The formula
Actual usage = Beginning inventory + Purchases – Ending inventory
Variance ($) = Actual usage – Theoretical usage
Variance (% of sales) = Variance ($) / Food sales x 100
Why two different usage numbers exist at all: theoretical usage is built forward from the menu, multiplying every dish sold by its recipe cost, so it describes a perfect kitchen where every portion is exact and nothing is wasted, dropped or given away. Actual usage is built from the storeroom, counting what physically left the shelves. Neither number is wrong; they measure different things, and the distance between them is everything that happened in the real kitchen that the recipe file does not know about: over-portioning, trim and spoilage, unrecorded waste, comps, staff meals and plain count errors.
Judging the gap as a percentage of sales keeps it comparable between a quiet week and a record week. Most well-run kitchens keep variance under 2% of sales; 2 to 3% is acceptable if stable, 3 to 5% signals something structural, and above 5% is a systemic problem. Full-service kitchens sit at the looser end of those bands; quick service should sit closer to 1.5 to 2%.
How to use it, with an example
- Choose one period and take every figure from that exact period, no mixing of weeks and months.
- Enter the beginning and ending inventory values from real counts, and total food purchases in between.
- Enter the theoretical usage from your recipe costing or POS report for the dishes actually sold in the period, and the food sales for the same dates.
- Read the variance in dollars, as a percentage of sales, and against theoretical usage, then use the verdict to decide whether to watch it or chase it.
Worked example: A full-service restaurant counts $8,500 of food at the start of the month, buys $24,500 during it, and counts $7,900 at the end, so actual usage is $8,500 + $24,500 – $7,900 = $25,100, an actual food cost of 27.3% on $92,000 of food sales. Its recipes say the dishes sold should have used $23,800, a theoretical food cost of 25.9%. The variance is $1,300 over theoretical, 1.4% of sales and 5.5% above the theoretical figure itself: inside the well-managed band, the ordinary friction of a real kitchen rather than a leak. The calculator above is prefilled with these same numbers, so you can see a healthy result before you enter a worrying one.
Variance only stays visible when counts, invoices and recipe costs are kept in one place. In Odoo for Restaurants, the Odoo Inventory module holds stock movements and product costs against the same items your recipes draw from, so actual usage per period is a report you can pull weekly.
Common mistakes
- Comparing periods that do not match. Theoretical usage for four weeks against actual usage for a month with five weekends manufactures a variance out of nothing. Lock the dates on both sides before drawing any conclusion about the kitchen.
- Trusting a count that was rushed. A sloppy ending count is the most common cause of phantom variance in both directions. Count in the same order, with the same sheet, by two people on high-value items, the discipline covered in restaurant inventory management tips.
- Chasing theft first. Theft exists, but it is rarely the first explanation. Outdated recipe yields, portion drift on two or three high-cost items and unrecorded waste explain most gaps, and they are all fixable without an accusation. Work the biggest ingredients first.
- Treating a negative variance as good news. Using less than theoretical usually means the count is wrong, invoices were booked into the wrong period, or recipes are costed too high. A variance you cannot explain in either direction is a data problem.
- Averaging the whole kitchen into one number. A 1.4% total can hide a 15% variance on beef inside accurate dry goods. Run the same maths per category, then per item, where the leak actually lives and where how to reduce restaurant food cost starts its fixes.
- Recording waste nowhere. Food that is cooked and binned without a waste log lands entirely inside variance, indistinguishable from over-portioning. A simple waste sheet separates the two, as set out in how to manage food waste in restaurants.
FAQs
What is an acceptable inventory variance for a restaurant?
Under 2% of food sales is well managed for most restaurants, and quick-service operations often aim for 1.5 to 2% given their tighter menus. Two to 3% is acceptable if it is stable rather than growing. Between 3 and 5% something structural is usually wrong, and above 5% of sales most operators treat it as a systemic problem to investigate immediately rather than a number to watch.
Can variance be zero?
No, and zero should make you suspicious rather than pleased. Real kitchens lose small amounts to trim, spillage, sampling and evaporation every day, so a true zero usually means the theoretical recipes are padded or the count was adjusted to fit. A small, stable, positive variance is the sign of honest data.
What are the most common causes of high variance?
In rough order of frequency: over-portioning on high-cost proteins, recipe yields that no longer match the product being delivered, waste and spoilage that never gets recorded, count errors, supplier short-weighting, and then theft. Price rises are not variance, since theoretical usage should already be costed at current prices.
Should variance be measured weekly or monthly?
Weekly for high-cost categories, monthly for the whole inventory at minimum. A weekly rhythm catches a portioning problem while it is still a few hundred dollars; a monthly-only rhythm finds the same problem after it has compounded for four weeks. The count does not need to be a full stocktake each week, just the main storage areas, valued consistently.
How do staff meals and comps affect variance?
They are real usage that no recipe sold, so they appear inside variance unless you record them. Ring staff meals and comps through the POS as zero-priced items with recipes attached, and the theoretical figure absorbs them, leaving variance to show only the genuinely unexplained gap.
