How to calculate and reduce food cost in your restaurant

Why food cost decides the year

Food cost is not an accounting category. It is the running record of every decision made between the delivery door and the plate — what you ordered, what you paid for it, what got prepped, what got binned, what was given away, and what quietly left the building. Individually those decisions are too small to argue about. Added together, they are usually the difference between a year that pays the owner and a year that does not.

Two restaurants can look identical from the dining room and sit four points apart on food cost. Same covers, same noise, same reviews, same Saturday. On six hundred thousand euros of food revenue, four points is twenty-four thousand euros. That is not a rounding error, it is a salary, and it is invisible to everyone standing in the room.

This is why food cost is the first number we tell an operator to get under control. Not because it is the most sophisticated measure — it is not — but because it moves fastest, responds fastest, and sits almost entirely inside your own control. Rent is a negotiation once every five years. Food cost is a decision you make forty times a day.

The formula is the easy part

Food cost percentage is opening inventory plus purchases, minus closing inventory, divided by food revenue for the same period. You almost certainly knew that. Most operators do, and most operators still cannot tell you what the number was last week.

That is the real problem, and it has nothing to do with arithmetic. The formula was never the obstacle. Having the number, soon enough to do something with it, is the obstacle.

A monthly figure that arrives with the accounts is describing a month that is already closed. The produce has been eaten, the waste is long gone, and the shifts you would have changed happened five weeks ago. You are reading a post-mortem and calling it a report. Weekly is not a more rigorous version of monthly — it is a different instrument, because a week is short enough that you still remember what went wrong in it.

• A monthly close tells you what happened. A weekly number lets you change what is happening.
• Same weekday, same method, same items — comparability matters more than precision.
• An imperfect count run every Monday beats a perfect count run twice a year.
• Trend is the signal. A single week on its own tells you almost nothing.

Actual, theoretical, and the gap between them

One number on its own will not tell you what to fix. Two will.

Your theoretical food cost is what your menu would produce if every plate went out exactly as it was costed, weighted by what actually sells. Your actual food cost is what really happened. The distance between the two is your operational leak, and it is denominated in euros.

That distance is worth more than either number alone, because it separates two problems that look identical on a profit and loss statement and have opposite fixes. A wide gap is an execution problem: portions, waste, spoilage, things not being logged. Repricing the menu will not touch it. A narrow gap sitting at a high number is a pricing and menu-mix problem. Tightening portions will not touch that one either.

Most operators guess wrong the first time, and spend a quarter carefully fixing the thing that was already fine. Having both numbers in front of you removes the guess. That is most of the value of measuring at all.

Where the reductions actually come from

When food cost climbs, the first instinct is to call suppliers. Prices do matter. But in our own restaurant they were rarely the largest line, and the bigger losses were internal, unglamorous, and entirely ours to fix.

• Portions that drifted. Ten grams heavy on a protein looks like nothing on the pass and costs thousands over a year.
• Prep for a day that never came — the Thursday prep for a Friday that stayed quiet.
• Spoilage from ordering for the week you hoped for rather than the week you got.
• Staff meals, comps and remakes that were never costed and never logged.
• Deliveries invoiced in full and delivered in part, signed for without being weighed.
• Theft — usually smaller than owners fear, and almost always larger than they check.

None of that arrives on an invoice

Every item on that list has one thing in common: no supplier will ever bill you for it. There is no line on any document in your office called “portion drift” or “prepped for a Friday that never came”. These losses show up in exactly one place — the gap between what you bought and what you sold.

Which is the thing a weekly food cost number measures, and the reason we keep insisting on it. You cannot reduce a loss you have never seen. Operators who start tracking weekly rarely discover one dramatic problem; they discover six small ones, none of which would have survived being looked at, and all of which had been running for years.

Why most food cost projects die in week three

Operators rarely fail at this for lack of a method. They fail because the method they were handed takes an hour a day, and restaurants do not have an hour a day.

The weeks you most need the number are the busy ones. Those are precisely the weeks nobody opens the spreadsheet. So a system that only survives quiet weeks is a system that reports on the wrong weeks — and worse, it leaves gaps in the trend exactly where the trend was about to tell you something.

Whatever you build has to survive a real service week: short, fixed, and boring enough to do on a Monday morning without deciding to do it. That constraint should drive every choice you make about tools, not the other way round. Sophistication that costs you consistency is a bad trade, every time.

What it was worth to us

We built our own version of this for our own restaurant, in a town of thirty thousand people, after years of running on instinct and closing the year with no clear sense of which decisions had helped. The first year we tracked properly, we saved €4,000. Not from a new concept, not from a price rise, not from a marketing push — from finally seeing where the money had been leaving all along.

That figure is not a projection and it is not a case study we bought. It is our own accounts, and it is the reason any of this exists.

Start with one number, this week

If you have never tracked food cost weekly, do not start by building a system. Get one number this week, with whatever you already have. Then get it again next week. The second number is where the value begins, because a direction of travel tells you something a level never can.

If you would rather not build and maintain the spreadsheet at all, that is the job master.restaurant does: you enter revenue, purchases, inventory and payroll, and it returns food cost, payroll cost and break-even without a formula anyone has to look after. The three-day demo runs on your own figures and does not ask for a card, and if it earns its place, lifetime access is €499, paid once.

Either way the discipline is the same, and it is the part that matters. See the number early enough to act on it, while there is still a month left to act in.