Why cash-on-delivery orders come back, and what actually reduces it

Return to origin — RTO — is the courier's term for a cash-on-delivery order that comes back instead of being delivered. In Indian ecommerce it is not an edge case. On COD orders it is routine, and on some categories and pin codes it eats a meaningful share of revenue before a single rupee is actually collected.

The cost is not just the missed sale. You pay to ship the parcel out, pay again for the courier to bring it back, repack whatever survives the trip, and the item sits unsold while it travels both ways. A COD book with a 20% RTO rate does not lose 20% of revenue — once both legs of shipping and the delay are counted, it loses considerably more.

Why the order comes back

Most RTO falls into a small number of causes, and they call for different fixes.

The customer never really decided to buy. COD has almost no commitment behind it — a tap can place an order the way a click adds to a wishlist elsewhere. Some share of every COD order was never a firm purchase.

The product does not match what they expected. Wrong size, wrong colour, a photo that flattered the item. This is a listing problem, not a courier problem, and no amount of confirmation calling fixes it.

The delivery attempt does not fit their day. A single attempt at a time nobody asked for, with no way to reschedule, produces refusals from genuinely interested buyers.

Fraud and habitual refusal. A small number of customers order repeatedly and refuse repeatedly, sometimes to check a product for free, sometimes for reasons that have nothing to do with the product. Small in count, disproportionate in cost.

What actually reduces it

Confirm the order before it ships, not after. A message asking the customer to confirm — WhatsApp works well for this, inside the rules that actually govern it — catches the orders that were never firm, before you have paid a courier to move one.

Charge a small token amount, even ₹49 or ₹99, rather than the full price. It is not about the money. Someone who has paid something has decided something, and the refusal rate on partially-paid COD orders is consistently lower than on fully COD ones.

Fix the listing before you fix the delivery. If a product returns often for the same stated reason — size, colour, quantity — that is written on the product page, not solved by calling the customer more insistently.

Track refusals by pin code and by customer, not just in aggregate. Courier partners already produce this data; most businesses never open the report. A handful of pin codes and a handful of repeat refusers usually account for a disproportionate share of the cost, and both can be flagged before dispatch rather than discovered after.

Give a real reason to pay upfront, even a small discount or faster dispatch. Shifting even a fifth of your COD mix to prepaid removes RTO from that fifth entirely.

None of this needs a platform. It needs someone to look at the courier's own NDR report every week and act on what it says.

What we would not build for this

A prediction model that scores every order for RTO risk before dispatch sounds like the serious answer, and for a large operation with real order volume it can be. For most small and mid-sized sellers it is not worth building. The courier partners' own non-delivery reports, a confirmation message before dispatch, and a blacklist of repeat refusers get most of the same result for a fraction of the cost, and they need no maintenance once set up.

Build something only once the manual version is genuinely running out of hours — the same test that applies to any automation decision. Before that point, the fix is mostly a process change: confirm before you ship, watch the report, and stop shipping COD to the pin codes and customers that are refusing it.

Where this sits against the rest of the store

RTO is an operations cost that shows up as a shipping line, which is why it survives unaddressed for years — nobody budgets a line for it, so nobody notices when it grows. If COD is a meaningful share of your orders, this is usually the cheapest fix available on the store as a whole, cheaper than most of what gets spent chasing more traffic to replace the orders that came back.