Prime cost and why it is the only number that matters some months

Everything else on a restaurant P&L is either fixed, seasonal, or too small to chase. You do not renegotiate rent in week two because food came in heavy. You do change what gets ordered, prepped, portioned, and scheduled.

That is the whole argument. Prime cost is the operator's number because it is the only block that responds to operating.

Why combine food and labor into one number?

Because they trade against each other, and watching them separately hides the trade.

Buy portioned proteins and your food cost rises while your prep labor falls. Butcher in house and the reverse happens. Bring in a prepared sauce and one line goes up, another goes down. Looked at individually, each of those decisions can be defended or attacked. Looked at together, you can see whether the trade was worth making.

The same applies to convenience across the board. Bagged salad, pre portioned desserts, a bakery contract instead of a baker. Every one of those is a deliberate purchase of labor with food dollars.

If you only manage food cost, you will make labor decisions by accident. If you only manage labor, you will make food decisions by accident. Prime cost forces you to see both sides of the same choice.

There is a second reason. Food cost is measured monthly if you count and estimated monthly if you do not. Labor is measured every pay period with real precision. Combining them gives you one number you can watch weekly with reasonable confidence, and weekly is the cadence where you can still do something.

How much of the dollar is prime cost supposed to take?

Roughly two thirds, and the published numbers are worth knowing before you judge your own.

The National Restaurant Association describes the pre-pandemic shape of a typical independent restaurant as food taking about 33 cents of every sales dollar and labor taking about 33 cents. That is prime cost at roughly 66 cents. Everything else, meaning utilities, occupancy, supplies, general and administrative, repairs and card processing, took about 29% of sales, which left a pre-tax margin near 5%.

Two thirds of your revenue is spent by people making decisions during a shift. That is why the weekly number matters and the annual one does not.

The labor half has a harder benchmark behind it. The Association's Restaurant Operations Data Abstract, 2025 edition, collected financial and operating data from more than 900 operators nationwide. Salaries and wages including benefits came in at a median of 36.5% of sales for full service restaurants in 2024. Split by result, operators reporting a pre-tax profit ran 34.2% and operators reporting a loss ran 42.9%. Limited service shows the same shape lower down, at 31.7% overall, 30.0% for profitable operators and 34.1% for losing ones.

The distance between a profitable full service restaurant and a losing one is 8.7 points of labor, which is most of a prime cost problem sitting on one side of it.

Two cautions before you use any of that. The Association states plainly that the Abstract is "not intended to represent standards or goals for individual restaurants." And the median is salaries and wages including benefits. If you compare a wages only figure against 36.5%, your restaurant looks better than it is.

How do you actually use it during the month?

Not by waiting for the statement.

Run prime cost weekly against that week's sales. Purchases as a stand in for cost of goods, actual labor from the timekeeping system, both divided by net sales. It is not accurate enough to publish. It is accurate enough to act on, and acting in week two is worth far more than being right in week six.

Two habits make it useful:

Track the trend, not the level. Your prime cost depends on your concept, your rent, your market, and your menu mix. Comparing yours to somebody else's is mostly noise. Comparing this week to your own last eight weeks is signal.

Investigate direction changes, not single weeks. One bad week is a holiday, a party that cancelled, or an invoice that landed on the wrong side of a Sunday. Three weeks moving the same direction is a condition. This is the same discipline as knowing when a metric that moved is telling you something and when it is noise, and it saves you from reorganizing the kitchen over a fluke.

What does one point of prime cost actually cost?

Here is the arithmetic on a restaurant doing $1.2M a year, which is about $23,000 in an average week. Substitute your own weekly sales and the shape holds.

Take one week. Purchases came in at $7,360, which is 32% of that week's sales. Labor with taxes and benefits was $8,395, which is 36.5%. Prime cost is $15,755, or 68.5 cents of every dollar.

Now compare it to your own eight week average rather than to anyone's benchmark. Say that average was 66 cents. You are running 2.5 points heavy.

Two and a half points of $23,000 is $575 in that one week. Held for a year, it is $29,900.

