· Orqestra Team · 7 min read
Theoretical vs Actual Food Cost: Formula and Variance
Theoretical vs actual food cost explained: both formulas, how to find the variance, what causes it, a worked weekly example, and the targets operators use.
Theoretical food cost is what your food should have cost based on recipes and what you sold: recipe cost per portion times portions sold, summed across the menu. Actual food cost is what it really cost, from inventory: beginning inventory plus purchases minus ending inventory. The difference is your food cost variance, and it is the single best measure of how tightly the kitchen is run. Well-run restaurants keep the variance under 2 percentage points.
Food cost percentage alone tells you the number is high. Comparing theoretical to actual tells you whether the problem is on the menu (pricing, recipes) or in the kitchen (waste, portions, theft). That split decides what you fix.
Theoretical food cost
Theoretical food cost assumes a perfect kitchen: exact portions, zero waste, no theft, every item rung up. Restaurant365 defines it as what your food costs "should be" for a period, and meez describes it as the benchmark built from recipe data, ingredient prices, and the POS sales mix.
To calculate it:
- Cost every menu item from current invoice prices, adjusted for yield (the usable part of what you buy).
- Pull item-level sales from your POS for the period.
- Multiply each item's recipe cost by the number sold.
- Add them all up. That is theoretical food cost in dollars.
- Theoretical food cost % = theoretical food cost ÷ food sales × 100
The quality of this number depends entirely on your recipes. A recipe costed in 2024 with 2026 prices, or one that ignores the trim on a case of chicken, produces a theoretical number that is too low, which makes the variance look worse than it is.
Actual food cost
Actual food cost comes from counts, not sales. The formula, as Supy and meez both give it:
Actual food cost = beginning inventory + purchases - ending inventory
Actual food cost % = actual food cost ÷ food sales × 100
This is the same number as your period food cost percentage. It captures everything that left the shelves, whether it was sold, wasted, overportioned, eaten by staff, or walked out the back door.
The variance
Variance = actual food cost % - theoretical food cost %
Express it in percentage points, and also in dollars (actual $ minus theoretical $) so the kitchen can see what it is worth. meez adds a second version, variance dollars ÷ total sales × 100, which gives the same answer when both costs are measured against the same sales.
The variance is almost always positive. A negative variance (actual below theoretical) usually means a counting error, a recipe costed too high, or inventory that was received but not entered as a purchase.
A worked example: one week
A 70-seat casual restaurant, the week of October 19 to 25, 2026. Food sales for the week: $22,800.
Theoretical side. The five best sellers are shown individually; the rest of the menu is costed the same way and grouped.
| Menu item | Recipe cost per portion | Sold | Theoretical cost |
|---|---|---|---|
| Burger | $4.05 | 310 | $1,255.50 |
| Chicken sandwich | $3.60 | 240 | $864.00 |
| Caesar salad | $2.80 | 150 | $420.00 |
| Fish tacos | $4.90 | 180 | $882.00 |
| Pasta | $3.20 | 200 | $640.00 |
| All other items (costed individually) | $2,950.00 | ||
| Theoretical food cost | $7,011.50 |
Theoretical food cost % = $7,011.50 ÷ $22,800 × 100 = 30.8%
Actual side. From the Sunday night counts and the week's invoices:
- Beginning inventory (Oct 19): $9,100
- Purchases received during the week: $7,350
- Ending inventory (Oct 25): $8,900
- Actual food cost = $9,100 + $7,350 - $8,900 = $7,550
- Actual food cost % = $7,550 ÷ $22,800 × 100 = 33.1%
Variance = 33.1% - 30.8% = 2.3 percentage points, or $7,550 - $7,011.50 = $538.50 for the week.
That $538.50 is food that was bought and used but never matched to a sale. Over 52 weeks at the same pace it is about $28,000. It is also a number the chef can act on, which "food cost is 33%" is not. The food cost calculator handles both sides of this math if you want to run it without a spreadsheet.
What the variance means: target ranges
There is no official industry standard for acceptable variance, so here is what the sources we opened say operators target.
| Variance (percentage points) | meez | Supy |
|---|---|---|
| Under 1 | well-managed | excellent, typical of daily counts and automated purchasing |
| 1 to 2 | well-managed | acceptable, monitor weekly |
| 2 to 3 | acceptable for most concepts, monitor trends | investigate, something is systematically off |
| 3 to 5 | common, but room for improvement | urgent (above 3) |
| Over 5 | investigate immediately | urgent |
Sources: meez, Supy. Restaurant365 says best-in-class operators aim for "around 1% or less."
Two caveats. These figures assume weekly measurement; Supy notes that monthly measurement masks problems. And they assume good recipes. If your theoretical number is built on stale prices, a 2-point variance might really be 1, or 3.
The restaurant in the example, at 2.3 points, is in the "look into it" zone by either source. Not a crisis, but worth a week of attention.
What causes the variance, and how to find each cause
| Cause | What it looks like | How to detect it |
|---|---|---|
| Overportioning | Actual above theoretical on specific proteins or cheese | Weigh ten plates of a top seller at random during service; compare to the recipe card |
| Waste and spoilage | Produce and dairy usage above sales | Keep a waste log for two weeks; count walk-in items by expiry date |
| Theft | Usage of high-value, easy-to-carry items (steaks, shrimp, liquor) above sales | Track those items daily against sales; check for voids and after-close activity |
| Unrecorded comps and staff meals | Food leaves the kitchen with no ticket | Require a POS ticket for every plate, including comps, so they appear in sales mix |
| Pricing or invoice errors | Purchases rise with no change in volume | Match every invoice line against the agreed price list; look for unit-size changes |
| Incorrect or stale recipes | Variance is steady across all items, not concentrated | Re-cost the top 20 items from current invoices and check yields |
| Inaccurate counts | Variance swings up and down week to week | Have the same person count at the same time with the same units; recount high-value items |
| Receiving errors | Inventory lower than purchases imply | Weigh and count deliveries against the invoice before signing |
Start with the item-level view. If the variance sits in two or three ingredients, the cause is in the kitchen and the fix is specific. If it is spread evenly across everything, the cause is usually recipes or counts, and the fix is in the office.
How to run this weekly without a full-time analyst
- Count the 30 to 50 items that make up most of your spend every week, on the same day, at the same time. Everything else gets a monthly count.
- Keep recipe costs current. Update the top sellers whenever a key invoice price moves more than a few percent.
- Pull item sales from the POS for the same seven days.
- Compute theoretical, actual, and variance. Review it with the kitchen manager on Monday.
- Pick one cause to chase each week. Do not try to fix all eight at once.
Keeping counts, purchases, and par levels in one place makes step 1 and step 4 fast; Orqestra's inventory module is built for that weekly loop, and the same discipline works in a spreadsheet. Setting par levels also helps, because overordering is the quiet cause of spoilage.
Common mistakes
- Measuring monthly. By the time a monthly variance shows up, four weeks of waste have already happened.
- Trusting old recipes. A theoretical number built on 2024 prices makes 2026 variance look worse than it is.
- Ignoring yield. Costing a chicken breast at the per-pound price of the whole bird understates theoretical cost on every chicken dish.
- Not costing the "other" items. If 30% of the menu is uncosted, theoretical cost is a guess and so is the variance.
- Letting comps skip the POS. A plate that leaves the kitchen without a ticket is pure variance.
- Inconsistent counts. Different people, different times, different units. The variance then measures your counting, not your kitchen.
- Treating 2 points as fine forever. Two points on $22,800 a week is over $23,000 a year. The target is to trend down, not to stay comfortable.