What a benchmark is for, and what it is not
A benchmark is not a target. Nobody has ever improved a restaurant by hitting 30% food cost on purpose. It is a sanity check — a way of asking whether the number in front of you is the kind of number a restaurant like yours normally produces, or whether something has gone wrong that nobody has noticed yet.
That distinction matters because the bands are wide. Seven points of food cost separates the top and bottom of most of them. On €600,000 of annual revenue, seven points is €42,000 — the difference between a restaurant that pays its owner and one that does not. A band cannot tell you where inside it you belong. Only your menu, your suppliers and your prep can.
So the useful question is never “am I inside the band?”. It is “if I am outside it, do I know exactly why?”. An operator running 38% food cost who can tell you it is a deliberately generous tasting menu at a €140 check is in control. An operator running 31% who cannot say why it was 28% in March is not, even though the second number looks better on paper.
Where these ranges come from
They come from the same place every operator's rules of thumb come from: hospitality teaching, trade press, supplier conversations and decades of people repeating figures that turned out to be roughly right. They are folklore. Useful folklore — but folklore, and we would rather call it that.
We are saying so plainly because the alternative is what most benchmark pages do: attach a year, a sample size and an authoritative-sounding source to a number nobody actually measured. We cannot verify an industry statistic, we did not run a survey of European restaurants, and we will not write a citation that points at nothing. If you need audited figures for a bank or an investor, these are not them.
What they are genuinely good for is the thing operators already use them for: a fast read on whether a number is plausible, and a way of arguing with your own P&L that has some shape to it.
Food cost: why the band is so wide
Kitchens sit in the high twenties to the mid thirties. Bars sit far lower, because pouring a spirit costs a fraction of plating a dish, and because a bar's cost of goods is concentrated in a handful of lines it buys every week rather than spread across a hundred perishable ones.
Inside a single band, the thing that moves you is menu mix, not skill. A restaurant selling a lot of beef and fish will sit at the top of its range while doing everything right; one selling pasta and vegetables will sit at the bottom while doing very little right. This is why comparing your food cost with another restaurant's is close to meaningless, and comparing it with your own last quarter is close to essential.
The gap between your theoretical food cost — what your recipes say the food you sold should have cost — and your actual is where the operational money hides. A band will never show you that gap. Only counting will.
- Menu mix — which dishes actually sell, not the ones you are proud of
- Purchase prices, tracked delivery by delivery rather than at the month end
- Portioning, which drifts every time a new pair of hands joins the line
- Waste, spoilage and trim, none of which appear on an invoice
- Staff meals and comps, which are real food cost wearing a different hat
Labour: the band you cannot read off a menu
Labour is quoted by service model rather than by cuisine, because what drives it is how many people the room needs to run, not what leaves the pass. A full-service dining room carries a section, a pass, a bar and a floor whether or not the tables fill. Counter service does not.
That is the whole reason full service sits several points above limited service in the conventional bands. It is structural. You do not get a served room down to counter-service labour by managing harder — you get there by changing the service model, which is a different decision with different consequences.
The bar band is the widest of the three, and honestly so: bar labour is almost entirely a scheduling question. Two people on a slow Tuesday and five on a Saturday is one venue producing a ten-point swing, which is why bar labour has to be read across a month rather than across a night.
One thing the bands almost never include, and should: your own pay. If you work the floor five days a week, your labour is a real cost whether or not it leaves the account. A labour percentage that quietly excludes the owner is a number designed to feel better than the restaurant is.
Prime cost is the number that decides the year
Prime cost is food plus total labour. In an independent restaurant it is the large majority of controllable spend, and it decides the year. If prime cost is right, most other problems are survivable. If it is wrong, nothing you fix downstream will be big enough to matter.
The prime cost chart above is the two bands added end to end, and that is arithmetic rather than a fourth benchmark. The bottom of your food band and the bottom of your labour band rarely occur in the same month — cheaper raw ingredients usually mean more prep, and more prep means more hours. Read the sum as an outer boundary, not as a range you should expect to sit inside comfortably.
The practical version: make prime cost the single number you look at first each month, and decompose it only when it moves. Two halves moving in opposite directions inside a stable prime cost is an ordinary restaurant. A prime cost climbing a point a month is a restaurant with a problem it has not identified yet.
