The two ways ordering by feel goes wrong (and why you only hear about one)

A supplier order is a decision made before you know exactly what the next service will bring. Your experience matters. You know the rhythm of the room, what the weather is doing, and whether a local event will change the mix. But memory is not a ledger. It can pull attention toward the last rush or the last time you ran short, while the quieter pattern of what sold gets harder to see.

There are two ways that can go wrong. If you order below demand, the kitchen may run out while guests still want the dish. If you order above demand, the surplus may expire or lose quality before it gets used. The first failure has a customer-facing sentence. The second often has no sentence at all; it is a smaller invoice margin, a discarded prep container, or stock carried into the next count.

The visible miss: not enough for service

A diner does not see your prep sheet or supplier order. They see the menu, place an order, and hear that the kitchen cannot make the item. In restaurant reviews, the phrases that show up include “they were out of it” and “hit or miss.” Those are real review phrases from diners, not testimony from StocksSense customers. They describe the experience, not its cause: a review cannot establish whether ordering, delivery, prep, or an unexpected rush was responsible.

That distinction matters. A stockout is not proof that the owner guessed badly. It is a signal to investigate. Check what sold, what was on hand, what arrived from the supplier, what was prepped, and whether the item was unavailable for another reason. If you only react by ordering more every time a guest misses out, you may fix one visible shortage by creating a quiet surplus elsewhere.

The quiet miss: too much for the week

Over-ordering has a different feedback loop. The food arrives, the invoice is paid, and the kitchen may use some of it. The remainder can be trimmed, spoiled, carried forward, or thrown away. Unless someone records the reason and quantity, the next person placing the order may never learn what happened. That does not mean every leftover is waste: some stock is useful protection against uncertain demand, and some carries safely into a later service. The point is to distinguish deliberate coverage from surplus that repeatedly goes unused.

This cost is easy to miss because it rarely announces itself. A guest might never know there was extra produce in the walk-in. The invoice records what you bought, but not automatically how much was discarded or why. If waste is not counted, the owner sees the purchase and may see the sales, but cannot reliably connect the gap to an ordering decision. That is a practical hypothesis about record keeping, not a claim about every restaurant’s books.

One demand pattern, two directions of error

Here is an illustrative example using a perishable ingredient ordered in cases. Assume the kitchen uses one case for every ten portions and each case costs an assumed thirty dollars. The weekly demand values below are invented solely for arithmetic. They add to eighty-four portions across four weeks, or twenty-one portions per week on average. A constant order of twenty portions a week would have been short in some weeks and over in others.

The table's arithmetic is direct: twenty plus twenty-four plus eighteen plus twenty-two equals eighty-four portions of example demand. Four weekly orders of twenty equal eighty portions ordered. In the example, week two is four portions short, week three has two portions beyond demand, and week four is two portions short. Across all four example weeks, eighty ordered minus eighty-four demanded equals a net shortage of four portions. The positive and negative weeks do not cancel in the kitchen: demand and supply have to meet in the right service, and a leftover from one week may not be usable for the next.

At the assumed conversion of ten portions per case, the example total demand is 8.4 case-equivalents and the example order is 8 case-equivalents. At the assumed thirty dollars per case, the purchase represented by eight cases is $240. If the four-portion shortage in the example could not be served, the purchase-cost equivalent is four-tenths of a case, or $12. This is not a measured lost sale: ingredient cost is not menu revenue, and the example says nothing about whether a guest chose another dish, returned later, or left.

Now imagine raising the example order to twenty-four portions every week to avoid the visible shortfall. Across four weeks that is ninety-six portions ordered against eighty-four portions of example demand, a twelve-portion surplus. At ten portions per assumed case, twelve portions are 1.2 case-equivalents. At the assumed thirty dollars per case, that surplus represents $36 of ingredient purchase cost if none of it is used. If it carries into later service or is used in another dish, it is not all waste. The arithmetic shows exposure, not a measured disposal bill.

Keep the two costs in view

A useful ordering review asks about both directions. For shortages, note the item, the service, whether it was unavailable, the quantity you had expected to need, and the cause you can verify. For surplus, note what remained, whether it was safely carried forward, what was discarded, and why. Separate a forecast miss from a delivery problem, a count error, a menu change, or an unusual event. Otherwise the next order is responding to a story rather than a record.

Use the same unit when comparing demand and stock. If your supplier sells by case but your recipe uses portions, write down the conversion and check it. In the example above, ten portions per case is an assumed conversion; if a case actually yields a different number after trimming or portioning, every case-equivalent changes. Check pack size, usable yield, shelf life, and whether the ingredient appears in more than one dish before treating the arithmetic as a purchasing instruction.

Then review the pattern over time instead of changing the order after one noisy service. A strong weekend, weather shift, event, or temporary menu feature may matter. So may a gradual change in demand or a recurring seasonal pattern. Write down what you knew when you placed the order, then compare it with what happened. That makes it possible to tell a reasonable decision under uncertainty from a repeatable miss.

What a forecast can and cannot do

StocksSense starts with sales history and produces a forecast and reorder quantity. Its statistical engine smooths level and trend, can select no seasonal pattern or an additive or multiplicative seasonal pattern, and can damp projected trend. It selects model settings using walk-forward errors across several forecast horizons. Before fitting, it checks for unusual monthly values and can damp flagged outliers. These steps describe the code; they do not mean the result knows tomorrow's covers or guarantees that a dish will stay available.

The reorder quantity also includes safety stock. The current code calculates that buffer as average demand multiplied by the order-period length and a safety multiplier. The multiplier can use cached, per-product reliability results when available; with no backtest result, the code uses a volatility tier from the trailing history. Reliability also gates which forecast horizon is treated as reliable. Those are decision aids, not a supplier order sent on your behalf, and the buffer is not a promise that stockouts or waste will disappear.

A sales CSV export works from any POS; catalog sync with a supported POS is available when enabled for your account. Once the history is in, check whether the forecast period matches your buying cadence and whether the item units line up. The order guide is a starting point for your judgment. You still account for supplier delays, storage limits, substitutions, and what the kitchen knows about the coming week.

For a fuller view of the input, adjustments, and reorder output, the How it works page walks through it end to end. For the arithmetic of turning one ingredient’s history into an order, the worked example linked below does it by hand.

The point is to see both misses

A shortage can leave a public sentence: “they were out of it.” A surplus can leave only a lighter bin. Neither signal tells the whole story by itself, and neither proves that ordering by feel caused the problem. Track what was unavailable and what went unused, check the reason, and compare both against the sales and stock you actually recorded. When you can see the two sides together, you can make the next order with a clearer account of what you are protecting against.