Restaurant prime cost: the formula, and what it tells you

Prime cost is cost of goods sold plus total labour, measured against sales. It is the number that covers both halves of what you actually control week to week, which is why operators who watch only food cost keep getting surprised by the P&L. Put your own week in the calculator, then read what it means.

Last updated October 15, 2026 · by the Final Menu team

The prime cost formula

Prime cost adds the two largest costs a restaurant can do something about in a given week, then reads the total against what the restaurant sold.

prime cost = cost of goods sold + total labour

prime cost % = prime cost ÷ total sales × 100

Cost of goods sold is food plus beverage for the period. Total labour is everyone on the payroll, kitchen and front of house, including you, plus the payroll taxes, benefits and insurance that ride on top of the wage.

A week, worked through

Here is the week the calculator loads with. Sales for the period were $42,000.

That leaves $14,200, or 33.8% of sales, to cover rent, utilities, insurance, equipment, repairs, marketing, card fees, accounting and whatever is supposed to be left over for you. Read the number that way and it stops being an abstraction: prime cost is the share of every dollar that is spoken for before the landlord is paid.

Why not just watch food cost?

Because the two halves move against each other, and watching one hides the trade. Prep more in house and food cost falls while labour rises. Buy portioned, pre-trimmed product and labour falls while food cost rises. Either move can look like a win on its own line while the combined number sits still or gets worse.

This is also why food cost percentage is a dish-level and period-level tool rather than a verdict on the business. It tells you what the plate costs. Prime cost tells you whether the operation can pay its rent.

What counts as labour

The common error is counting wages and stopping. Fully loaded labour includes:

• Hourly wages and salaries, kitchen and front of house.
• Your own salary, whether or not you currently pay yourself one.
• Employer payroll taxes.
• Health insurance and any other benefits you fund.
• Workers compensation insurance.
• Overtime, which is where a scheduling problem shows up as a cost problem.

Leave the burden out and prime cost reads several points better than the business is really running, which is the most comfortable kind of wrong number to publish to yourself.

Run it weekly, not monthly

A monthly prime cost arrives after the month it describes is over. Nothing about that month can be changed by then. A weekly figure lands while the next week is still being ordered and scheduled, which is the only point at which either half of the number is still movable.

Two rules keep weekly numbers honest. Use the same period boundaries for costs and sales, or you are dividing one week by another. And count invoices by the week the product arrived rather than the week you paid them, or a slow payment run will make one week look brilliant and the next look broken.

Moving the number

The cost-of-goods half

This half drifts without anyone deciding it should. Suppliers reprice, recipes stay the same, and menu prices set six months ago quietly stop clearing their target. Recosting recipes against current invoice prices is the work, and it is the half this site is about. The four menu pricing methods all take plate cost as their input, so a stale plate cost makes every price derived from it stale too. Most products that automate this differ mainly in where they put invoice capture, which the recipe costing comparison sets out tier by tier.

Over-ordering shows up here too, as product bought and not sold. That is an inventory problem rather than a pricing one, and the purchasing-led tools are built for it.

The labour half

This half you schedule, which makes it feel more controllable than it is. Most of the movement here is overtime, overlapping shifts at changeover, and holding a full line through a shoulder hour that does not earn it. Sales per labour hour, read by daypart, finds more of it than cutting a shift at random.

What counts as a good prime cost

Start with the number you have probably already read: prime cost should run between 55% and 65% of sales. No association, government body or published study states that range. The earliest source that can be attributed for it is a restaurant consultant writing in a trade magazine in 2016, who gave 65% as a full-service average and 55% as his own target for restaurants above $850,000 in sales.[5] It has been repeated between blogs ever since without anyone naming a study.

Here is what is actually published. The National Restaurant Association surveys operators and reports medians. For 2024, full-service respondents reported food and non-alcohol beverage at a median 32.0% of sales[2] and labour at a median 36.5% of sales.[3] Add those and the typical full-service prime cost lands near 68.5%, above the band that circulates as a target. Limited-service respondents reported prime cost at a median of 65 cents in the sales dollar.[1]

The more useful cut of the same survey compares operators by outcome rather than by average. Among full-service operators who turned a pre-tax profit in 2024, labour ran a median 34.2% of sales,[4] against 36.5% across all respondents. The gap between those two numbers is a better target than any round figure, because it describes what the profitable half actually did rather than what a rule of thumb says they should have.

Use your own history instead

Your format, your rent, your market and your menu all move the number that works for you. A steakhouse and a cafe have no business sharing a target. The number worth watching is your own prime cost over the last thirteen weeks: the direction it is moving, and what moved it. A restaurant that cut prime cost from 71% to 67% has learned something useful. A restaurant sitting at 64% and drifting up has learned something more urgent, and neither fact is visible in a benchmark.

Why the number drifts when you do nothing

Menu prices rose 3.5% over the year to September 2026.[6] Wholesale food prices did something different: the producer price index for final demand foods read 283.977 in September 2026 against 286.618 a year earlier, which is a fall of roughly 0.9%.[7] Those two facts together are the opposite of the usual story about costs climbing relentlessly.

They are also averages across an entire economy, which is why they cannot price your menu. Beef and eggs can run hard in a year when the aggregate index falls. The index describes a national basket; your invoices describe what you actually bought. Only one of those two sets your plate cost, and it is the one sitting in a pile by the office door.

What prime cost does not tell you

It says nothing about volume. A restaurant can hold a healthy prime cost percentage and still lose money by not selling enough to cover fixed costs, because every percentage on this page is a ratio and rent is a number. Read prime cost next to total sales and your fixed monthly costs, never on its own.

It also says nothing about menu mix. Two restaurants with identical prime cost can have completely different problems: one with a few badly priced dishes selling in volume, another with sound pricing and a staffing issue. The combined figure tells you something is wrong without telling you where.