Your P&L says food cost is 34%. Your menu math says it should be 29%. Somewhere between the recipe card and the trash can, five points of margin disappeared — and nothing on the P&L tells you where.
That gap has a name: variance. Measuring it is the single most direct way to find money your kitchen is losing, and you don't need enterprise software to do it.
Two numbers, one gap
Actual food cost is what you really spent relative to what you sold. It comes from inventory counts and purchases — reality, with all its waste, over-portioning, and mystery shrinkage baked in.
Theoretical food cost is what you would have spent if every plate went out exactly to recipe. No drift, no drops, no staff grazing the line. It comes from your recipes and your sales mix.
The difference between them is recoverable profit. A healthy kitchen runs 2–3% variance — some gap is unavoidable. Above 5%, money is walking out the door through a systemic problem, not bad luck.
If you haven't diagnosed the basics of a high food cost yet — supplier creep, stale recipe costs, over-ordering — start with the food cost guide. Variance analysis is the next level: it tells you not just that you're losing money, but on which products and how much.
How to calculate theoretical food cost
You need three things: recipes with exact quantities, current ingredient costs, and a count of what you sold.
Then for every menu item sold, multiply each ingredient quantity by its cost, sum the plate, and multiply by units sold. Add it up across the whole menu and you have what the period should have cost.
Here's a burger, costed honestly:
| Ingredient | Quantity per burger | Unit cost | Plate cost |
|---|---|---|---|
| Brioche bun | 1 each | $0.55 each | $0.55 |
| Ground beef | 6 oz | $4.80/lb | $1.80 |
| American cheese | 2 slices | $0.18 each | $0.36 |
| Lettuce, tomato, onion | 2 oz | $1.60/lb | $0.20 |
| Fries | 5 oz | $1.10/lb | $0.34 |
| Total | $3.25 |
Sell 450 burgers in a week and your theoretical spend on burgers is $1,462.50. Your theoretical beef usage alone is 450 × 6 oz = 168.75 lb. Hold that number — it matters in a minute.
Where most independent restaurants get stuck isn't the math. It's the inputs: recipes that live in the chef's head instead of a system, ingredient costs from when the menu launched two years ago, and sales data trapped in the POS. Solve those three and the calculation is just multiplication.
How to calculate actual food cost
This one you may already know:
Beginning inventory + purchases − ending inventory = actual usage
The formula is simple, but it's only as good as the counts behind it. Skip the counts and you're measuring purchases, not usage — which hides exactly the problems you're trying to find. If your counts are inconsistent or slow, fix that first; the inventory counting guide covers how a small team counts a full kitchen in under an hour.
One rule trips people up: the period has to match. If your sales data covers Monday through Sunday, your beginning count is Monday morning and your ending count is the following Monday morning, with every invoice in between. Mismatched windows produce variance numbers that look alarming and mean nothing.
Reading the variance: what the numbers tell you
Run both calculations per product and compare. Back to that beef.
Say your Monday count showed 80 lb of ground beef, you received 200 lb during the week, and the next Monday you counted 95 lb. Actual usage: 80 + 200 − 95 = 185 lb. Theoretical usage was 168.75 lb.
That's 16.25 lb of beef — about $78 — gone in one week with no burger to show for it. Roughly $4,000 a year, from one ingredient.
Positive variance (actual above theoretical) is the common direction. Causes: over-portioning, unlogged waste, staff meals nobody records, theft.
Negative variance (actual below theoretical) looks like good news and usually isn't. It means under-portioning — your guests are getting shorted — or your recipes overstate what the kitchen really uses, which corrupts every other number downstream.
The most important discipline: rank by dollar impact, not percentage. A 50% variance on parsley is a rounding error. A 3% variance on ribeye is real money. Sorted by dollars, your top five products usually account for most of the total gap — which is why Rinvy's Variance Report ranks every product by dollar impact rather than burying you in a 200-row spreadsheet.
The five most common causes of high variance
- Over-portioning of proteins. The 6 oz spec poured at 8 oz. Most expensive ingredient, most expensive habit.
- Unlogged waste. The dropped tray and the wilted case both raise actual usage. If they're not logged with a reason, they're indistinguishable from theft.
- Recipe inaccuracy. The recipe says one thing; the line has quietly evolved. Variance analysis assumes recipes reflect reality today.
- Unrecorded staff meals and comps. Food consumed, never rung in. Theoretical usage stays flat while actual climbs.
- Theft. Persistent variance concentrated on specific high-value items — steaks, liquor, seafood — that doesn't respond to retraining is a different conversation.
Setting it up at your restaurant
You can do this with spreadsheets. Plenty of operators start there. The setup is the same either way:
- Digitize your recipes with exact quantities. Start with your ten best sellers — they drive most of your usage.
- Connect recipes to menu items so each POS item maps to a costed plate.
- Keep ingredient costs current. Costs should come from what you actually paid on recent invoices, not from launch-day pricing. If invoice entry is the bottleneck, the invoice automation post covers how to get line items in without typing.
- Get your sales data out of the POS. A weekly CSV export of items sold is enough.
- Count inventory on a consistent schedule with matching period boundaries.
The spreadsheet version of this takes a few hours a week and breaks the first time a recipe changes and nobody updates the formula. The software version automates the multiplication: Rinvy's Variance Report computes theoretical usage from your sales and recipes, actual consumption from your counts and receipts, and ranks the gap by dollar impact. It needs the same inputs — digitized recipes linked to menu items, a sales CSV, consistent counts — and it's on the Pro plan at $69/month with a 60-day free trial, which is long enough to find out whether your gap is bigger than the subscription.
From variance to action
The report doesn't fix anything by itself. The fixes map to the causes:
- Over-portioning — portion tools sized to spec, recipe cards at every station, weigh a plate mid-service once a week.
- Unlogged waste — log everything thrown away, with a reason. Two weeks of data tells you whether the problem is ordering, prep, or storage.
- Recipe drift — watch the dish get made, update the recipe to match reality or retrain to the standard. Pick one; don't let them disagree.
- Suspected theft — count the high-variance items more often. Daily counts on five products take minutes and change behavior fast.
Then set a target — under 3% is a good one — and review weekly. The first report is usually the ugly one. That's the point. Variance you can see is variance you can close; the version you can't see just keeps compounding, a few invisible pounds of beef at a time.