How much to order under uncertain demand
Compares order sizes by expected profit: ordering less than demand loses sales, ordering more leaves stock that clears at a lower price.
At a glance
- What question does it answer?
- Which order size gives the highest expected profit if demand may vary.
- What data is needed?
- Price, cost and the clearance price of leftovers.
- How to read the result?
- The best order is chosen by expected profit among demand scenarios and your plan. Look at the chance of leftovers, expected leftover and lost sales: choosing an order balances them. Example on the default values: Price RUB 4,900, cost RUB 2,200, clearance price RUB 1,200 (below cost) and three demand scenarios. Your plan of 3,500 units gives an expected profit of RUB 7.14m and the best order by expected profit is 3,000 units (RUB 7.18m). The probability of leftovers is 75%, expected leftovers are 625 units and lost sales 125 units. If the clearance price is above cost, leftovers do not lose money and the model always advises ordering more: check the leftover price and the delay of cash. Next: “Unit price by run tiers and the minimum run” and “First order and reserve for reorders”.
Why and how to reason
A seasonal item cannot be reordered: order too little and you lose sales, order too much and the leftover goes at a reduced price. This is the classic “newsvendor” problem. Demand is unknown, so scenarios with probabilities are taken, and for each order size the expected profit is computed.
In each scenario: sold (the smaller of order and demand) x price + leftover x clearance price - order x cost. Expected profit is the average over scenarios weighted by probability.
The optimum follows a simple rule: order while the probability of selling one more unit is above the share of overage loss, that is until it falls below 1 - 73% = 27%. Critical ratio = underage / (underage + overage) = 2,700 / (2,700 + 1,000), about 73%: this is the required cumulative probability, and you take the smallest order at which the probability that demand does not exceed the order is at least 73%. In the example it is 25% at an order of 2,000 and already 75% at 3,000, so 3,000 is the optimum; the chance of selling a unit beyond 3,000 is only 25%, below the 27% threshold.
Example. Scenarios: 2,000, 3,000, 4,000 with probabilities 25 / 50 / 25%. At an order of 3,000: 4.4 + 8.1 + 8.1, weighted RUB 7.175 m. At your plan of 3,500 expected profit is RUB 7.14 m.
When to use it
It is useful for seasonal styles when there is no repeat order or the lead time is too long. Suitable for the first buying decision. Not suitable for items with steady demand: use the reorder point there. Assumptions: scenarios and probabilities are your estimates, the clearance price is known. The better demand is estimated, the more reliable the result; widen the scenarios if in doubt.
How to read the result
Compare the expected profit of your plan with the best order. The difference is small (7.14 against 7.18 m), so the plan is near the optimum; if large, re-check the order. The critical ratio shows how much risk you should take: the higher the margin and the lower the loss on leftover, the bolder the justified order. The probability of leftover (75%) and expected lost sales show your risk on both sides. Decision: choose an order considering your attitude to risk, not only the expected value. If in doubt, narrow the scenario spread and see how the best order changes.
Common mistakes
Typical trap: entering probabilities that do not sum to 100%; taking only the optimistic scenario; not noticing that the clearance price is below cost; ignoring that stock-outs lose future customers too. Also: confusing the leftover price with the in-season discount price. Also: taking expected demand as the only one.
What data is needed?
- Price, cost and the clearance price of leftovers.
- Demand scenarios (weak, base, strong) and the probability of each.
- Where to take it: selling price and cost from the price list and costing; the clearance price of leftovers from the results of past clearances (not from hope); demand scenarios from sales of similar seasons, the marketing plan and buyers’ views, with probabilities adding up to 100%. Scenarios first, then your own order plan for comparison.
Inputs
| Field | Unit | Value in the example | Meaning |
|---|---|---|---|
| Selling price (excl. VAT) | ₽ | 4,900 | Selling price per unit excl. VAT in a normal sale. |
| Unit cost | ₽ | 2,200 | Full unit cost excl. VAT. |
| Clearance price for leftovers | ₽ | 1,200 | What the leftover really sells for at the end of the season. If it is above cost, leftover loses nothing and ordering more always wins: check that the exit price is realistic. The example value is illustrative. |
| Your planned order | units | 3,500 | How many units you plan to order; the calculation compares this plan with other order sizes. |
How it is calculated
| Step | Formula |
|---|---|
| Profit in a scenario | Sold × price + leftover × clearance price − order × cost |
| Expected profit | Sum of (probability × scenario profit) |
Worked numbers
Default values from the app; in the interactive version you enter your own.
- Expected profit at your plan
- 7.14 m ₽
- Best order by expected profit
- 3,000 unitsA full scan of order sizes from zero to the largest scenario demand (plus your plan).
- Expected profit of the best order
- 7.18 m ₽
- Critical ratio
- 73 %(Price − cost) ÷ (Price − clearance price). It pays to order so that the chance demand does not exceed the order is at least this level.
- Chance of leftovers
- 75 %
- Expected leftover
- 625 units
- Expected lost sales
- 125 units
Expected profit by order size
| Order, units | Expected profit |
|---|---|
| 0 | 0 ₽ |
| 2,000 | 5.4 m ₽ |
| 3,000 | 7.18 m ₽ |
| 3,500 | 7.14 m ₽ |
| 4,000 | 7.1 m ₽ |
By expected profit the best order is 3,000 units (7.18 m ₽). At your plan of 3,500 units the expected leftover is 625 units and lost sales 125 units.
Nuances and limits
- Estimate probabilities from sales history; they should total 100%, otherwise they are scaled.
- Probabilities entered by eye give a pretty but unreliable figure: rely on past seasons and revisit after the first weeks.
- Money in a larger order costs interest and warehouse space: compare the gain with the cost of frozen cash.