Export order-level return data from each marketplace, map every return to a SKU (size and colour level), standardise the reason codes into a few groups such as fit, quality, damage, wrong item and change of mind, then calculate return rate and reason share per SKU. Rank SKUs by total return cost, not just percentage, and act on the top few first.
Why a blended return rate hides the real problem
A brand-level return rate of, say, 20% tells you very little. It averages a kurta that is returned for fit every second order with a plain tee that is almost never sent back. When you look by SKU, the picture usually narrows to a handful of styles, sizes or colours that cause most of the cost.
Reason-level analysis adds the second dimension. A SKU returned for size and fit needs a size chart fix. A SKU returned as damaged or defective points to packaging or the factory. A SKU returned as not as described points to photos or copy. Each reason has a different owner and a different fix.
Pull the right data from each marketplace
You need order-level, not summary-level, data. From your seller panel or report section, download the returns or claims report along with the sales and settlement reports for the same period. Report names and columns differ by marketplace and change over time, so check what your panel currently offers.
Aim to capture these fields for every return:
- Order ID and return or RTO ID
- SKU or style code, plus size and colour if available
- Order date, return initiated date and return received date
- Marketplace return reason or sub-reason
- Return type: customer return, RTO (undelivered) or exchange
- Quality check outcome, if the marketplace shares it
- Any return-related deductions in the settlement
Use a consistent period, ideally by order month rather than return month, so returns are matched to the sales that generated them. Returns often arrive weeks after the sale, so recent months will look better than they are until the window closes.
Clean the data and group the reasons
Marketplace reason lists are long and sometimes overlap. Before counting, map them into a small set of your own groups. A practical starting set:
- Size and fit
- Quality or defect
- Damaged in transit
- Wrong or missing item
- Not as described or shown
- Change of mind or no longer needed
- Other
Keep a mapping table so the same marketplace reason always lands in the same group. Also standardise SKU codes: the same product may appear with different formats across marketplaces, so build a master SKU map before you combine channels. Finally, tag each row as customer return or RTO, because they behave differently, as covered below.
Calculate the metrics that matter per SKU
Build a pivot with SKU in rows and reason groups in columns. Then compute:
| Metric | How to calculate | What it tells you |
|---|---|---|
| Return rate | Returned units / units sold | How often the SKU comes back |
| Reason share | Returns for one reason / total SKU returns | What is driving the returns |
| Return cost | Returns x (fees, reverse shipping, write-offs per return) | Which SKUs hurt margin most |
| Sellable recovery | Units restocked as sellable / units returned | How much value comes back |
Take per-return costs from your settlement reports and rate card rather than assumptions, since fees and reverse logistics terms depend on your agreement and may change. Rank by return cost, not only percentage. A 12% return rate on a high-volume, high-price SKU can cost more than a 30% rate on a slow seller.
Separate RTO and logistics issues from product issues
RTO (return to origin) happens when a shipment is not delivered, so the customer never saw the product. Customer returns happen after delivery. Mixing the two leads to wrong fixes: you cannot repair an RTO problem with a better size chart.
For RTO, look at pin code, payment mode, courier and delivery attempts alongside SKU. For customer returns, focus on reason groups. Also watch for returns where the item received back is different, damaged or missing. Those are not product problems; they may be claim cases. Check your seller agreement for the claim window and the evidence required, and file within it.
Turn the analysis into fixes and recovery
Pick the top five SKUs by return cost and write one action for each, with an owner and a review date:
- Fit-driven returns: revise size charts, add model measurements, review grading with the manufacturer.
- Quality or defect: raise with production, add inspection before dispatch.
- Not as described: reshoot images, correct colour, fabric and dimension details.
- Transit damage: improve packaging and compare courier performance.
- Wrong or missing item or tampered returns: build the claim file with order, return and QC evidence.
Recheck the same SKUs after one or two full return cycles to see whether the reason share moved. Tools such as MarketplaceMetrics by Fundle can reconcile returns against settlements and track recovery cases, so deductions and claims are not lost in spreadsheets. Confirm which marketplace reports are supported for your channels.
Worked example
Illustrative example only. All numbers below are made up to show the method, not benchmarks.
| SKU | Units sold | Returns | Return rate | Top reason |
|---|---|---|---|---|
| KURTA-BLU-M | 120 | 30 | 25% | Size and fit (18 of 30) |
| TEE-BLK-L | 150 | 12 | 8% | Damaged in transit (5 of 12) |
| JKT-GRY-S | 40 | 10 | 25% | Not as described (6 of 10) |
Assume an illustrative all-in cost of ₹150 per return. KURTA-BLU-M costs 30 x ₹150 = ₹4,500, TEE-BLK-L costs ₹1,800 and JKT-GRY-S costs ₹1,500. Though KURTA-BLU-M and JKT-GRY-S share a 25% return rate, the kurta costs three times as much because of volume. The first fix is therefore the kurta's size chart (60% of its returns are fit-related), then the jacket's photos and colour description.
Questions and answers
How often should I review return reasons by SKU?
Review monthly, by order month, and revisit older months once their return window has closed. For fast-moving or new launches, a weekly check in the first few weeks helps you catch fit or quality problems before large volumes ship.
What is a good return rate for a SKU?
There is no universal benchmark. Return rates vary by category, price and marketplace. Compare each SKU against your own category average and its own history, and focus on SKUs that are well above peers or rising month to month.
Should I analyze RTO and customer returns together?
Keep them in the same dataset but tag them separately. RTO reflects delivery, address, payment mode and courier issues, while customer returns reflect product, listing and expectation issues. Analysing them separately avoids applying the wrong fix.
Where do I find the cost of each return?
Check your seller agreement, the marketplace rate card and the settlement report lines for returns. Fees, reverse shipping and deductions can change, so confirm the latest terms and take actual deductions from settlements instead of assuming a fixed figure.