Restaurant Inventory Variance: The Five Places Stock Leaks

Stock variance is not a mystery, it is an unanswered question. Product left the building without the sale that should have accompanied it, and the number on the report tells you how much but never where.

Finding where is the entire job. This guide traces variance back through the five points a product passes on its way from a supplier's van to a customer's plate, because the leak is always at one of them.

What is stock variance in a restaurant?

Variance is the difference between theoretical usage, calculated from your recipes and sales, and actual usage, calculated from your stock counts and purchases. When those two disagree, the gap is variance.

It gets reported as a percentage or a cash figure, and both are useful for tracking. Neither is useful for fixing anything, because a total tells you nothing about its own composition. A three point variance made entirely of one over-portioned protein and a three point variance spread evenly across four hundred lines are the same number and completely different problems.

Why is theft usually the wrong first assumption?

It is the first thing most operators say and rarely the largest contributor. Theft is deliberate, which makes it relatively concentrated and relatively detectable. The five points below are all accidental, all continuous, and all invisible in the moment they happen.

Starting from theft also has a cost beyond wasted time. It puts the investigation on the team rather than the process, which is the fastest way to lose the cooperation you need to actually find the leak.

The Five Point Trace

Trace the product forward along the path it physically travels. Each point has its own signature in the data, so you can usually narrow the field before you set foot in the building.

1. Point one, delivery. Product you paid for that never arrived, or arrived as something else.

2. Point two, storage. Product that arrived and spoiled before it could be sold.

3. Point three, preparation. Product consumed by yield and trim that no recipe accounts for.

4. Point four, service. Product plated beyond the specified portion.

5. Point five, transfer. Product moved between sites, or to staff and comps, without a record.

Work them in order. The earlier points are cheaper to check and, in most sites, more likely to be responsible than the later ones.

Point one: what goes wrong at delivery?

Deliveries arrive during prep, which is the worst possible moment for anyone to check them properly. Someone signs, the boxes go into the walk-in, and the reconciliation against the order happens later or never.

Short deliveries are rarely deliberate. Substitutions, split cases, an out-of-stock line quietly dropped. All produce the same result: you paid for product you did not receive, and by invoice stage nobody can recall what came through the door.

The signature in the data is variance concentrated in specific supplier lines rather than spread across a category. If one supplier's products account for most of the gap, start at the delivery door.

Checking against the order at the point of delivery takes minutes and is the highest-return habit in the back of house. Counting straight into a phone rather than onto paper removes the second chance to make an error, which is what the free stocktake app exists to do.

Point two: how much is lost in storage?

Product that arrives correctly and then spoils is a pure loss. It was paid for, counted in, and never generated a sale.

According to WRAP, around 920,000 tonnes of food is wasted at UK hospitality and food service outlets each year, and roughly 75% of it was avoidable and could have been eaten. Avoidable is the operative word. Spoilage is not a fixed cost of doing business, it is a rotation and ordering problem.

The signature is variance concentrated in short-life products, and it moves with the trading calendar. A quiet week after a heavy order is the classic cause.

Point three: what does preparation actually consume?

Preparation is where stock legitimately becomes something smaller. A case of produce becomes trimmed product and a bin of offcuts. That is not loss, it is the job.

It becomes variance when the yield is not in the recipe. A dish costed against raw purchase weight rather than plated portion will understate its consumption on every single cover, permanently and invisibly.

Illustrative example, not an attributed client figure. If a braise loses roughly a third of its raw weight in cooking and the recipe was costed on the raw figure, every portion sold widens the gap. Nobody in the kitchen has done anything wrong. The recipe is simply describing a dish that does not exist.

The signature is a variance that is stable, proportional to covers, and confined to specific dishes. Stable variance is almost always a recipe problem rather than a behaviour problem.

Point four: how does portion drift happen?

Portion drift is the point nobody wants to raise, because raising it sounds like an accusation.

It is not really about carelessness. A chef plating the same dish at seven o'clock and at ten o'clock on a busy Friday is working at a different speed with a different queue behind them. Without a scale on the pass or a spec that anyone consults, portions move, and they move upward far more often than down.

The signature is variance that correlates with trading volume and with specific shifts. If your gap widens on your busiest days and narrows on quiet ones, you are looking at service, not storage.

Point five: what leaves without a record?

Anything that exits the kitchen without a costed sale attached to it creates variance that has nothing to do with the kitchen. Staff meals, comps, tastings, recorded waste that never got recorded.

For multi-site groups, transfers are the largest version of this. Illustrative example: a keg goes across for an event, a case is borrowed on Thursday and returned on Monday, or not returned at all. Off-record transfers are why group operators end up comparing venues that were never counting the same things. One site looks like it is haemorrhaging stock and the other looks unusually efficient, and neither figure means anything.

Our guide to reducing inventory variance across multi-location restaurants covers the group-level version of this problem in more depth.

How do you tell which point is responsible?

Each of these checks costs under an hour. Running all five costs a day and will usually locate the majority of the gap, which is a better use of management time than another month of reporting the total.

What should you do once you find it?

Work the largest cash contributor first, not the largest percentage. A two percent gap on your highest-volume protein is worth more than a twenty percent gap on a garnish, and fixing the second one feels productive while changing nothing.

Then re-measure. A fix that is not followed by a measurable movement in the next period's variance was not a fix, it was a theory.

Operators using StockTake Online typically identify up to 3 to 8% in recoverable food cost within the first 60 days of going live. The range reflects the starting point: a site already counting weekly with current recipes will find less than one running on memory.

Holding all five points in view at once, across every site, is what separates a restaurant stock control system from a better spreadsheet. The spreadsheet can hold the count. It cannot tell you which of the five points the gap came from.

If you want to run the Five Point Trace against your own numbers rather than a worked example, book a demo and bring one closed period with you.

Key takeaways

• Variance tells you how much stock leaked. It never tells you where, and where is the entire job.

• Trace the product forward through five points: delivery, storage, preparation, service, transfer.

• Theft is the common first assumption and rarely the largest contributor.

• Each point leaves a distinct signature in the data, so you can narrow the field before walking the building.

• Stable variance is usually a recipe problem. Variance that moves with trade is usually a service problem.

• Work the largest cash contributor first, then re-measure to confirm the fix was real.

Frequently asked questions

What causes stock variance in a restaurant? Variance leaks at five points in the product journey: deliveries that were short or substituted, product spoiling in storage, yield and trim that no recipe accounts for, portions drifting above spec during service, and product transferred or given away without a record. Theft is possible but is rarely the largest contributor.

What is an acceptable level of stock variance? There is no universal figure and any number presented as a standard should be treated with caution. As a rule of thumb rather than a sourced benchmark, the useful test is whether your variance is stable and explainable rather than whether it matches someone else's target. A gap you can account for is healthier than a smaller one you cannot.

How do I find out where my stock variance is coming from? Read the signature in the data before you walk the building. Variance concentrated in one supplier's lines points to delivery. Variance that is stable and dish-specific points to recipes. Variance that widens on busy shifts points to portioning. Each first check costs under an hour.

Does stock variance mean someone is stealing? Usually not. The five leak points are all accidental and continuous, which is exactly why they persist and why they add up. Starting from theft puts the investigation on the team rather than the process and tends to cost you the cooperation you need to find the real cause.

How often should I review stock variance? Weekly for most sites. Monthly variance is accurate enough for the accounts and far too slow to change anything, because the trading period that caused the gap has already closed by the time you see it.