Business context
Why Refund Rate matters
A rising rate can point to a new quality or expectation issue. A falling rate is positive only if refund access and customer treatment have not worsened.
- Business question
- What share of orders resulted in a refund?
- Teams that use it
- Finance, e-commerce, operations, customer support, product, and fraud teams.
- Decisions it supports
- Product quality reviews, description changes, return policies, supplier discussions, and support interventions.
Calculation
Refund Rate formula
(Refunded Orders ÷ Total Orders) × 100
Formula components
- Refunded orders
- Distinct orders receiving a full or qualifying partial refund under the documented rule.
- Total orders
- Completed orders eligible to be refunded for the same population.
- Refund window
- The period after purchase during which a refund can still be recorded.
How to calculate Refund Rate
- Choose whether partial refunds count as refunded orders or are measured separately by value.
- Select an order cohort old enough to pass through the normal refund window.
- Count distinct refunded orders and total eligible orders.
- Divide refunded orders by total orders and multiply by 100.
Worked example
Refund Rate example
A retailer reviews 3,800 fulfilled orders after the full return window has passed. Seventy-six orders received a qualifying refund.
- Refunded orders = 76.
- Total eligible orders = 3,800.
- Refund Rate = 76 ÷ 3,800 × 100 = 2.0%.
Refund Rate is 2.0%.
Two of every 100 eligible orders resulted in a qualifying refund under this definition.
How to interpret the result
Break the rate down by product, supplier, reason, channel, region, and fulfilment method. A small category can create a large company-wide problem if its rate is unusually high.
Normal levels vary by product category, return policy, season, payment method, and whether partial refunds are included. Compare cohorts after an equal refund window.
Common mistakes and limitations
- Immature order cohorts
- Recent orders have not had equal time to be refunded.
- Counting refund transactions
- One order can create several refund records and should not be duplicated in an order-based rate.
- Mixing order and revenue rates
- Refunded-order percentage and refunded-value percentage can tell different stories.
- Ignoring reason codes
- The overall rate does not identify whether the cause is quality, fraud, delivery, or preference.