The discount you sign off in October is a January cost. Festive buyers who came in on 30% off, paid cash on delivery and shipped to a pincode you've never served before are the cohort that generates refusals, RTO freight and a returns queue two months later. So the honest answer to how to reduce RTO rate on a Shopify store in India is not a single app or a tighter policy page. It's pricing: work out what a returned-to-origin COD order actually costs you, then spend some of that number buying the order back as prepaid before it ever ships. Everything else — OTP, address checks, NDR calling, shorter return windows on sale collections — is cleanup on the orders you couldn't convert.
There's a timing reason to do this now rather than in week one of the sale. Shopify's late-September release lets return rules vary by market and apply to specific collections, which means a 10-day window on your Festive Sale collection and 30 days everywhere else is finally a settings change instead of a custom build. It has to be configured before orders land. Retro-fitting a policy to orders already placed is a chargeback waiting to happen.
What a COD return to origin costs per order
Most founders quote RTO as a percentage and stop there. The percentage is useless without a rupee figure attached. Here's the calculation on a fairly ordinary apparel or accessories order — plug your own numbers in, the structure is what matters.
Order value ₹1,499. COGS ₹600. Forward freight ₹75. COD remittance fee ₹35. Blended CAC from Meta and Google during festive, ₹350.
A delivered COD order contributes 1,499 − 600 − 75 − 35 − 350 = ₹439.
An RTO costs you the forward leg you already paid (₹75), the return leg (₹75), packaging (₹20), inbound QC and restocking labour (₹15), a write-off allowance because roughly one in twelve returned units comes back unsellable — 8% of ₹600 = ₹48 — and the ad spend that bought the order, ₹350. Total: ₹583 gone, per RTO.
Run 100 COD orders at a 28% RTO rate. Seventy-two deliver: 72 × 439 = ₹31,608. Twenty-eight come back: 28 × 583 = ₹16,324. Net ₹15,284, or ₹153 per order placed. Your dashboard says AOV ₹1,499. Your bank says ₹153.
Prepaid discount vs COD order margin: how much you can actually afford to pay
Now price the same order prepaid. Payment gateway fee around 2% on Razorpay or Cashfree, no COD remittance charge, and RTO drops to roughly 4% because the customer has already parted with money.
At a 5% prepaid discount — ₹75 off, order value ₹1,424 — a delivered order contributes 1,424 − 600 − 75 − 28 (gateway) − 350 = ₹371. Ninety-six deliver: ₹35,616. Four come back and cost 150 + 20 + 15 + 48 + 28 + 350 = ₹611 each, so ₹2,444. Net ₹33,172, or ₹332 per order.
Same product, same ad spend, more than double the contribution. Push the question further: what's the largest prepaid discount that still leaves you no worse off than COD? Solve for the discount d where the prepaid cohort nets the same ₹15,284 per hundred, and you get roughly ₹265 — about 18% off. That is your ceiling, not your target. You should not need to pay 18% to move a customer from COD to UPI. But knowing the ceiling changes the argument in the marketing meeting, because it reframes a 7% prepaid incentive from "margin we're giving away" to "RTO reserve we're spending early".
Two caveats worth saying out loud. First, this assumes conversion holds when you nudge toward prepaid, and in genuinely COD-dependent segments it doesn't — tier-3 apparel under ₹800 is the obvious case. Second, the 28% and 4% are illustrative. Pull your own from Shiprocket or Delhivery panel exports, split by payment mode, for the last six months.
Reducing the RTO rate starts at the COD button
The cheapest intervention is not showing cash on delivery to the orders most likely to come back. In practice, four rules cover most of the damage during a festive sale:
- Cap COD by cart value. Above roughly 1.8× your AOV, refusals climb sharply and the cash loss per RTO scales with it. A ₹4,000 COD order that bounces costs you the same freight as a ₹900 one and four times the working capital.
- Turn COD off on the deepest-discount collection. If a line is at 40% off, the buyer has very little skin in the game and plenty of competing carts open.
- Blacklist your worst pincodes. Every courier panel will give you RTO by pincode. Take the bottom 2% by volume-weighted RTO and make them prepaid-only for the sale period.
- Charge a COD fee of ₹49–₹79 and waive it on prepaid. Framed as a waiver rather than a surcharge, it converts a meaningful slice without reading as a penalty.
The awkward part: native payment customisations on Shopify run on Functions, and Functions-based payment rules are a Plus feature. On a standard plan you'll do this with a third-party checkout app or a COD-control app, which is fine and costs a fraction of the upgrade. Do not move to Plus for this reason alone — we've talked merchants out of it more than once. If you're already on Plus, or close to the GMV where the rest of the platform case stands up on its own, a custom payment function gives you rules a generic app can't express, like hiding COD only when the cart contains a sale SKU and the shipping pincode sits in your high-risk list. That's a small custom build, not a project.
Address verification and OTP confirmation on COD orders
Checkout-stage OTP on COD works, and it costs conversion. Both things are true. It filters the junk orders — mistyped numbers, abandoned-cart bots, the same phone number ordering four times — but it adds friction at the exact moment your mobile-first, low-bandwidth buyer is already struggling with a slow 4G connection in a crowded festive week.
