RTO, returns and cancellations are different problems

Ask a brand for its return rate and you often get one blended number. Inside that number are three different events with different causes, different costs and different owners. Mixing them up means fixing the wrong problem.

Cancellation happens before dispatch. The customer (or the brand) cancels before the parcel ships. Cost: almost nothing. Some payment gateway and admin overhead at most.

Customer return happens after a successful delivery. The customer accepted the parcel, then decided to send it back: wrong size, changed mind, product did not match expectations. You pay reverse shipping and restocking, but the sale was real and the customer experience was completed. In the Shorages dataset of roughly 30,000 UAE orders, returns after delivery ran at about 2 to 3% for both COD and prepaid orders.

RTO (return to origin) means the parcel never reached the customer at all. It was shipped, delivery failed, and the courier brought it back. No sale, no payment collected, and you paid freight both ways. In the same Shorages dataset, 19.58% of COD orders ended in RTO against 2.85% of prepaid orders. One boundary: Shorages is UAE data, not Indian data, and we cite it only to show the shape of the payment-mode split. The Indian evidence sits in Shipway's FY25 shipment data, which shows the same shape at far larger scale: about 26% of non-prepaid orders returned against under 2% of prepaid.

A high customer-return rate points at product and expectation problems: sizing, product photos, descriptions, quality. The fix lives with the product and content team.

A high RTO rate points at order intent, address quality and delivery operations. The fix lives with checkout, verification and logistics. Almost all of it sits in COD orders.

If your dashboard blends the two, a product problem can look like a logistics problem, or the other way round. We have seen brands invest in sizing content when the real leak was unverified COD orders, and brands buy RTO tools when the real leak was a misleading product page.

There is also a reporting trap. Some platforms count an order that the customer refused at the door as a "return", and some count it as RTO. The same event lands in different buckets depending on the tool, which makes month-on-month comparisons meaningless if you switch tools or if the tool changes its definitions.

Split your numbers three ways for the last 90 days: cancellations before dispatch, RTO, and post-delivery returns. Compute each as a share of the orders that reached that stage (RTO as a share of shipped COD orders, not of all orders). Most brands can pull this from their courier aggregator and store platform in an afternoon.

Then look at the three numbers separately. Each has its own owner and its own fix. The rest of the posts in this series are about the RTO bucket, because for most Indian D2C brands it is the largest and the least examined.