A café's order quantity starts with expected ingredient use, the next delivery opportunity and usable stock. Weekly sales help estimate use, but a weekly ordering schedule does not automatically mean a seven-day planning horizon: supplier lead time also matters.
This guide works through an illustrative milk order. The numbers are planning assumptions, not a food-safety shelf life or a guaranteed service outcome.
Convert drinks into ingredient units
Use the actual recipe quantity, then reconcile estimates with counts and adjustments. For this example, all ounces are US fluid ounces.
Drink
Servings
Milk per serving
Estimated milk use
Latte
245
8 fl oz
1,960 fl oz
Cappuccino
110
5 fl oz
550 fl oz
Hot chocolate
65
8 fl oz
520 fl oz
Steamed milk
30
6 fl oz
180 fl oz
Read before ordering
A dense operator briefing for teams that need sharper buying, cleaner supplier follow-up, and fewer expensive surprises.
Total
3,210 fl oz
At 32 fluid ounces per quart, that is 100.3125 quarts for the week, or about 14.3 quarts per day. That is theoretical recipe usage. Spills, steaming waste, staff drinks, returns and recipe changes can make physical usage different.
Choose the review and protection periods
Assume you review orders every seven days and delivery takes two days. For an order-up-to policy reviewed on order day, the protection period is seven plus two = nine days. The order must cover the time until the following reviewed order can arrive.
A target defined on delivery day has a different reference point. Label it before comparing “PAR” numbers between a spreadsheet and software. For a simpler count sheet, see the restaurant order guide.
Add an explicit buffer
For this simplified example, assume independent daily demand with standard deviation of three quarts, fixed lead time and a normal approximation. A 95% cycle-service target uses approximately 1.65 as the z-score:
Round the planning target to 144 quarts before applying the supplier's pack constraints. A 95% cycle-service target is a modeling target for a cycle without a stockout under these assumptions. It is not a guarantee of exactly one stockout every twenty weeks and is not the same as 95% unit fill rate.
If lead time is variable or day-of-week demand is strongly seasonal, use a model that reflects that. Do not treat the formula as universally calibrated because it contains a z-score.
Subtract the relevant stock position
Suppose the café has 30 usable quarts, 24 quarts already on a confirmed incoming order that will arrive within the relevant window, and six quarts reserved for a separately planned event.
At twelve quarts per case, that is eight cases. Confirm that the incoming order is not being counted twice and that it arrives before the demand it is intended to cover.
Also check the first two days before the new delivery. An order-up-to calculation can look sufficient overall while stock runs short before any receipt arrives. Use a dated stock projection where timing matters.
Perishability can change the ordering schedule
Do not assign milk a universal daily decay percentage or treat a statistical spoilage allowance as a safe-use rule. Actual product labels, storage, receiving condition and applicable operating procedures determine usable stock.
If the calculated quantity cannot be used while acceptable, reduce the batch or arrange more frequent deliveries rather than simply adding more milk to offset modeled losses. Keep waste records so the buying policy learns from the actual operation.
Keep a reorder point separate
A continuous-review reorder point uses demand during lead time plus a buffer. With the same simplified assumptions and a two-day lead time:
ROP = 14.3 × 2 + 1.65 × 3 × √2 ≈ 35.6 quarts
That is about 36 quarts of inventory position under the model. On a weekly schedule, reaching it between order days is a signal to investigate coverage and the next feasible delivery, not proof that an emergency order can arrive in time.
Put the calculation into a repeatable workflow
Validate the top ingredients, recipe quantities, supplier packs, counts and delivery dates first. Compare the recommendation with actual usage over several representative cycles. Review promotions, closures, special events and changes in supplier lead time explicitly.
LineNow's café ordering workflow connects the buying decision to supplier cases, changes, receiving and a later stock count. The PAR calculator helps explore inputs; your actual settings and data determine the recommendation.
