Returns rate
The share of what you sold that came back.
01 What it is
Why anyone looks at this number
The share of what you sold that came back.
A return is a sale reversed and a cost added, twice. It is also the cheapest quality signal you own, because the customer has already told you what was wrong — if anyone recorded it.
02 The formula
How it is worked out
Returns rate = units_returned ÷ units_sold grain : month × category × channel unit : percentage source : Returns processing, joined to the original order line and the product master
The reason code is the whole value of this metric. Free text, or a default code the warehouse clicks through, turns a diagnostic into a cost line. Join the return back to the product attributes and the pattern usually names itself.
03 Worked example
The same number, with real inputs
| Units sold | 208,000 |
| Units returned | 12,400 |
| Top reason: fit / sizing | 38% |
| Calculation | 12,400 ÷ 208,000 |
| Result | 6.0% |
Nearly two fifths of returns are sizing, concentrated in two ranges. That is a product-data fix upstream, not a logistics problem to absorb downstream.
04 What moves it
Four things that actually change this number
Product data quality
Sizing, images and specifications that set the wrong expectation.
Fulfilment accuracy
Wrong item, damaged in transit, late enough to be unwanted.
Returns policy
Generous windows raise the rate and often raise lifetime value with it.
Category mix
Apparel and electronics behave nothing like consumables.
05 Where the number lives
The system, the record and the fields
| System of record | Key record | Fields you need |
|---|---|---|
| Returns processing, joined to the original order line and the product master | Return joined to Order line, Product and Customer | return_id, order_line_id, reason_code, condition_grade, restock_flag, refund_amount, return_date |
The reason code is the whole value of this metric. Free text, or a default code the warehouse clicks through, turns a diagnostic into a cost line. Join the return back to the product attributes and the pattern usually names itself.
06 How it goes wrong
Three ways this metric misleads people
No usable reason code
The rate is known, the cause is not, and the same returns arrive next season.
Fix: Constrain the code list, and make it the customer's reason rather than the handler's.Counting refund value instead of units
Price changes and discounts move the metric with no change in behaviour.
Fix: Measure units for the operational read, value for the P&L one.Ignoring the cost of the return leg
Handling, grading, repackaging and markdown are absorbed invisibly.
Fix: Cost the whole return, including the recovery rate on resale.08 Questions
Frequently asked
Is a low return rate always good?
Not in categories where a generous policy drives the purchase. The pairing to watch is return rate against retention and lifetime value — tightening the policy usually improves one and quietly damages the other.
Where should returns data be joined?
To the order line and the product attributes. A return that cannot be traced to what was actually shipped can be counted but never diagnosed.
One definition, everywhere it is used
SCIKIQ stores this metric once and serves it to every dashboard, board pack and agent that asks.