Three to nine percent is a warning, not a target
The general net margin band is low enough to surprise everybody outside the industry and nobody inside it. Three percent means a €600,000 restaurant clears €18,000 in a year — less than the owner would earn working in somebody else's kitchen, for a fraction of the risk and none of the lease.
Bars are quoted higher, and the reason is the food cost band above: a structurally lower cost of goods on the thing you sell most drops straight through to the bottom line. That is an advantage of the format, not evidence that bar operators run tighter ships.
The band that matters more than any of these is the one underneath it — whether the number is positive at all, and whether it stays positive after the owner is paid properly. A restaurant reporting a 7% margin while paying its owner nothing is reporting a 7% margin on a business that does not work.
European context: what we can and cannot tell you
This is a European company — registered in the Czech Republic, working in Czech, English and Italian — so the obvious thing to publish here would be Czech, Slovak and Italian benchmark figures. We are not going to, because we do not have them, and inventing three plausible-looking national ranges would undo everything the rest of this page is trying to do.
What can be said without inventing anything is structural, and it matters more than a national average would. The first thing is VAT. A band means nothing unless you know whether the revenue underneath it is gross or net, and European rates differ enough by country and by category — eat-in against takeaway, food against alcohol — that one restaurant can produce two noticeably different food cost percentages depending on which revenue figure it divides by. Choose net of VAT, and use it every month.
The second is the cost of employing somebody. Across much of Europe, employer social contributions add a substantial amount on top of gross pay, and the proportion is set nationally rather than by you. A labour percentage built from gross wages alone will sit comfortably inside the band and still be wrong. Use the full employer cost, contributions and holiday pay included, or your labour figure is measuring a different thing from the one the band describes.
The third is seasonality, which catches out mountain towns and coastal ones alike. An annual band spread evenly across twelve months describes almost no European restaurant. Fixed costs run all year; revenue does not. The month-by-month picture is the one to act on, and the annual figure is the one to sign a lease against.
None of that is a number, and we would rather hand you three questions you can answer about your own accounts than three country averages you could not check.
How to use a band without lying to yourself
The first rule is definitional consistency. However you decide to count — VAT in or out, owner's pay in or out, delivery commission as lost revenue or as a cost — count it the same way every month. A comparison between two differently defined numbers is worse than no comparison, because it produces confident conclusions out of noise.
The second is that direction beats level. A restaurant at 34% food cost drifting up half a point a month is in more trouble than one sitting flat at 36%. The band tells you where you are. Only a series tells you where you are going, and where you are going is the part you can still change.
The third is to move one thing at a time. Re-cost the menu, renegotiate two suppliers and rebuild the rota in the same week and you will get a result you cannot attribute — which means you cannot repeat it either.
- Fix your definitions once, and write them down
- Measure weekly; a month is long enough to hide a bad week inside it
- Compare against your own last quarter before you compare against a band
- Change one variable, then wait a full period before reading it
- Count the owner's pay as a cost, always
The one number here that is ours
We built this software for our own restaurant first. In a town of 30,000 people we grew a million-dollar operation, and for years we ran it the way most independents do — on instinct, with the real numbers arriving six weeks after they could have been useful. The first year we tracked properly, we saved €4,000. Not from a clever new idea. From finally seeing where the money was leaving.
That figure is ours, it is measured, and it is the only one on this site that is. Everything above is the industry's conventional wisdom, offered as a shape to check yourself against. Your own numbers, counted the same way every week, will tell you more in a month than any benchmark will tell you in a year.
Get the benchmark report
A longer written version of this page: the bands, the reasoning behind each one, and the questions worth asking when your own numbers land outside them.
It is not finished, so there is nothing to download and no PDF sitting behind this box. Leave your address and it goes out when it is written. If that is not worth an email address, close the tab with our blessing — the whole page is above and none of it is gated.
Questions operators actually ask
What is a good food cost percentage for a restaurant?
Is prime cost more important than food cost?
Why is the restaurant profit margin band so low?
Do these benchmarks apply to European restaurants?
What should I do if my restaurant sits outside every band?
See your own numbers instead
A band is somebody else's restaurant. The dashboard shows you yours — food cost, labour, break-even and the direction all three are travelling — in a few minutes a week. Three days free, no card.
No signup