How to reduce RTO, in the order that actually works

Every COD store eventually assembles the same list: confirm orders, block bad pincodes, push prepaid, call the NDRs. All of it works. What decides whether your RTO rate actually moves is the order you do it in — because each step changes the population the next step is measured against.

Step 0: measure RTO where it happens

A single store-wide RTO percentage is a number you can't act on. Split it the way the losses split: by payment method (COD versus prepaid), by pincode cluster, by product and price band, by traffic source, and by the day of week the order was placed. The first honest cut usually shows the problem is concentrated, not spread — a handful of pincodes, one discount campaign, one product that attracts impulse orders.

Step 1: stop the orders you never wanted

The cheapest RTO is the one that was never accepted. Verifying the phone number with an OTP before the order exists (1-click checkout) removes the entire category of fake and mistyped numbers — the orders nobody can call back, and the ones that were never placed by a real buyer at all. Address quality follows the same logic: a structured, autofilled address beats a free-text box that produces 'near the big temple' as a delivery instruction.

Step 2: price the risk instead of refusing it

The blunt version of RTO control is switching COD off. It works, and it costs you the shoppers who only buy on COD. The graded version scores each order — history on the phone number, pincode performance, cart value, distance, product type — and applies a different rule per band. Low risk keeps COD untouched. Middle risk gets nudged: a prepaid discount, a small COD fee, partial COD where a token advance is paid online. Only the top band loses COD entirely.

• A nudge you can measure beats a ban you can't: track conversion and RTO per risk band, not just RTO overall.
• Keep the shopper's exit visible — a risky order that pays an advance is revenue kept, not a customer refused.
• Review band thresholds monthly. Courier performance and your own catalogue both drift.

Step 3: confirm before dispatch, not after

Once a parcel is in the network, cancelling it costs you both legs of shipping. A confirmation message sent within minutes of a COD order — on WhatsApp, with the order summary and a one-tap confirm or cancel — moves that decision to the cheapest possible moment. Cancellations here look like lost orders on a dashboard. They are avoided losses on the P&L.

Step 4: work the NDRs like a queue, not a report

A non-delivery report is a live window, not an archive entry. The reattempt window is short, and the parcels that get a fast, specific follow-up — correct the address, agree a time, confirm the buyer is home — come back from the brink. The ones that sit in a spreadsheet until Friday become RTOs.

Step 5: change the mix, not just the rules

Every step above manages COD risk. The structural fix is having less of it: prepaid incentives, UPI as the default, offers that reward paying now. Prepaid share is the one number that improves RTO, cash flow, ad measurement and remittance timing at once.

You don't cut RTO with a single rule. You cut it by moving each decision earlier — to the moment when changing your mind is still free.

Run the steps in this order and each one is measured on a cleaner population than the last. Run them backwards and you'll spend your effort chasing NDRs generated by orders that should never have been accepted.