Turn return reasons into one product improvement

Start with the return line, not the customer

Working with Shopify data? Prepare your return CSV and check which export fields you can use before importing.

Use one stable return-line ID, order reference, SKU, reason, quantity, refund amount, handling cost, date and currency. You do not need customer names, addresses or payment details for this diagnostic.

A line may contain several returned units. Keep quantity separate from the number of rows. Enter the total refund and total handling cost for that line, not a per-unit amount unless the line contains only one unit.

Normalise reasons without hiding the detail

“Too small,” “size small” and “tight fit” may belong in one working group, but keep the original wording available in the source export. Separate fit, damage, wrong item and preference where those distinctions support different actions.

Check the relevant SKUs and dates before combining files. Review duplicate return-line IDs instead of counting the same return twice. Keep currencies in separate totals; adding USD and GBP amounts does not produce a meaningful refund number.

A worked SKU example

Refunds are not the same as lost profit or money that can be recovered. Some returned stock may be resold; some costs are already incurred. Use the result to select an investigation, not to claim a savings figure.

Use the right denominator

Without shipment data, report returned units, reasons, refunds and handling costs. Do not call the result a return rate.

If a period has 5 returned units and 100 shipped units of the SKU, the period ratio is 5%. Those returned units may have shipped in an earlier period. The ratio is therefore not a matched-order return rate and should not be presented as one.

Whyset can compare selected-period shipments and returns; it does not match each returned unit to its original shipment cohort. A proper cohort analysis requires that additional linkage and enough time for returns to occur.

Make one change you can inspect later

Choose a specific action, an owner and a review date. For a fit issue, that might be checking the size chart against measured garments and adding an accurate measurement note. For a color issue, inspect photographs and actual batches before editing the listing.

Record when the change went live and what else changed during the next period, such as sales mix, promotions or suppliers. Compare like with like and leave time for orders to arrive and be returned. A smaller refund total on fewer sales does not establish an improvement.

The free CSV diagnostic runs in your browser. Saved workspaces keep imports and assigned actions for later review. Start with one recurring product issue rather than changing the whole catalogue at once.