Your waste bin is a forecasting report

Every service produces two outputs. One goes to the customer and gets recorded in the POS. The other goes in the bin and gets recorded nowhere.

The second one is arguably more informative. Sales tell you what people bought from what was available. Waste tells you where your prep didn’t match demand — which is the gap margin disappears into.

Three kinds of waste, three different fixes

Lumping it together as “food cost” hides the causes. Separate it:

Prep waste. Prepped more than sold. A forecasting problem — you predicted a busier night, or a different mix.

Spoilage. Bought more than used before it turned. A purchasing and rotation problem, usually traceable to order quantities set once and never revisited.

Plate waste. Came back from the table uneaten. A portioning or recipe problem, and the one most likely to be telling you something about the dish rather than the operation.

Same cost line, three different interventions. A kitchen tackling spoilage when its real issue is over-prep will work hard and not move the number.

The cheapest measurement that works

You don’t need scales on every bin. For two weeks, at close, write down the five items that went in the waste, roughly how much, and which category.

Five lines a night. Ten minutes a week to review. Almost every operator who does this finds the same thing: a small number of items account for most of the waste, and they’re usually the ones with the shortest shelf life and the least predictable demand.

Once you know which five items they are, prep decisions stop being a general worry and become five specific questions.

Forecasting by weekday is the first real win

Most kitchens prep from a mental average of “a normal Tuesday”. But demand isn’t normally distributed around a weekly mean — it’s structured by day, by weather, by what’s on locally, by payday.

Pull eight weeks of sales by item and by weekday. You’ll typically find variation large enough that prepping to the weekly average guarantees waste on slow days and running out on busy ones. Prepping to the day’s pattern fixes both at once, with no new system.

Then layer in the things that move the number: weather, local events, school holidays. You don’t need a model. You need the previous four equivalent days in front of you when you decide prep.

Menu pricing without the guesswork

The common approach is a target food cost percentage across the menu. It’s simple and it systematically misprices.

Two dishes at the same food cost percentage can contribute very different amounts of cash. The one with a lower percentage but lower price can make less per plate than one with a worse ratio and a higher price. You bank cash, not percentages.

The more useful exercise, done quarterly:

1. For each dish: contribution per plate (price minus food cost), and units sold.
2. Plot them: high margin / high volume, high / low, low / high, low / low.
3. High margin, high volume — protect these; don’t change the recipe to save pennies.
4. High margin, low volume — a menu description and placement problem, not a food problem.
5. Low margin, high volume — the dangerous one. Busy and unprofitable. Reprice, re-engineer the portion, or accept it as a loss leader deliberately.
6. Low margin, low volume — cut it. It’s also carrying unique ingredients that become your spoilage.

That last point connects the two halves of this: menu complexity and waste are the same problem. Every dish with an ingredient nothing else uses is a spoilage risk with a menu line attached.

Start here

If you change one thing this month, record waste in five lines a night for two weeks. It costs nothing, it needs no software, and it will tell you which of the three problems you actually have.

Most operators discover they’ve been solving the wrong one.

---

Dinemetric delivers demand forecasting and menu optimisation for restaurants. More at dinemetric.com.