Your food cost is probably wrong. Here's where it breaks.

Most kitchens cost a recipe using the price printed on the invoice. That number is almost never what the ingredient actually costs you on the plate — and the gap is bigger than most operators think.

Ask a chef what their salmon costs and you'll usually get the invoice price. $9.85 a pound. It's the number on the sheet, it's the number in the spreadsheet, and it's wrong — not because anyone made a mistake, but because the invoice describes what the vendor sold you, not what ends up on a plate.

Food cost goes wrong in four specific places. None of them are exotic, all of them are fixable, and together they routinely account for several points of food cost percentage.

1. Yield: you throw away part of what you buy

This is the big one. A whole salmon arrives at 10 pounds. After you remove the head, frame, collar, belly trim and pin bones, you have somewhere around 6.2 pounds of portionable fillet. You paid for ten pounds. You can sell six.

The industry terms are AP (as purchased) and EP (edible portion), and the relationship is simple:

At 62% yield, that $9.85 salmon is really:

Three dollars a plate. On a 200-cover banquet that's $604 that never appears in your costing, on one ingredient, on one menu item.

And yields vary far more than people assume — not just by ingredient but by season, by vendor, and by who's holding the knife. A cook who takes the fillet close to the bone and a cook who doesn't can be six or seven points apart on the same fish.

Where the trim goes matters too

If those salmon frames become stock and the belly trim becomes rillettes, your effective yield is higher than 62% — you're selling part of what you'd otherwise write off. Most costing systems have no way to represent that, so kitchens that use their trim well get no credit for it in the numbers.

2. Unit conversions: the invoice and the recipe don't speak the same language

Here's a costing problem that shows up in every kitchen and almost never gets solved properly.

You buy heavy cream by the gallon — say $8.00. Your panna cotta recipe calls for cups. Your broccolini comes in a case at $23.48; your recipe calls for bunches. Your flour arrives in a 50 lb bag; your pastry recipe calls for 200 grams, or sometimes 2 cups.

Three different conversion problems sitting in one walk-in:

• Within a family — gallons to cups, pounds to ounces. Fixed math, but somebody still has to do it, every time, correctly.
• Across families — weight to volume. There is no universal answer; it depends on the ingredient. A cup of flour is about 0.28 lb. A cup of heavy cream is about 0.52 lb. You need a density for each item.
• Vendor units — a "case" of broccolini is 18 bunches at your vendor and 24 at someone else's. That conversion is specific to the item and the supplier, and it changes.

When these get handled by mental arithmetic under pressure, they don't get handled consistently. The panna cotta gets costed once when the recipe is written and never again.

3. Sub-recipes: the cost hides one level down

Very little on a menu is made from raw ingredients alone. An entrée contains a sauce, which contains a stock, which contains mirepoix and bones. A composed salad contains a vinaigrette made in two-quart batches.

Each of those layers has its own yield and its own conversions. A stock that reduces from 6 quarts to 4 has a yield of 67%, and if you cost it at 6 quarts you've understated every dish that uses it.

Butchery is the clearest case. If your whole salmon goes to a butcher who cuts 8 oz portions, the recipe doesn't call for "salmon" — it calls for a portion that already has trim loss baked into it. Treating that portion as its own item, with its own cost derived from the whole fish, is the only way the math stays honest when the line cook pulls one from the reach-in.

4. Price drift: you costed it once

The fourth problem is time. A recipe costed in March using March prices is quietly wrong by June, and badly wrong by the following March. Proteins and produce move constantly, and nothing in a spreadsheet tells you when it happened.

The practical damage isn't the drift itself — it's that you find out about it at the end of a period, in aggregate, when your food cost percentage is up two points and you have to go hunting for the cause.

One caution on recosting

There is a real tension here. You want current prices for menu engineering and purchasing decisions. But if you've already quoted a banquet or printed a menu at a given cost, silently recalculating that historical number destroys your ability to look back and understand what actually happened. Live costs for decisions, frozen costs for history.

What to actually do about it

You don't need software to make progress on this. In order of return:

1. Yield-test your ten highest-volume proteins. Weigh in, break down, weigh the usable product. Ten items, one afternoon. This single exercise typically uncovers more cost than everything else combined.
2. Write the conversion down once, next to the ingredient. Cups per gallon, bunches per case, grams per cup. Not in someone's head — on the item.
3. Cost your sub-recipes separately and use the result inside the parent recipe, rather than costing the parent from raw ingredients.
4. Recost your top twenty sellers quarterly. Not everything. The twenty items that drive your volume.

Everything above is arithmetic. It's not difficult — it's just relentless, and doing it by hand across a few hundred recipes is where kitchens give up. That's the part worth automating, and it's the reason we built PlateCost.