Key takeaways
- Pick one denominator. Units and orders answer different questions; mixing them creates a number nobody can compare.
- Use the same period. Returns arrive after shipping, so a mismatch between numerator and denominator distorts the rate.
- Exchanges are not returns. Track exchanges separately, or state clearly that they are included.
- Segment before judging. Category, channel, price band, fit risk, and customer type can matter more than policy.
- Return rate is not return cost. A lower rate can still cost more per return if the remaining returns are expensive to process.
The formula
The basic formula is straightforward:
Return rate = returned units ÷ shipped units × 100
For an order-level view, replace units with orders:
Order return rate = returned orders ÷ shipped orders × 100
The denominator is where most comparisons go wrong. Shipped units include orders that may still be in transit; delivered units exclude them. If returns are counted when the customer requests them, pair that numerator with shipped or delivered units from the same cohort, not with this month’s shipments. A dashboard that divides this month’s returns by this month’s shipments is a ratio of two different cohorts.
A worked example
A store ships 4,850 units in a month and receives 291 returned units. The unit return rate is 291 ÷ 4,850 × 100 = 6.0%. If 4,700 of those units were delivered and the store prefers a delivered denominator, the rate is 291 ÷ 4,700 × 100 = 6.2%. The difference is small here, but it grows when shipping times are long or a large batch is still in transit.
| Measure | Numerator | Denominator | Result |
|---|---|---|---|
| Unit return rate | 291 returned units | 4,850 shipped units | 6.0% |
| Unit return rate, delivered | 291 returned units | 4,700 delivered units | 6.2% |
| Order return rate | 234 returned orders | 3,900 shipped orders | 6.0% |
None of these is “the” return rate. Each answers a different question, and the store should publish the definition next to the number. A 6% rate is not automatically good or bad: a fashion retailer with free returns and a narrow size run may see much higher rates, while a consumable replenishment product may see low single digits.
What actually moves the number
Return reasons are more useful than the headline rate. Split returns into fit, damage, late delivery, description mismatch, changed mind, and wrong item. Fit and description problems point to sizing charts, photography, and product copy. Damage and late delivery point to packaging and carrier performance. “Changed mind” may point to price, impulse purchase, or a generous policy.
Then segment by channel and customer type. Marketplace returns may be recorded differently from direct-to-consumer returns; first-time buyers often return more than repeat buyers; promotional periods can shift both volume and mix. A single blended rate can hide a worsening cohort behind a stable average.
Limits
- It is not a cost. A 5% return rate on low-value items can cost less than a 3% rate on high-value, fragile items.
- It ignores resale value. A returned item that is resold as new is not the same as one that is scrapped.
- It ignores exchanges. If customers exchange rather than refund, revenue may be retained while the return rate still rises.
- It depends on policy. A longer window or free returns can increase the measured rate without changing product quality.
- It is lagged. Returns arrive weeks after shipment, so the most recent period is usually incomplete.
Pair the rate with shipping cost per order and contribution margin to see whether a change in returns actually improves the economics. The AOV calculator is useful too, because a high return rate on large orders can erase more contribution than a higher rate on small ones.
Use the calculator
Run your own numbers with the return rate calculator. Enter returned and shipped units or orders, and use the result as a starting point for the cost and cohort questions above, not as a grade.
Frequently asked questions
Is return rate the same as refund rate?
No. A return can be exchanged, restocked, or refused; a refund is a payment reversal. Refund rate usually has a smaller numerator and a different cost profile.
Should exchanges count as returns?
Only if you define them that way. Most operators track returns and exchanges separately because an exchange often keeps the revenue and changes the fulfilment cost instead.
What is a good return rate?
There is no universal benchmark. Compare like categories, channels, price bands, and policies over the same season. A rate that is stable and well understood is more useful than a rate that looks low because the denominator is wrong.
Further reading