The short answer
Food cost is what one plate costs you to make. Write down what a single portion uses, price each ingredient at what you actually paid, add it up. That is the whole method, and you can do the first dish in ten minutes.
What is worth your attention is not the formula. It is the two mistakes that make the answer wrong while it still looks right:
- A gram measured against a kilo. You buy mango by the kilo and cook with it by the gram. Get that conversion wrong and the cost is out by a factor of a thousand.
- An ingredient nobody has priced. If one component has no purchase rate, it contributes nothing — and a dish missing its most expensive item looks almost free. This one does not announce itself. It gets believed, and then it gets priced against.
The rest of this is how to get both right, with the arithmetic done in rupees.
A worked plate, in rupees
Take a Mango Lassi. You buy mango at ₹120 a kilo and milk at ₹4 a litre. One glass uses 100 g of mango and 200 ml of milk.
| Ingredient | One glass uses | You buy at | Cost |
|---|---|---|---|
| Mango | 100 g = 0.1 kg | ₹120 / kg | ₹12.00 |
| Milk | 200 ml = 0.2 L | ₹4 / litre | ₹0.80 |
| One glass costs you | ₹12.80 | ||
Sold at ₹30, that glass earns you ₹17.20 — a 57% margin. Sold at ₹20 it earns ₹7.20, a 36% margin. Same glass, same kitchen; the difference is entirely in what you decided to charge, and until you did this arithmetic you had no way to see it.
Notice the small grey text in the second column. That is the whole job: 100 g is 0.1 kg. Everything difficult about food costing is that one step, repeated.
The two ways it goes wrong
1. The unit error, which multiplies by a thousand
A cook thinks in grams and millilitres. A purchase bill is in kilos and litres. Somebody has to reconcile the two, and if it is the owner doing arithmetic in their head at data-entry time, sooner or later a recipe ends up with 0.1 in a field labelled grams, or 100 in a field the software reads as kilos.
That second one costs ₹12,000 a glass instead of ₹12. You would catch that. The dangerous version is subtler: a spice measured in grams against a packet you buy by the packet. There is no conversion between those two — a packet is not a mass — so the honest answer is that the line cannot be costed until you say how many grams are in a packet.
The same is true of pieces and bottles. Three bottles is not three packets, and no software should guess that it might be. If your costing tool silently converts across unit families, it is inventing numbers.
2. The unpriced ingredient, which is worse
Suppose the lassi above also takes 2 g of saffron, and saffron has never been entered on a purchase bill. Its rate is zero. Multiply it out and it contributes ₹0.00 — so the glass still reads ₹12.80, confidently, and nothing anywhere says a word.
At ₹400 a gram, that saffron is ₹800. The real cost of the glass is not ₹12.80. It is not even close.
A food cost quietly missing its most expensive ingredient is worse than having no food cost at all, because you will believe it and price against it. An unpriced ingredient must make the whole dish read as incomplete and say which ingredient it is waiting for — never as a number you can act on.
Half and Full are two dishes
If a dish is sold in sizes, one cost cannot describe it. A Half biryani and a Full biryani use different amounts of the same rice and the same meat, and they sell at different prices. Costing the dish once and comparing it to one of the two prices produces a number that is wrong for both.
It gets worse if the size is the thing that is actually sold. On most tills, the moment an item has sizes the cashier has to pick one — so the "base" price on the menu card is a price the shop never charges. A margin measured against it is a percentage of a number that does not exist.
Cost every size against the price it is rung up at, then look at the worst one. That is the plate quietly funding the others, and it is usually the one the kitchen is proudest of.
Your food cost moves, and it should
Onions were one price last month. Tomatoes will be another price next week. If your food cost lives in a spreadsheet you filled in once, it stopped being true the day after you filled it in — and it will be most out of date exactly when prices are moving fastest, which is precisely when you needed it.
The fix is not discipline. It is arithmetic that happens on its own: derive each dish's cost from what you last paid for its ingredients, so a rate that moves moves every dish that uses it, the same day, without anyone remembering to do anything. A weighted average across your recent purchases is the sane basis — one expensive emergency purchase of coriander should nudge your cost, not redefine it.
This is also why a cost stored on the dish is a trap. Freeze the number and you have frozen last month's onion price into this month's margin, where it will sit looking authoritative and being wrong.
What the food cost percentage is actually worth
Food cost percentage is cost divided by selling price. The lassi above is 42.7% at ₹30. You will find plenty of articles telling you what that number should be. Treat all of them carefully, for one reason: almost none of them says what was included.
Some count only raw ingredients. Some add packaging. Some add cooking gas. Some add staff meals and wastage. A percentage worked out one way is not comparable to a percentage worked out another, so a benchmark without a definition is not a benchmark — it is a number with a percent sign.
Two things are worth more than any published figure:
- Your own number, measured the same way every month. The level tells you little; the direction tells you everything. A cost creeping up three months running is a real signal, whatever the absolute figure is.
