Theoretical vs. Actual Food Cost: How to Find the Variance

Most operators track food cost. Far fewer track the gap between the food cost they should have and the food cost they actually have. That gap, called variance, is where most of the avoidable losses hide.

Your food cost percentage tells you whether costs are high. Variance tells you why, and points you at the exact items to fix. Without it, you can see that costs are up but not where the money is going.

The Two Numbers

Whether you say "theoretical vs. actual" or "actual vs. theoretical," the comparison is the same, and it fits in one sentence: theoretical food cost is what your sales say you should have used; actual food cost is what your inventory counts say you really used; the difference is your variance.

Theoretical usage is what you should have used, based on what you sold. If you sold 200 burgers and each burger spec calls for 6 oz of beef, you should have used 1,200 oz of beef. Do that for every ingredient across every dish and you get your theoretical usage for the period.

Actual usage is what you really used, measured by counting:

Actual Usage = Beginning Inventory + Purchases - Ending Inventory

This is the same formula behind COGS. It captures every ounce that left the shelf, for any reason.

Variance is the gap:

Variance = Actual Usage - Theoretical Usage

As a percentage:

Variance % = (Actual - Theoretical) / Theoretical x 100

Theoretical assumes every plate hits spec, nothing spoils, and nothing leaves out the back door. Actual is what really happened. The gap between them is normal for any working kitchen, and most of it is something you can control.

A Quick Example

One ingredient, one week of beef:

Line	Value
Burgers sold (POS)	200
Spec per burger	6 oz
Theoretical usage	1,200 oz
Beginning inventory	900 oz
Purchases	1,600 oz
Ending inventory	950 oz
Actual usage	1,550 oz
Variance	+350 oz
Variance %	+29%

You sold 1,200 oz worth of burgers but used 1,550 oz of beef. That extra 350 oz, nearly 22 pounds, became nothing you got paid for. At $6/lb, that is about $131 gone in one week on one item. Annualized, one ingredient is leaking roughly $6,800.

Your recipe costing will never show that loss. You only catch it by comparing what you used against what you sold.

What the Gap Is Made Of

A positive variance (you used more than you sold) comes from a predictable set of causes. In rough order of how often they are the culprit:

• Over-portioning. The single most common cause. A 6 oz spec that drifts to 7 oz is a 17% overage that nobody notices, because every plate still looks right.
• Yield and trim loss. Recipes are often written in edible-portion weight, but you buy and count in as-purchased weight. If your recipe does not account for trim, the variance absorbs it. (This is why yield factors matter.)
• Waste and spoilage. Over-prepped items that do not sell, produce that turns, proteins past their window. Real, but usually smaller than people assume until they measure it.
• Comps, staff meals, and tastings. Product that left the inventory but never hit a sales line. Legitimate, but it has to be recorded or it shows up as variance.
• Receiving errors. Short deliveries and wrong substitutions you paid for but never received. The invoice says 40 lbs, the box held 36.
• Theft. Real and worth taking seriously, but usually last on the list, not first. Chase the boring causes before you suspect people.

A small variance is normal. A large or growing one usually traces back to a specific cause you can find.

What Counts as a Good Variance

These are working benchmarks, not hard rules. The right target depends on your concept and how tightly you portion.

Variance (actual vs theoretical)	What it means
Under 2-3%	Normal. Hard to eliminate. Do not chase it to zero.
3-5%	Worth a look. One or two items are usually doing the work.
5-8%	A real leak. Something specific is broken and findable.
Over 8%	Urgent. Start with your highest-volume, highest-cost items.

Bars run tighter because liquor is easy to over-pour and easy to walk. A pour cost variance above 5% on spirits is a common early sign of free pours or unrecorded comps. High-volume kitchens should also aim tighter, because a 3% variance on a large base is a large dollar figure.

Use the percentage to spot the problem, then rank the fixes by dollars.

Find the Money, Not the Percentage

The mistake operators make is sorting their variance report by percentage. A 40% variance on cinnamon costs you almost nothing. A 4% variance on your top protein can be thousands of dollars a year. Always rank by dollar impact, not percent.

The fastest way to find a leak:

1. Pull your POS product-mix report for the period. This is your sold quantity per menu item.
2. Multiply each item by its recipe cost to get theoretical usage and theoretical cost.
3. Compare to actual usage from your counts.
4. Multiply each variance by unit cost and sort high to low.
5. Work the top five. Eighty percent of your recoverable variance lives in a handful of items, almost always your proteins, dairy, and high-pour liquor.

For each of those five, walk the causes in order: weigh ten plates against spec (portioning), check the waste log (spoilage), verify last week's deliveries against invoices (receiving), and only then look harder at access and theft.

Why This Is the Number Software Exists For

Anyone can calculate food cost from invoices. Variance is harder, because it requires three things at once: accurate counts, complete recipes, and real sales data, all in the same units. That is exactly the work traditional inventory software was built to do, and exactly where it tends to break:

• Unit mismatches. You buy beef by the case, portion it by the ounce, count it by the pound. If those units do not reconcile, the variance is garbage and you cannot trust it.
• Stale counts. Variance needs a fresh beginning and ending count. Skip one and the next two periods are wrong. (See the weekly counting habit.)
• Incomplete recipes. A dish with no recipe contributes sales but no theoretical usage, so its ingredients all look like variance.

Get those three right and you can run variance every week instead of once a quarter. What you get back is a short list: which five items to check, and roughly how many dollars you can recover by fixing them.

Start Small

You do not need to compute variance for your entire catalog to get value. Pick your five most expensive, highest-volume ingredients, the ones that are most of your food cost. Get clean recipes and clean counts for just those. Run the comparison for one period.

If the gap is under 3%, you are in good shape and can move on. If it is 5% or more on even one of them, you have found real money, and the cause is almost always sitting in plain sight on the line.

---

Further Reading

• What Is COGS in a Restaurant?. The actual-usage side of the equation, in detail.
• How to Calculate Food Cost Percentage. The number variance explains.
• Restaurant Prime Cost Formula & Benchmarks. Where food cost fits in the bigger profitability picture.
• How to Calculate Plate Cost. Recipe costing and yield factors, which feed theoretical usage.
• Restaurant Inventory Counting: A Practical Guide. Accurate counts are what make variance trustworthy.

---

Sources

• Restaurant365, "Theoretical vs. Actual Food Cost": definitions and variance benchmarks for full-service operations.
• Toast, "Restaurant Food Cost": product-mix and theoretical usage methodology.
• National Restaurant Association, 2026 State of the Restaurant Industry: cost pressure context.