What we'd do instead on most stores: let the order through, then confirm within 30 minutes on WhatsApp. One message with the item, the payable amount and two buttons — confirm or cancel. Cancellations at that stage cost you nothing but a picklist edit. An RTO costs ₹583. Trading a cancelled order for a prevented RTO is a good trade every single time.
Underneath that, run basic address hygiene before the manifest goes out. Flag pincode-to-city mismatches, addresses under about 25 characters, phone numbers that repeat across different names, and apartment-less addresses in metro pincodes. Most courier aggregators now also publish an RTO risk score per order; treat it as a sorting tool for who gets a human call, not as gospel. The one-line version: verify the address before dispatch, confirm the intent after the order, and don't do both at checkout.
Shorter return windows on sale items, now settable per market
Your standard 30-day window is priced for full-price demand. On festive discount SKUs it's a subsidy. A buyer who paid ₹899 for something listed at ₹1,499 has every incentive to hold it through the season and return it in the first week of January, by which point the item is seasonal dead stock and you're selling it at clearance.
Shopify's return rules live in your returns settings, and the September update means they can now differ by market and be scoped to particular collections. The configuration that works:
- Build a dedicated Festive Sale collection before the campaign goes live. Automated collection, condition on a tag you apply at discount time, so adding a SKU to the sale automatically pulls it into the shorter-window rule.
- Set 7 or 10 days on that collection against a 30-day default. Ten days is defensible to a customer; three days reads as hostile and generates support tickets that cost more than the returns.
- Apply a restocking fee or customer-paid return shipping on the sale collection if your category justifies it. Apparel usually can't. Electronics accessories and home usually can.
- Set the rule separately for your UAE, UK or US markets if you sell cross-border. Return economics on a Dubai order are nothing like a Bengaluru one, and until this update most merchants were running one blunt policy across all of them.
Then make the written policy page, the product page line and the order confirmation email say the same thing. Indian e-commerce rules require return and refund terms to be clearly disclosed before purchase, and a shorter window buried only in settings is the kind of thing that loses you a payment dispute. If your product pages don't currently surface the return window at all, add it near the add-to-cart. It cuts pre-purchase support queries and it protects the policy.
The benchmark question, answered honestly
People search for an RTO benchmark after a festive sale, hoping for a number to measure themselves against. We won't give you one we can't source, and any agency quoting you a tidy national average for post-Diwali RTO is quoting a vibe.
The only benchmark that means anything is your own. Export COD orders for 10 October to 5 November, compute RTO for that cohort, and compare it to your trailing six-month COD RTO on non-discounted orders. In our experience that festive cohort lands several points above a store's own baseline, and the gap widens the deeper the discount and the newer the pincode. If the gap is under two points, your COD controls are already working and you should spend your October energy elsewhere. If it's into double digits, the COD button is doing damage that no amount of NDR calling will fix.
Cutting COD refusals during the Diwali delivery window
Refusals during the festive fortnight have causes that are logistical rather than behavioural. People travel. Cash in hand gets spent on other things. Courier networks run at capacity and attempt deliveries at odd hours. Four things help:
Send a WhatsApp message the evening before the out-for-delivery scan with the exact payable amount. Most refusals we see on first attempt are a buyer who didn't have ₹1,499 in cash at 11am on a Tuesday. Twelve hours of notice solves that.
Work your NDR queue inside 24 hours, not 48. A second attempt arranged on the same day converts far better than one scheduled after the customer has mentally written the order off.
Avoid dispatching high-value COD into a region the day before its major local holiday. Delhivery and Blue Dart will still attempt it; nobody will be home.
And split carrier allocation by pincode performance rather than by flat rate card. The cheapest courier on a ₹75 slab that RTOs 8 points higher is not the cheapest courier.
Planning for the January returns spike
Two separate waves hit in January and it helps to model them apart. The first is end-of-window returns from the late-November and December cohort, which is predictable: take your December delivered volume, apply your category return rate, and you have the unit count. The second is end-of-season sale returns, which arrive fast and skew toward sizing and buyer's-remorse reasons.
Three things to put in the October plan:
- Cash. Hold an RTO and refund reserve rather than treating November's settlement as profit. If you're running the numbers above, the reserve is the RTO cost per order multiplied by projected COD volume. That money is already spent, you just haven't paid it yet.
- Reverse-logistics capacity. Inbound QC is the bottleneck, not pickup. A returned unit that sits unQC'd for eleven days is a unit you can't resell in the January clearance window where it still has value.
- Accounting. Credit notes and the GSTR-1 amendments that follow are genuine January admin load. If one person does this manually and they also run customer support, budget for the collision.
What to set up this week
Pull your last six months of COD orders out of your courier panel and compute RTO by pincode and by discount depth. Build the festive collection and set its return window before any discount goes live. Decide your COD cap and your prepaid incentive in the same meeting, using the same arithmetic, because they're the same decision. If the prepaid incentive is funded from the RTO line rather than the margin line, nobody has to argue about it in November.
If you want a second pair of eyes on the checkout rules and returns configuration before the sale starts, our free store audit covers both. It's a better use of an hour in October than it is in December.