- The spread across your own menu. This is where the money is. Nearly every restaurant has a dish it is proud of that earns almost nothing, and a dull one carrying the whole month. You cannot see either until each plate is costed separately.
Keep gas, packaging and staff meals out of the plate figure so dishes stay comparable to each other, and account for them where they belong — packaging against parcels, the rest as running costs for the month.
Doing it without a spreadsheet
The spreadsheet version works and it is a fine place to start. It fails the same way every time: it is accurate the week you build it, nobody updates it when prices move, and six months later it is quietly wrong in a direction nobody has checked.
The alternative is to let it fall out of records you already keep. If your POS knows what you paid for onions — because you entered the purchase — and knows what one plate uses, it can do this arithmetic continuously and without being asked.
That is what Dinelax POS does. Open a menu item, add what one portion uses from your own inventory, and the cost, the selling price and the margin appear immediately. Units convert against the stock item's own unit, so grams against kilos is handled and cross-family pairings are refused rather than guessed. Anything with no purchase rate is named as unpriced instead of counted as free. A dish with sizes is costed per size, against the price actually charged, and the menu grid shows the worst margin of the lot — so the plate losing you money says so from across the room.
One thing it deliberately does not do: it does not deduct anything from your stock when you sell. What a plate costs and what you have left are two different questions, and the first is useful on its own. Billing, GST bills, kitchen tickets and reports are ₹149 a month; everything, including recipe costing, is ₹999. Fourteen days free, no card.
Where to start tomorrow
- Do your ten best sellers first. They are most of your revenue and the answers arrive fastest.
- Enter purchase rates before recipes. A recipe against unpriced stock cannot tell you anything.
- Weigh one portion, once. Guessing the quantity is the one place the whole exercise can go wrong that no software can catch for you.
- Look at the spread, not the average. The outlier dish is the finding.
Frequently asked questions
helpHow do you calculate food cost for a restaurant dish?
Write down what one portion actually uses, price each ingredient at what you paid for it, and add it up. A Mango Lassi using 100 g of mango bought at Rs 120 a kilo and 200 ml of milk bought at Rs 4 a litre costs Rs 12.00 plus Rs 0.80, so Rs 12.80 a glass. Sold at Rs 30 that is a 57% margin. The arithmetic is simple; what goes wrong is the units and the ingredients nobody has priced.
helpWhat is a good food cost percentage for an Indian restaurant?
Be careful with any number quoted as a standard, including the ones you will find repeated across the internet, because almost nobody says what they included. Some figures count only raw ingredients, some add packaging, some add gas and staff meals. A percentage calculated one way is not comparable to a percentage calculated another. The number worth having is your own, measured the same way every month, so the direction it moves means something. A dish-by-dish cost tells you more than any benchmark: it tells you which plate is the problem.
helpWhy is my food cost wrong when the maths looks right?
Two reasons, in this order. First, units: a recipe written in grams against stock bought by the kilo will be out by a factor of a thousand if nobody converts it, and 100 g of a Rs 120/kg mango reads as Rs 12,000 instead of Rs 12. Second, and quieter, an ingredient with no purchase price yet contributes zero, so a dish whose most expensive component has never been entered looks almost free. The first error is obvious the moment you see it. The second one is believed.
helpShould Half and Full be costed separately?
Yes, and against the price each one is actually sold at. They use different amounts of the same things, so one cost cannot describe both, and a margin measured against a price you never charge is fiction. If a dish has sizes, cost every size and look at the worst one — that is the plate quietly funding the others.
helpHow often should food cost be recalculated?
It should not be a recalculation at all. If the cost is derived from what you last paid for each ingredient, then onions getting dearer this week moves every dish that uses onions, by itself, the same week. A food cost you have to sit down and redo is a food cost that is out of date between sittings — and it will be most out of date exactly when prices are moving fastest.
helpDo I need to count stock to know my food cost?
No, and the two questions are worth keeping apart. What a plate costs is a recipe question, answered from your purchase rates. What you have left is a stock question, answered by counting. The first is useful on its own and can be set up in an afternoon; the second is a bigger job. Start with the plate.
helpWhat about gas, packaging and staff meals?
They are real costs and they are not ingredient costs. Keep them out of the plate figure so it stays comparable dish to dish, and account for them where they belong — packaging against parcel orders, gas and staff meals as running costs for the month. Mixing them in is how two restaurants end up quoting food cost percentages that cannot be compared.
helpCan a POS calculate food cost automatically?
It can, if it knows two things: what each dish is made of, and what you paid for each ingredient. Dinelax does this from the purchase entries you already make — write the recipe once and the cost, the margin and the worst-margin size appear against the dish and move on their own as your rates change. It costs Rs 149 a month for billing and Rs 999 for everything, with fourteen days free.