How to lower food cost with actual vs theoretical variance

Theoretical food cost is what food should have cost given what you sold: each item's recipe cost times the number sold, added up. Actual food cost comes from inventory: what you started with, plus what you bought, minus what's left. The difference is variance — product that left the building without being sold.

Variance is the most useful food cost number a store has, because it's the part the store can fix. Supplier prices and menu mix move food cost too, but those show up in theoretical as well. Variance is what's left after the menu has been accounted for: waste, portions, receiving and counting.

The formulas

• Theoretical food cost = sum of (recipe cost per item × units sold) for every item sold.
• Theoretical food cost % = theoretical food cost ÷ net sales × 100.
• Actual food cost = beginning inventory + purchases − ending inventory.
• Actual food cost % = actual food cost ÷ net sales × 100.
• Variance = actual food cost − theoretical food cost, in dollars or in percentage points.

Recipe cost means the cost of every ingredient in one item as built to spec, priced from current invoices. A burger's recipe cost includes the patty, bun, cheese, sauce, produce and wrapper. Many POS and inventory systems calculate theoretical cost for you once recipes are entered. If yours doesn't, start with your top sellers.

Greeley's menu should cost 28.5% of sales. It actually cost 31.2%. The 2.7-point gap is $675 a week. Nobody at the store would notice $675 in a week, but at that rate it's $35,100 a year, which is often a large share of what a single store earns.

Where variance usually comes from

• Waste that isn't logged: expired product, dropped food, remakes.
• Portions: an extra ounce across hundreds of orders.
• Receiving: short deliveries signed for, or substitutions at a higher cost.
• Recipe costs not updated after a supplier price change.
• Comps and staff meals that weren't rung in.
• Counting errors, which make variance jump from week to week.

Portions are usually the biggest and quietest. Nobody sets out to over-portion. A scoop gets replaced with a slightly larger one, a new cook builds by eye, a spec card falls off the wall. Each ounce is small, but it repeats on every order.

That one portion drift accounts for about 40% of Greeley's weekly variance. It's also one of the easiest to fix: a scale at the station, a portion check at the start of each shift, and the spec posted where the cook can see it.

How to run counts without drowning

• Count your top 15–20 items by cost every week. They're most of your spend.
• Same day, same time, same person, in units rather than dollars.
• Do a full count monthly.
• Log waste on a sheet at the line, every shift.
• Track variance per item. The total says there's a gap; the item says where.

A weekly count of 15 to 20 items takes most stores well under an hour once it's routine. The time to count is before the first delivery of the week and before prep starts, so product isn't half in the walk-in and half on the line.

Reading variance by item

A total variance of 2.7 points says something is wrong. It doesn't say what. For each counted item, compare how much should have been used (units sold × recipe quantity) with how much was used (beginning + received − ending). Then rank items by dollar gap.

The pattern tells you where to look:

• One protein far over, everything else close: portions or waste on that item.
• Many items slightly over: a counting or receiving problem, or recipe costs that are out of date.
• An item over one week and under the next: counting error. Check units and where product is stored.
• Paper and packaging over: delivery and takeout mix changed, or extra bags and cups are going out.

Fixes that usually work

Once an item points somewhere, the fix is usually a habit rather than a new system.

• Receiving: one trained person checks each delivery against the invoice before the driver leaves, weighs protein, and writes any shortage or substitution on the invoice. Credits get claimed the same day.
• Waste: a sheet at the line with item, quantity and reason. The reason matters most. Expired product points at prep pars that are too high; remakes point at ticket accuracy or training.
• Prep pars: set them from a forecast for that day, not from habit. A par built for a busy month keeps producing waste after sales soften.
• Portions: scales and portion tools at the station, spec cards posted, and a quick check at the start of each shift.
• Recipe costs: update them when invoices change, or theoretical drifts away from reality and the variance stops meaning anything.
• First in, first out: date labels and older product in front, so the oldest stock is used first.

What good looks like

Many operators aim to keep actual within one to two points of theoretical. The trend matters more than the level: a store at 1.5 points every week is in control, and a store that moves from 1 to 3 has changed something.

Compare theoretical food cost with your brand's median too. If theoretical itself is high, the problem isn't the store's execution. It's menu mix or pricing, and the fix is a menu or price review rather than tighter counts. How to raise menu prices without losing customers.

If variance points at missing product, look at process first — receiving, portioning, waste logging — and follow your HR policies and local law before drawing conclusions about any person.

A weekly food cost checklist

1. Update recipe costs for any item whose supplier price changed.
2. Check every delivery against the invoice at the door: count, weigh and note substitutions.
3. Keep a waste sheet at the line and have the shift lead sign it at close.
4. Count the top 15 to 20 items by cost, same day and time, in units.
5. Calculate actual and theoretical food cost and the variance in points.
6. Rank counted items by dollar gap and pick the top one or two.
7. Make one change for the coming week, such as a portion check on the item with the biggest gap.
8. Check that item's variance next week before moving on to the next.

Common mistakes

• Looking at actual food cost alone. A rise can be a supplier increase that hit theoretical too, which counting harder won't fix.
• Pricing counts at old costs, which hides the increase that caused the rise.
• Counting in dollars instead of units. Unit errors are easy to see; dollar errors aren't.
• Changing who counts, or when, every week. The variance then measures the counting, not the store.
• Reacting to one week's jump before checking for a count error.
• Treating comps and staff meals as waste. Ring them in so they're accounted for and visible.
• Jumping to theft. Most variance is process, and the fix is a scale or a receiving habit.

Food cost is half of prime cost, so check it with labor beside it. How to calculate prime cost.

Where Franlever fits

With an inventory or accounting connection, Franlever reads food cost beside each store's sales and labor, so it can be seen weekly rather than only when the P&L closes. It compares each store with its own normal range and with the brand's median, and when food cost moves, AI COO can carry it as that store's one lever for the day, sent to the GM by email or text. See how it works for multi-unit operators.

General information for restaurant and franchise operators, not legal, tax, accounting or financial advice. Figures in examples are illustrations, not predictions or promises of results. Laws and platform policies change and differ by location, so check the current rules and talk to a qualified advisor about your situation.