A trial batch at a supplier: when the trial pays off, what a mistake without it costs and at what probability of defects it breaks even
From the size of the full order, the unit price, the prepayment, the size and markup of the trial batch, the trial costs, the delay, the probability of detecting a problem and a table of supplier classes (probability of failure and the loss share) it counts what the trial costs, what losses it prevents and at what probability of defects it pays off. It suits China and any other country: take the full unit cost with logistics and duty from the “Full cost of an imported garment” calculation.
At a glance
- What question does it answer?
- Whether the trial pays off: what it costs, what losses it prevents for your supplier class and at what probability of defects it breaks even.
- What data is needed?
- The full order, the full cost of a unit in the warehouse, the prepayment.
- How to read the result?
- Example on the default values (an order of 1,200 units at RUB 1,450, prepayment 50%, a trial of 150 units with a 15% markup): the full order RUB 1.74 million, the prepayment at risk RUB 870,000; the trial costs RUB 183,225 (10.5% of the order). The trial pays off in 1 of 3 classes, the largest gain RUB 22,313, for a checked supplier it is −RUB 174,661. Example on the default values: Cost of the full order: RUB 1.74 million; Prepayment at risk: RUB 870,000; Purchase of the trial batch with the markup: RUB 250,125; Markup for a small batch: RUB 32,625. The figures are illustrative: enter your own to see your result.
Why and how to reason
A trial batch costs money, and the question is whether it is cheaper than a mistake. If the supplier turns out bad you will lose part of the full order: defects, missed terms, a lost prepayment. The expected loss without a trial equals the probability of failure × the loss share × the cost of the order. With a trial you pay for it: a markup for a small batch, extras for samples and inspection and the weeks of delay in which part of the sales is lost. In return the trial detects the problem with some probability, and then the full order is not placed. The trial pays off when the prevented losses exceed its costs. On the default values the full order is RUB 1.74 million, the prepayment at risk RUB 870,000, the trial costs RUB 183,225 (10.5% of the order). For an unknown supplier with a 30% probability of failure the gain is RUB 22,313, for a checked one with 3% the trial loses money: −RUB 174,661.
When to use it
Apply it before ordering from a new supplier, especially abroad, when the prepayment is large and the factory cannot be checked in person. It suits China and any country: take the full cost of a unit from the “Full cost of an imported garment” calculation. It is not needed if you have checked the supplier in person and the batch is small. The supplier classes and the probabilities of failure are your own estimates from the checking checklist: there is no norm. For footwear, accessories and children’s the loss share is higher because of sizes and certification, and the trial takes longer.
How to read the result
Look at the conclusion for each class and at the break-even probability of failure: it is the border above which the trial pays off. For the riskiest class it is 26.7%: if you estimate the risk higher, take the trial. The trial pays off in 1 of three classes. Where the trial is not worth it, do not give up protection altogether: place the order with sampling inspection, staged payment and a right to refuse. The trial costs, 10.5% of the order, consist mostly of the weeks of delay: shorten the delay (parallel checking of samples, partial shipment) and the trial becomes more worthwhile.
Common mistakes
They count the trial as free and forget the markup and the weeks of delay. They believe the trial as a guarantee, while quality drops on the run. They set one probability of failure for all suppliers. They forget the prepayment at risk. They take too large a trial, and it becomes a risky order itself. They do not set in advance what counts as the success of the trial and then argue about the result.
What data is needed?
- The full order, the full cost of a unit in the warehouse, the prepayment.
- The size and markup of the trial batch, the extras, the probability that the trial shows the problem.
- Supplier classes: the probability of failure and the loss share from your check; the price of the weeks of delay.
