Calculate the two outcomes before choosing a return rate
Start with a fully kept order and include its VAT, platform, creator, product, fulfilment and advertising costs. Then calculate the full-refund outcome separately, using only the fee reversals that apply and the inventory value actually recoverable.
The expected-value model here assumes full refunds and retained outbound fulfilment and advertising. Partial refunds, exchanges, returnless refunds and different fee reversal terms need separate cases if they materially affect the store. A single return percentage cannot encode all those outcomes.
Expected contribution per original order = (1 − full-return rate) × kept-order contribution − full-return rate × full-return lossIllustrative cohort with €8.60 kept-order contribution
Assume a kept order contributes €8.60. A returned order loses €20.60 after recovering 80% of an €18 product cost and retaining €6 outbound and packaging cost, €6 ads and €5 reverse-processing cost. Assume full relevant platform and creator commission reversal.
At a 10% full-return rate, 100 original orders contain 90 kept outcomes and ten returned outcomes. Contribution is 90 × €8.60 − 10 × €20.60 = €568, or €5.68 per original order. These are illustrative outcomes, not observed return behaviour.
| Full-return rate | Contribution / original order | Margin / original gross sales |
|---|---|---|
| 0% | €8.60 | 14.33% |
| 5% | €7.14 | 11.90% |
| 10% | €5.68 | 9.47% |
| 20% | €2.76 | 4.60% |
Find the break-even return-rate threshold
Set expected contribution to zero and solve the equation. With kept profit K and positive return loss L, the break-even return probability is K / (K + L). In this example, €8.60 / (€8.60 + €20.60) is about 29.45%.
That threshold leaves no contribution for unallocated overhead and is not an acceptable operational target. A required profit margin produces a lower permitted return rate. The threshold also changes when acquisition cost, fee reversal or stock recovery changes.
Full-return rate at zero contribution = kept-order contribution / (kept-order contribution + full-return loss)Measure the rate and stock recovery from compatible cohorts
Use a cohort old enough for the represented after-sales events to emerge, and keep the denominator stable. An order-return rate, unit-return rate and refunded-sales share can all differ. Compare like with like before updating the model.
Track stock disposition separately from the refund itself. A physically returned item is not necessarily ready to resell at full cost. Record inspection, repair, repacking or write-off, and avoid using a temporary warehouse receipt as proof of full recovery.
- Choose a product and destination cohort with sufficient observation time.
- Define the original-order denominator and full-refund outcome.
- Match refunds and commission reversals.
- Record returned-stock disposition and handling cost.
- Update both return probability and loss severity.
- Recalculate contribution and the required margin threshold.
Should the team reduce return frequency or return severity first?
The model can compare a lower probability with a lower loss per return. Better recovery or handling may improve contribution even when the return rate is unchanged, while preventing a return saves its entire expected outcome difference. The achievable effect requires a measured operational test.
Keep interventions and product mix visible. A shift toward less expensive products can lower refund value without reducing the number of returned orders, so one headline return metric can conceal what actually improved.