Inventory Shrinkage: Finding the Missing Stock

The gap nobody can explain

Every operator has had this month. Sales were fine, prices did not move, the menu did not change, and food cost still landed two or three points above where it should have. You go through the invoices and everything reconciles. You check the menu prices and they are correct. The money left, and there is no document that says where.

That gap has a name. Inventory shrinkage is the difference between the stock your sales should have consumed and the stock your count says actually moved. It is not a cost line you can look up — it is a subtraction, and the only way to see it is to calculate both sides and compare.

The scale is easy to underestimate because the percentages sound small. A site doing $70,000 of monthly food sales with 1.5% unaccounted shrinkage is losing $1,050 a month, or $12,600 a year, entirely from net profit. At a 10% net margin, replacing that through trading would take $126,000 of additional sales. Most operators would work extremely hard for that revenue and barely glance at the shrinkage.

The reason it persists is that it hides inside a number everyone already watches. It shows up as a slightly worse food cost percentage, which gets attributed to supplier price rises, and the investigation stops there.

Building both sides of the comparison

Actual usage is the easier half. It comes from the count and it cannot argue with you: opening stock plus purchases minus closing stock. Whatever that number is, that much product left the building during the period.

Theoretical usage is the half that takes work. You need every item sold in the period, multiplied by the current costed recipe behind it. Sell 310 burgers at a costed $3.40 of ingredients and you should have consumed $1,054 of stock on burgers. Do that across the menu and you have what the period demanded.

This is where most shrinkage projects fall over, because the recipe file is the weakest document in the building. It was built when the site opened, updated selectively, and it is now missing three specials, two seasonal swaps, and every price rise since spring. A theoretical figure built on that file measures the age of your paperwork rather than the state of your stock room.

Two fixes carry most of the weight. Re-cost anything whose main ingredient has moved more than 10% since it was last touched, and make sure every recipe uses net yields rather than as-purchased weights. Costing a whole fish at purchase weight turns every gram of legitimate trim into phantom shrinkage, which is covered in more depth in our guide to yield percentages.

One more trap: sales channels. Delivery, catering, functions and staff tills all consume stock. If your sales extract misses one, that channel’s usage has nowhere to go except your variance, and you will spend weeks hunting a loss that was a reporting gap.

What is actually inside the gap

Once you have a variance you trust, the question becomes what is in it. Four things, in roughly this order of size.

Over-portioning

The largest and least dramatic cause. A dish specified at 140g of protein going out at 165g is running 18% over on its most expensive component. It happens without intent: scales disappear during a rush, a new cook calibrates their hand against a busy pass, and the overrun becomes the house standard within a fortnight. No system catches it, because the stock left for a legitimate reason and the sale was rung correctly.

Waste that never got written down

Product that spoiled, burned or came back from a table and went straight into the bin. This is waste in disguise rather than true shrinkage, and the fix is a better log rather than tighter security. Anything you record moves from the unexplained column to the explained one, which is exactly what waste tracking is for.

Receiving and invoice errors

Charged for 20kg, delivered 18kg. Invoiced at the old price after a negotiated reduction. A credit note agreed on the phone that never appeared. These errors are rarely random in direction, and each one inflates purchases without inflating what arrived — landing squarely in the variance. Weighing deliveries on high-value items, rather than all items, catches most of it for very little effort.

Theft

Real, usually the smallest of the four, and almost always the first assumption. Investigate it last. Acting on a suspicion of theft when the cause turns out to be a 25g portion drift costs you trust across the whole team and recovers none of the money.

Narrowing from a number to a cause

A single site-wide shrinkage figure proves money is leaving but points at nothing. The way to make it actionable is to stop calculating one variance and start calculating several.

• Split by category. Proteins, produce, dairy, dry goods, beverage. Shrinkage concentrates — one category usually carries most of it, and knowing which one halves the search immediately.
• Then take the top ten items by usage value. These typically represent the bulk of your exposure, and a small percentage error on an expensive item is visible where the same error on flour is not.
• Count that short list weekly. Ten items counted weekly beats four hundred counted monthly, because a four-week-old signal has lost the context that would let anyone explain it.
• Weigh portions on the three highest-volume dishes. One week of spot checks will either confirm portioning as the cause or eliminate it, and both outcomes are useful.
• Check receiving on the worst category for a fortnight. Weigh in, check prices against the agreed list, chase every credit note.

Faster, smaller counts have a second benefit. They give you the data to order properly, which is where par levels and stock turnover start doing work. Over-ordering feeds spoilage, spoilage feeds unlogged waste, and unlogged waste feeds shrinkage. The chain runs in both directions.

When the number is lying to you

Two results should make you suspicious before they make you angry.

A variance above 3% of sales is more often a broken calculation than a broken control. The usual causes are stale recipe costs, a count and a sales report covering different dates, sub-recipes counted twice, or a popular new dish that was never entered into the recipe file — which removes its usage from the theoretical side entirely while the actual side keeps counting it. Check all four before escalating anything.

A negative variance, where actual usage came in below theoretical, is not a saving. It normally means the closing count was overvalued, an invoice has not landed yet, or your recipe yields are more pessimistic than reality. Stock does not multiply on a shelf. Treat it as a data error and go looking for the cause with the same energy you would apply to a loss.

The habit worth building is comparing periods rather than hitting a target. A shrinkage figure of 1.1% that has held steady for six months is a known, quantified cost you can decide whether to attack. The same figure after three months at 0.4% is a live problem that started somewhere specific, and the date it started is usually the strongest clue you will get. Consistency in how you count matters more than precision in what you count, because only consistency makes that comparison honest. The same principle runs through our wider inventory management guide.

Frequently asked questions

How much inventory shrinkage is acceptable?

Under 0.5% of food sales is tight and about as good as most sites can practically get. Between 0.5% and 1.5% is normal without formal portion controls. Anything above 3% should be treated as a measurement error until the recipe file, count dates and sales channels have all been verified.

Do I need a full stock count to calculate shrinkage?

For a site-wide figure, yes — actual usage depends on accurate opening and closing counts. But a partial count on your ten highest-value items gives you most of the useful signal at a fraction of the effort, and it can be run weekly rather than monthly.

Why does my shrinkage change so much month to month?

Usually counting inconsistency rather than real volatility. Different people counting, a count taken mid-service, or stock held in a second location that gets included some months and not others will all produce swings larger than the loss you are trying to measure.

Should staff meals be treated as shrinkage?

Track them separately. They are genuine consumption with no matching sale, so they belong in the variance, but they are authorised and predictable. Folding them into unaccounted shrinkage makes a growing staff-meal programme look like a control failure.

How long before fixing shrinkage shows up in food cost?

One to two inventory cycles, so typically four to eight weeks. Portioning fixes show fastest because they affect every cover. Receiving improvements take a full supplier cycle to appear. If nothing moves after two cycles, the shrinkage you thought you fixed was probably a calculation artefact.

Is beverage shrinkage measured the same way?

The arithmetic is identical, but drinks need their own variance rather than being blended into food. Free pour, spillage and comped drinks behave nothing like kitchen portioning, and mixing them together hides both problems.

What if I have no recipe file at all?

Start with your twenty highest-volume dishes rather than the whole menu. Those twenty will cover most of your usage, and a theoretical figure built on partial but accurate recipes is far more useful than one built on a complete but outdated file.

Run the numbers

Use the free Inventory Shrinkage Calculator to apply everything above to your own figures.