Prime cost: the one number that tells you whether the week worked

Prime cost is cost of goods sold plus total labor cost, measured against sales for the same period. Everything else on a restaurant P&L — rent, insurance, utilities, the loan — is largely fixed for the period you are in. Prime cost is the portion that responds to decisions made this week.

That is why it deserves the attention. Not because it is the biggest number, though it usually is, but because it is the biggest number you can still do something about.

The calculation

Prime cost = COGS + total labor

Prime cost percentage = prime cost ÷ total sales

Two words in that formula do the damage, and they are total and COGS.

Total labor means all of it. Hourly wages, salaried management, payroll taxes, workers' compensation, benefits, and any contract or agency labor. A prime cost built on hourly wages alone is not prime cost; it is a fragment, and it flatters. Management salary and employer taxes are the lines that move when you add a shift leader or when a state unemployment rate resets.

COGS means cost of goods actually used, not goods purchased. Those are the same number only if inventory did not move, which it did. Purchases in a week when you built stock overstate cost; purchases in a week when you ran stock down understate it. The correction is the standard one:

Opening inventory + purchases − closing inventory = cost of goods used

Which means prime cost is only as honest as your inventory count. An operation that counts monthly can produce a monthly prime cost. An operation that wants a weekly prime cost has to count weekly — at minimum on the categories that move: proteins, produce, liquor. On the beverage side that same weekly count is what makes a pour cost variance visible while there is still time to act on it.

Why a monthly figure is nearly useless

A prime cost calculated at month-end and delivered mid-way through the following month describes a period that finished weeks ago. Every schedule it might have changed has already been worked, and every order it might have adjusted has been delivered and consumed.

The value of the number is entirely in its timing. A weekly prime cost, produced within a few days of the week closing, still has decisions attached to it: next week's schedule is not yet locked, next week's order is not yet placed. The same figure a month later is history — accurate, and worth nothing operationally.

This is the single most common reason operators say they "watch prime cost" and yet nothing changes. They are watching a photograph of something that has already left.

Targets, and why yours may not match anyone else's

Operators conventionally aim to keep prime cost around the low-to-mid 60s as a percentage of sales for full service, and lower for limited service, where labor is thinner. Those are working conventions rather than laws, and they are only useful once you know what your own model can carry.

A concept with high rent needs a lower prime cost to survive than one with cheap rent, because the fixed base beneath it is heavier. A bar-led operation carries a different cost of goods profile than a food-led one. Comparing your prime cost to a published benchmark tells you less than comparing it to your own last four periods and to the same period last year, which is where a real trend lives.

The coding decisions that distort it

Prime cost is sensitive to where things are posted, and a chart of accounts assembled without this in mind will produce a number that moves for reasons nobody can explain.

• Comps and voids treated as discounts. A comped meal consumed food. If the cost stays in COGS while the sale disappears, the percentage moves without any operational change. Comps need their own treatment and their own visibility — they are a management decision worth measuring, not noise to be netted off.
• Employee meals. Real food cost, not a sale. Left inside COGS against a sales line that never existed, they inflate food cost every period by roughly the same amount, which is worse than being wrong — it is being wrong invisibly and consistently.
• Delivery commission. Marketplace commission is not cost of goods. Where the gross order value is booked as sales and the commission buried in COGS, food cost percentage is wrong in both directions at once.
• Salaried management left out of labor. The most common single omission, and the one that makes a prime cost look survivable when it is not.

None of these are exotic. They are the ordinary consequence of a chart of accounts built for tax filing rather than for operating, which is what most restaurants have.

Making it operational

A prime cost that changes behavior has three properties: it arrives weekly, it splits into COGS and labor rather than arriving as one figure, and it is reported per unit rather than for the group. A consolidated prime cost across five units is an average, and averages hide the unit that is failing behind the four that are not.

The reporting is the easy part. The count is what most operations have to build.

---

Related: Restaurant accounting services · Food cost percentage · Restaurant P&L statement · Chart of accounts

Common questions

What is prime cost in a restaurant?

Prime cost is cost of goods sold plus total labor cost, measured against sales for the same period. It captures the largest and most controllable part of a restaurant's cost base.

What is included in total labor for prime cost?

All of it: hourly wages, salaried management, payroll taxes, workers' compensation, benefits and any contract or agency labor. Leaving out salaried management and employer taxes is the most common error and it makes prime cost look better than it is.

Should prime cost use purchases or inventory usage?

Usage. Opening inventory plus purchases minus closing inventory gives cost of goods used. Using purchases alone makes prime cost swing with stock building and drawdown rather than with what was actually sold.

How often should prime cost be calculated?

Weekly. A monthly prime cost delivered mid-way through the following month describes shifts already worked and orders already placed. A weekly figure still has decisions attached to it.