Inputs
| Field | Unit | Value in the example | Meaning |
|---|---|---|---|
| The full order | units | 1,200 | How many units you will order if the supplier fits: the main batch of the season. The example value is illustrative. |
| Full cost of a unit in the warehouse | ₽ | 1,450 | The price of a unit in your warehouse without VAT together with delivery, duty and marking (see “Full cost of an imported garment”). The example value is illustrative. |
| Prepayment to the supplier | % | 50 | What part of the order you pay before shipment: this money is at risk until the goods arrive and are checked. The example value is illustrative. |
| The trial batch | units | 150 | How many units you take for the trial: enough to check quality, sizes and terms and little enough not to risk. The example value is illustrative. |
| Markup for a small batch | % | 15 | By how much the unit price in the trial is higher than in the full order: a small batch costs more. The example value is illustrative. |
| Extra costs of the trial | ₽ | 45,000 | Samples, an independent inspection, the minimum delivery fee of a small batch: what will not come back with the sale. The example value is illustrative. |
| Probability that the trial shows the problem | % | 80 | If the supplier is bad, how often it shows in the trial batch: defects and terms do not always appear. The example value is illustrative. |
| Delay of the full order because of the trial | weeks | 4 | By how many weeks the trial postpones the main batch: making, delivery and checking the trial. The example value is illustrative. |
| Sales a week with the full order | units | 40 | How many units a week sell when the goods are in stock: the weeks of delay lose part of these sales. The example value is illustrative. |
| Contribution per unit | ₽ | 2,200 | How much you earn per unit after all costs (see “Marketplace unit economics”). The example value is illustrative. |
| What part of the postponed sales is lost for good | % | 30 | Demand does not always wait: some buyers will go to others. The rest will sell later. The example value is illustrative. |
How it is calculated
| Step | Formula |
|---|---|
| The full order | units × the full cost of a unit |
| Losses without the trial | probability of failure × loss share × the full order |
| Trial costs | markup for a small batch + extras + weeks of delay × sales a week × contribution × lost share |
| Losses with the trial | probability × share × (trial items + the undetected part of the full order) |
| The gain and the threshold | losses without the trial − losses with the trial − costs; the probability threshold = costs ÷ (share × (order × detection − trial)) |
Worked numbers
Default values from the app; in the interactive version you enter your own.
- Cost of the full order
- 1.74 m ₽
- Prepayment at risk
- 870,000 ₽
- Purchase of the trial batch with the markup
- 250,125 ₽
- Markup for a small batch
- 32,625 ₽
- Margin lost because of the delay
- 105,600 ₽
- Trial costs that will not come back
- 183,225 ₽
- Trial costs to the full order
- 10.5 %
- Classes where the trial pays off
- 1
- The largest gain of the trial
- 22,313 ₽
- The smallest gain of the trial (minus: the trial is not worth it)
- −174,661 ₽
- Break-even probability of failure for the riskiest class
- 26.7 %
- The class with the largest gain of the trial
- Unknown: no history, a price well below the market
- Without the trial
- With the trial: losses and costs
Data as a table
| Measure | Without the trial | With the trial: losses and costs |
|---|---|---|
| Checked in person: worked with, trial and sample passed | 13.1k | 188k |
| By recommendation: documents, reviews and a sample | 83.5k | 212k |
| Unknown: no history, a price well below the market | 313k | 291k |
Data as a table
| Measure | Gain |
|---|---|
| Checked in person: worked with, trial and sample passed | −175k |
| By recommendation: documents, reviews and a sample | −128k |
| Unknown: no history, a price well below the market | 22.3k |
The trial by supplier class
| Class | Probability of failure | Losses without the trial | Losses with the trial | Gain of the trial | Break-even probability | Conclusion |
|---|---|---|---|---|---|---|
| Checked in person: worked with, trial and sample passed | 3 % | 13,050 ₽ | 187,711 ₽ | −174,661 ₽ | 64.2 % | order at once with inspection |
| By recommendation: documents, reviews and a sample | 12 % | 83,520 ₽ | 211,935 ₽ | −128,415 ₽ | 40.1 % | order at once with inspection |
| Unknown: no history, a price well below the market | 30 % | 313,200 ₽ | 290,888 ₽ | 22,313 ₽ | 26.7 % | the trial pays off |
A trial is not insurance against everything: it finds defects and missed terms but does not guarantee that the full order will repeat the quality of the trial batch (factories economize on the run). So for the full order allow for sampling inspection and staged payment. The trial items themselves sell, so their cost is not counted as a cost apart from the markup. For footwear, accessories and children’s the loss share is higher (sizes, certification), the supplier classes are their own.
Nuances and limits
- Counting the trial as free: the markup, the extras and the delay cost money.
- Believing the trial as a guarantee: quality may drop on the run, inspection is needed.
- Setting one probability of failure for all suppliers: a checked one and an unknown one differ many times over.
- Forgetting the prepayment: it is at risk until the goods are accepted.