Ecommerce returns reporting

Return rate can mean several things. Unit return rate = returned units / shipped units. Order return rate = orders with at least one return / shipped orders. Value return rate = value of returned goods / value shipped. Refund rate counts refunds, which is different because an exchange has no refund.

State the definition and the denominator on every report. A unit rate and an order rate can differ widely when a few orders contain many items, and a change in the mix of measures can look like a change in customer behaviour.

Return rate also depends on the time window. A 30-day window and a 90-day window produce different numbers for the same product, so fix the window before comparing periods.

Report returns against the month the order was placed, not the month the return arrived. That cohort view lets you compare products fairly and stops a seasonal spike from landing in the wrong period.

Keep a second view based on the return date for cash and stock planning. The cohort answers why; the cash view answers when. Both are useful, but they should not be mixed in one chart.

A cohort table with order month as rows and return month as columns shows how quickly returns arrive. If most come back within three weeks, a shorter window is enough for product decisions.

A refund reverses revenue, not the costs already incurred. Outbound shipping, return shipping, processing labour, repackaging, and any resale markdown stay with the business.

A hypothetical order: revenue 120, cost of goods 60, outbound shipping 8, return shipping 8, processing 3. If the item resells for 90, the order contributes 11 after the return. If it cannot be resold, it loses 79. The headline return rate shows neither outcome.

The cost of a return is not limited to the refund. In many cases the largest cost is the markdown required to sell the item again, which is easy to miss because it appears later in the stock report.

One row per product and variant, with units shipped, units returned, return rate, refund value, net revenue, and contribution after return costs. Add size, colour, channel, and reason as filters rather than as separate reports.

Sort by contribution lost rather than by return rate. A high rate on a low-value accessory matters less than a moderate rate on a high-value item with expensive handling.

Add a column for the return reason at the top level, even if the detailed codes live in a filter. The reason is usually more actionable than the rate, because it points to a specific fix.

Thresholds should come from your own history, not a generic benchmark. A fashion business and a parts supplier have different normal ranges, and a product with a short history needs a wider band.

Each exception should point to a likely cause and an owner. A size concentration points to the sizing chart; a damage concentration points to packaging or the carrier; a fault concentration points to the supplier.

Review the rules quarterly. A threshold that never triggers is either too loose or the process is not capturing the data, and both cases need attention.

Most returns are predictable. Sizing charts, photographs, measurements, and accurate stock availability reduce fit and expectation returns. Packaging reduces damage. Supplier quality conversations reduce faults. A clear policy reduces abuse.

Track whether the fix worked by comparing cohorts before and after the change. If the return rate stays flat, the cause was misdiagnosed, and the next change should be tested separately.

Returns also affect inventory. A returned item that cannot be resold as new should be valued at its resale value, not its original cost, or the stock report will overstate the asset.