Order return rate measures affected orders; item return rate measures affected units. They answer different questions and can diverge for multi-item baskets.
Define the numerator and denominator
Order return rate equals returned orders divided by fulfilled orders for the same cohort. Item return rate equals returned units divided by fulfilled units. Revenue return rate compares returned merchandise value with eligible sales value.
Never label one metric simply 'return rate' in a shared report. State the unit, cohort date and event status. A returned order with three items counts once in the order metric but may contribute one, two or three units to the item metric.
Match returns to the original cohort
Dividing returns received this month by orders shipped this month becomes distorted when return windows cross months. A cohort view assigns each return to the original fulfillment period.
Use a mature cohort after most of its return window has elapsed. If an early estimate is necessary, mark it provisional and compare it with cohorts at the same age.
- Exclude cancellations that never shipped.
- Define whether exchanges count.
- Apply one rule to partial multi-item returns every period.
Choose the metric for the decision
Order rate predicts support contacts, labels and cases. Item rate is better for product quality, fit and unit disposition. Revenue rate highlights expensive merchandise flowing back even when unit count is low.
Match the cost driver: use returned orders for case-level label and support cost, returned units for inspection or depreciation, or split the model into both components.
Worked multi-item example
Suppose 1,000 orders contain 1,400 units. Eighty orders generate returns involving 110 units. Order return rate is 80 ÷ 1,000 = 8.0%; item return rate is 110 ÷ 1,400 = 7.86%.
If the 80 cases cost $18 each in labels and case handling, that component is $1,440. Unit-level depreciation belongs across 110 units separately. This prevents multiplying an order-level average by an item-level count.
Put this guide into practice
- Name the metric and unit.
- Use matching cohorts.
- Keep cancellations separate.
- Match case and unit costs correctly.
- Mark immature cohorts provisional.
Sources and further reading
External sources provide regulatory or industry context. Refund or Return independently prepares the framework and examples above.
Prepared by the Refund or Return Editorial Team under our editorial policy. This educational guide is not legal, tax or accounting advice.
Last reviewed: August 21, 2026