Divide returned units by units in the same order cohort for each SKU. Use dispatched units as the denominator for total return rate, but also split RTO (undelivered, over dispatched units) from customer returns (over delivered units). Group by order month, wait out the return window, and go down to size level where your reports allow it.
What return rate by SKU actually measures
Return rate by SKU tells you what share of a product's orders came back. A store-wide figure averages a good product with a bad one. A SKU-level figure shows which style, size or colour is leaking margin through reverse shipping, damaged stock and lost sales.
The basic formula is:
Return rate (%) = Returned units ÷ Units in the cohort × 100
The hard part is choosing the cohort (the denominator) and deciding what counts as a return. Get those two choices right and the rest is arithmetic.
Data you need and where to find it
Pull these fields for the same period from your marketplace seller panel. Report names and columns differ by marketplace, so check what your panel exports.
- Order or sales report: order ID, SKU or style ID, size, quantity, order date, dispatch status, selling price.
- Returns report: order ID, SKU, quantity returned, return date, return type, return reason.
- Settlement report: the amounts actually paid or deducted against each order and return.
Join the sheets on order ID and SKU. Check for duplicates, partial returns (one unit of a multi-unit order) and SKU codes that differ between sheets, such as a marketplace style ID versus your internal code. Map them to one master SKU first.
Step-by-step: calculate return rate by SKU
- Choose a cohort. Use orders placed in a given month, not returns received in that month. Otherwise a sales dip makes return rates look worse.
- Count dispatched units per SKU in that cohort.
- Count delivered units (dispatched minus RTO and in-transit).
- Tag each return as RTO (never delivered) or customer return (delivered, then sent back).
- Calculate three rates per SKU: RTO rate, customer return rate and total return rate.
- Repeat by size for SKUs with enough volume to be meaningful.
- Add a value view by multiplying returned units by selling price, so a high-priced SKU is not treated like a low-priced one.
Keep RTO and customer returns separate
They have different causes and different fixes. RTO is often linked to address quality, delivery attempts or order cancellations at the doorstep. Customer returns are more about fit, quality, description or expectation.
| Metric | Formula | What it points to |
|---|---|---|
| RTO rate | RTO units ÷ dispatched units | Delivery, address and courier issues |
| Customer return rate | Customer-returned units ÷ delivered units | Fit, quality, listing accuracy |
| Total return rate | (RTO + customer returns) ÷ dispatched units | Overall leakage from the SKU |
Using delivered units for customer returns avoids penalising a SKU for parcels the customer never received. Whichever denominator you pick, use it consistently so month-to-month comparisons hold.
Handle the timing lag
Returns arrive days or weeks after the sale, so a recent cohort always looks better than it will end up. Wait until the return window for your category has closed before treating a cohort as final. Confirm the window in your seller agreement or the marketplace's current return policy, since it can vary by category and change over time.
For recent months, label the figure as provisional and restate it later. Also watch for returns that are booked in a different month from the original sale in your settlement report, which can confuse a simple month-on-month view.
Turn the numbers into action
- Rank SKUs by total return rate and by returned value. Review the top few first.
- Check sizes. One size with a far higher rate than its neighbours usually signals a size chart or fit issue.
- Read return reasons to separate listing problems from product defects.
- Check the money trail. Compare returns against settlement deductions, and look for returns that came back damaged, wrong or not at all, which may be recoverable under marketplace claim processes. Confirm eligibility and timelines in your agreement.
This is where a tool helps once volumes grow. MarketplaceMetrics by Fundle reconciles sales and settlements, checks fees against agreed terms, and tracks returns and recovery cases. Myntra Excel reports are reviewed today, and other marketplaces are in scope discussion.
Worked example
Illustrative example only. The figures below are made up to show the method, not benchmarks.
| Item | SKU A | SKU B |
|---|---|---|
| Dispatched units | 400 | 200 |
| RTO units | 40 | 8 |
| Delivered units | 360 | 192 |
| Customer-returned units | 72 | 19 |
| RTO rate | 10.0% | 4.0% |
| Customer return rate | 20.0% | 9.9% |
| Total return rate | 28.0% | 13.5% |
For SKU A: 40 + 72 = 112 returned units, and 112 ÷ 400 = 28%. At an assumed selling price of ₹800 (illustrative), returned value is 112 × ₹800 = ₹89,600. Both SKUs may look acceptable in a blended figure of (112 + 27) ÷ 600 = 23.2%, yet SKU A is clearly the one to investigate, starting with its size-level split.
Questions and answers
Should I use dispatched or delivered units as the denominator?
Use dispatched units for total return rate, since RTO is also lost sales. Use delivered units for customer return rate, since only delivered orders can be returned by customers. State your choice in the report and keep it fixed over time.
How many units does a SKU need before the rate is reliable?
There is no fixed threshold. With very few units, one return swings the percentage sharply. Group small SKUs by style or colour, combine several months, or show the unit count beside the percentage so readers can judge reliability themselves.
Why does my return rate keep changing for past months?
Returns are booked after the sale, so older cohorts keep receiving returns until the return window closes. Calculate by order-month cohort and treat recent months as provisional. Restating them later is normal, not a data error.
Where do I find the return type and reason?
Look in the returns or RTO reports in your marketplace seller panel. Column names vary, so check each export. If a field is missing, ask your marketplace account manager or support which report carries it, and confirm the latest report formats.