KPIs your schedule sets

Labor cost percentage, prime cost, sales per labor hour, average labor cost per hour and overtime hours are all decided the moment you finish next week’s schedule. Every one of them is built out of the hours and the pay on it.

Labor cost percentage: the number on the P&L and the number on the schedule

Labor cost percentage is total labor divided by net sales, times 100. The full walk-through is in how to calculate labor cost percentage. In a restaurant there are two versions of it: the full one on your profit and loss, and the scheduled-wages one your schedule actually moves.

For example, here is the same week both ways.

The National Restaurant Association put full-service labor, including benefits, at a median of 36.5 percent of sales in 2024. It also split operators by whether they made money. Full-service operators that lost money spent a median of 42.9 percent of sales on labor, and the ones that made money spent 34.2 percent. The example restaurant sits at 35. Your schedule controls the $26,000, which is the biggest line in that ledger and the only one you can change this week. For the ways to bring it down, read how to reduce restaurant labor cost.

What is prime cost in a restaurant?

Prime cost in a restaurant is the cost of the food and drink you sold plus your total labor. Prime cost percentage is that sum divided by net sales, times 100. It puts your two biggest bills side by side.

For example, the example week’s food and drink plus its labor adds up to a prime cost of $65,000, which is 65 percent of its sales. The Association found limited-service restaurants spent a median of 65 cents of every sales dollar on prime cost in 2024. For full service it publishes the two halves on their own, 32.0 percent for food and 36.5 percent for labor, and two medians cannot simply be added together.

Every shift added after the schedule is built lands straight on prime cost, and the profit and loss does not show it until the month closes. Hold the labor half to a weekly limit that refuses the shift, and the only half of prime cost still moving is the food.

How to calculate sales per labor hour

To calculate sales per labor hour, divide net sales by the labor hours worked on the same days. It leaves pay rates out completely, so it answers one question: did the hours on the schedule match what the room sold?

For example:

1. 1

   Net sales for the week

   $100,000

2. 2

   Hours on the schedule

   1,250

3. 3

   $100,000 ÷ 1,250 hours

   $80 of sales for every hour

Staff to how busy last Saturday felt instead of what it sold, and every extra 10 hours needs $800 more in sales just to stay at $80.

Your sales figure comes from your register. The hours come from Workforce Insights in XShift AI. It counts only published shifts and takes unpaid breaks out, so the hours under that division are the hours you actually put on the schedule. Pick one location or one role, and the hours are counted again for just that slice.

Overtime hours: the restaurant KPI that costs the most to miss

Restaurant overtime hours are the hours an employee works past 40 in a workweek. The Department of Labor states the rule this way: “employees covered by the Act must receive overtime pay for hours worked in excess of 40 in a workweek at a rate not less than time and one-half their regular rates of pay.” Each of those hours costs half again what a regular hour costs.

Overtime builds one shift at a time inside the week. For example, a cook works four 10-hour days from Monday to Thursday and reaches 40. Every one of Friday’s 10 hours lands past the line.

An outdated scheduling tool does not add the hours up as you build, so nobody sees the cook cross 40 until Friday’s shift is already on the board. Ten cooks doing that is 100 overtime hours, and 100 hours times $10 extra is $1,000 of premium in a single week. That is $1,000 times 52 weeks, or $52,000 a year at one location, and $416,000 across eight.

Now put it next to the profit. Without the premium, the week would have kept $3,800. The premium takes $1,000 of it, and you keep $2,800.

XShift AI catches overtime before payroll in two ways. The first happens before the shift exists. Tell Autopilot that nobody goes past 40, and a shift that would take somebody to 41 is refused before it saves, even when their first 40 hours were at another location. It adds the new shift’s hours to everything that person already has that Sunday to Saturday week, and turns the shift away if the total passes 40.

The same goes for how many days in a row you allow. At eight hours a shift, five days is already 40 hours, so a sixth day in the same week is paid entirely at time and a half. Write a rule with the most days in a row you allow, say five, and a shift that would make it a sixth straight day is refused. It counts every day the person works at any of your locations. Two shifts on the same date count as one day. If the new day would join two runs together, the whole joined run is what gets counted.

The second is the overtime agent, for a week that is already running. A manager runs the scan from the Autopilot screen, and Workforce Insights has to be switched on. The scanner reads the current Sunday to Saturday week, flags everyone past 40, and prices those hours at time and a half. For each person it picks the smallest shift that would bring them back to 40. Then it looks for somebody to take that shift who passes all six checks below, starting with whoever has the fewest hours.

The swap goes in front of a manager to approve or dismiss, and nothing moves until somebody says yes, so the overtime is caught before payroll runs. Approving it moves the shift and tells both people. If the premium is the number you want gone first, read how to prevent restaurant overtime.

Average labor cost per hour

Average labor cost per hour is your labor cost divided by your labor hours. It rises whenever more of your hours are paid at time and a half, or go to your highest-paid people. That is why it catches creep an hours report misses: the hours can stay flat while the cost climbs.

For example, the 1,250 hours times $20 is $25,000 at straight time, or $20.00 an hour. Add the overtime from the section above and the same hours cost $26,000, which is $26,000 divided by 1,250, or $20.80.

The same 1,250 hours move from $20.00 to $20.80 an hour, and not one person got a raise. Workforce Insights shows that average for the week, with every hour past 40 already priced at time and a half, salaried people at their salary divided by 2,080 hours, and overtime-exempt people at straight time.

How XShift AI holds your restaurant labor KPIs to a budget

XShift AI holds your restaurant labor KPIs by pricing every shift before it is allowed to save, and refusing the one that would take the week past the limit you wrote in Autopilot.

Start with the number. The straight-time wage bill from the section above is a sensible weekly limit for the example restaurant, because it is the same week with the overtime taken out. A budget kept in a spreadsheet is only checked after the shifts are already saved. In Autopilot you write it as one sentence: “Keep labor under $25,000 a week at the downtown location.”

From then on, every time a shift is about to save, it adds up what everyone already scheduled that week at that location is paid. It prices the new shift for that person, with time and a half on any part past 40, counting their hours at every location. It adds the two and compares the total with your limit. If a manager is saving the shift by hand, it refuses it and shows the numbers, or it warns and lets a manager with permission save anyway, depending on how you wrote the rule. If the schedule builder or call-off coverage is choosing, it passes over that person and tries the next one who fits.

For example, the week already holds $24,800 and a Friday close still needs a cook.

The first cook’s close is refused, the second cook takes it, and the week stays under $25,000.

Five locations means five sentences and five limits, and each one is priced on its own. You can also give a single day its own limit, so a slow Tuesday cannot spend what Saturday needs. Before you save a rule, it lists anyone at that location who has no pay entered yet.

The weekly KPI sheet for a restaurant manager

A restaurant manager should check table turnover every shift, net sales and average check every day, the four labor KPIs every week, and food cost, prime cost and net profit margin every month. Each number sits where you can still act on it. Turnover can change tonight. Labor can change next week. Food cost needs a stock count, so it waits for the month.

Look at all ten every day and you will end up acting on none of them. Two of the four weekly numbers are already on one screen in Workforce Insights, and the other two need only your sales figure.

Sample KPIs for restaurant staff

The best KPIs for restaurant staff are the ones a person can move during their own shift, so they change by role.

A KPI a person cannot move on their own shift is not a target, it is just a number to be blamed with. Workforce Insights marks anyone who went past 40 in any week of the period, so the overtime KPI names the person and not only the total.

Frequently asked questions