The plain-language guide
How to forecast demand for a small food business.
You do not need a spreadsheet to predict what to order next week. Four numbers and a short method get you there.
The short answer
Four numbers matter: your average daily sales, your busiest day, how long your supplier takes, and how much you can afford to overstock. Track those and you get close. Your best seller stays in stock. You stop overbuying.
What "close enough" looks like
Take a burger stall that sells 40 burgers on a normal day. On Friday that jumps to 90. The patty supplier needs two days' notice to deliver more. Order for a 40-burger day, and the stall runs out of its best seller by 6pm. Order 90 every day, and it is throwing out unsold buns by Sunday. Nobody hits the exact number Friday will bring. But the gap between running out and throwing food away is wide. Aim for the middle of it.
The 4 numbers that matter
Four numbers from your own sales get you most of the way there. No algorithm needed, no guessing.
- Average daily sales: what you sell on a normal day, not a busy one. This is your floor.
- The busy day: the day your best seller runs out and the line gets long. This is the day you plan for.
- Supplier lead time: how many days pass from order to arrival. Miss this and nothing else matters.
- How much you can overstock: what spoils, what you have room for, what ties up your money. This sets your ceiling — and it's a different number from the buffer that protects you when a supplier runs late; that one gets its own guide.
Try it with your numbers
50
units of cushion between an average day and your busiest
Turning the numbers into an order
- Find your average day. Add up last week's sales and divide by the days you were open. That is your baseline.
- Add a cushion for the busy day. Look at your best day, and give yourself enough room to cover it. Use the real peak, not a guess.
- Order far enough ahead. Your supplier has a cutoff date. Count back from the day you need stock, and place your order before that date.
- Stop at your ceiling. Order too much and you pay in spoilage and tied-up cash. Order too little and you lose sales. Find the number between the two.
The one number a spreadsheet gets wrong
A spreadsheet takes a seven-day average and calls it done. But your week is not flat. You plan around the busy day, and you order around the supplier cutoff. A plain average hides both. That is how a good week ends with your top seller sold out on Saturday.
Where ForeFlux picks up
ForeFlux reads the same four numbers and does the arithmetic for you. It works per item and per branch. It names what to order and the deadline to order it by. It also tells you plainly when it is already too late to fix a shortage. The rule stays simple: we suggest, you decide. Every number stays overridable, and your call wins. This is a forecast, not a reorder point: see reorder point or forecast for what separates the two.
It asks for a little sales history first. A brand-new account says "not enough data yet" instead of printing a made-up figure. That honesty is the point.
Once your ordering runs on these four numbers, the next blind spot is stock you have ordered but not yet received. That one gets its own guide.
Questions owners ask
Past the basics.
What if I don't have much sales history yet?
Two or three weeks is enough for a rough average day and busy day. You can even do it by hand. ForeFlux says "not enough data yet" instead of guessing when an item's history is too thin.
Does this method work for a weekend-only or irregular-hours business?
Yes. Base your average and busy day only on the days you are open. A weekend pop-up forecasts off its own two-day week, not an average seven-day week. ForeFlux does this automatically per branch.
What if my supplier's lead time changes?
Recount your order-by date from the new lead time. Say your supplier used to take two days and now takes four: your cutoff moves two days earlier. Missing that shift is one of the most common causes of a stockout.
What evidence-based methods exist to forecast demand for a small restaurant?
Three methods have evidence behind them and fit on paper. First, a moving average of the last two to three weeks per item, split by weekday. That is what most inventory apps run underneath. Second, a par or reorder point built on that average plus measured lead time, the standard textbook method. Third, a newsvendor-style lean, ordering short or long depending on margin and spoilage cost. Bigger operators add seasonal models and machine learning, but those need months of clean history. Champions 12.3's study of 114 restaurants found the gains came from measuring waste and rethinking purchasing, not from complex models. Start with the four numbers on this page. ForeFlux runs the weekday-split average per item and per branch, and says "not enough data yet" when history is too thin.
Identify the core components of a demand forecasting model for restaurant inventory.
Five parts. A demand history per item, per day, ideally split by weekday. A baseline, usually a recent average, that says what a normal day sells. A shape on top of the baseline for busy days, payday weeks and holidays. Supplier lead time, so the forecast turns into an order-by date and not just a number. And a cushion rule that decides how far above the baseline to order, set by how much spoilage you can carry. Everything else, from seasonality to weather, is a refinement on those five. A model missing lead time is a sales report, not a forecast. ForeFlux is built on exactly these five and shows every one of them so you can override it.
See the next order before you need it
Know what you need, before you need it.
The demo needs no signup and no card. It is the real product running on demo data.
Open the order sheet