That is the number that makes the conversation concrete. Not "food seems high." A specific annual figure, attached to a specific week, produced from three inputs you already have.

Then split it. If labor is the half that moved, from 36.5% to 38%, that is 1.5 points, or $345 of the $575. If purchases moved the rest, you have a food question. The split tells you which department is having the conversation, and the split is available weekly while the P&L is not.

Run this for eight consecutive weeks before you change anything. One week is a story. Eight weeks is a shape.

Which side do you attack first when it is high?

Labor, almost always, and not because labor is the bigger problem. Because it is the faster one.

A labor change lands in days. You post a different schedule and the number moves next week. Most scheduling waste is not people being lazy, it is scheduling decisions that quietly cost a shift its margin and never get reviewed, so the fix is often available immediately.

Food takes longer. Recipes get retested, portions get re specified, staff get retrained, and vendor pricing takes a cycle to move. It is worth doing and it is not this week's lever.

There is an order that works:

1. Fix the schedule where sales per labor hour is worst 2. Count inventory so the food number stops being a guess 3. Take the count to your vendors with actual usage in hand 4. Then change recipes, portions, and prices

Step three is where most independents leave money. You are not going to out negotiate a national chain on volume, but there are real ways an independent operator gets better vendor terms that have nothing to do with size. Almost all of them require knowing your usage, which is step two.

And step two is where most operators stall, because a full inventory program feels like a second job. It does not have to be. Inventory systems for small restaurants that are actually worth the effort usually track a short list of high value items rather than every jar in the building.

Where does prime cost stop being the right number?

Three places, and knowing them keeps you from managing the wrong thing hard.

When part of the labor line is set by law rather than by you. California's AB 1228 established a $20 an hour minimum for fast food workers at chains with more than 60 locations nationwide, effective 1 April 2024. If a rule like that reaches your business, rate is not a lever and your only remaining moves are hours, throughput, and price. Be careful about the conclusions you draw from that law, because the evidence on its employment effect is genuinely split. Research summarised by the Cato Institute and circulated as NBER working papers found California fast food employment down 2.7% against the rest of the country over a year, while other work using a synthetic difference in differences approach produced an employment elasticity of −0.04, statistically indistinguishable from zero. Both camps report the same finding on one point: the separation rate fell.

When your concept deliberately trades one side for the other. A commissary model, a heavy purchase of prepped product, or a bar led operation with high margin liquid sales will sit somewhere unusual on both halves and be perfectly healthy. The combined number is still useful. The comparison to a median is not.

When the cost structure is fine and the sales are not. The Association's July 2026 analysis estimates total expenses for an average restaurant rose 36% between 2019 and 2026, and reports that 42% of operators said their restaurant was not profitable in 2025. In an environment like that, a restaurant can hold prime cost exactly where it belongs and still lose money, because 5% of margin does not absorb a 36% cost increase without more sales behind it. Squeezing a line that is already at benchmark is how good operators burn a quarter of management attention for nothing.

What to do this week

Calculate prime cost for the last eight weeks. Purchases plus labor divided by net sales, one row per week. A spreadsheet is enough. Do not build a dashboard yet.

Plot it. You are looking for the shape, not the value. Flat, drifting up, or bouncing.

Add two more columns while you are in there: food as a share of sales and labor as a share of sales. When the combined number moves, you want to see instantly which half moved it.

If it is drifting up, look at labor first. If it is bouncing wildly, your inventory is not being counted and the food number is fiction. If it is flat and you are still not profitable, prime cost is not your problem and you should be looking at rent, volume, or your menu prices.

Set one number as the weekly review point and put it in front of your management team every Monday. Not a report. One number, out loud, with one sentence about which direction it moved.

Questions about your prime cost?

Email me at eric@seod.com with eight weeks of purchases, labor, and net sales. Three columns, no formatting required. I will send back the trend, which side is driving it, and the one change I would make first.

Sixteen years running multi unit restaurants and more than $54M in annual P&L means I have watched this number weekly for a long time, including in the months it was the only thing standing between a location and a bad quarter.

Otherwise there is more on running the operation here.