COD, returns and your real ROAS: the number Indian D2C dashboards never show

Every ad platform reports ROAS on the value of orders it thinks it caused, at the moment they were placed. In most Western markets that is a reasonable approximation of revenue. In India it is not, because a large share of orders are cash on delivery, and a large share of those never complete. The gap between the two numbers is not a rounding error. For some brands it is the whole margin.

This post is the arithmetic, the operational changes that move it, and what makes food, FMCG and nutrition different from the fashion brands most returns advice is written for.

How big the gap is

Two datasets are worth anchoring on. The payments and checkout company GoKwik, which handles COD risk for Indian D2C brands, states in its 2026 COD guide that around 60% of Indian ecommerce orders are cash on delivery, rising to as high as 90% in tier 3 and tier 4 cities, and that without controls return to origin can run 30 to 50%.

Unicommerce's India D2C Report 2026, built from its own platform transaction data covering 6,000+ brands and 410 million shipments, is more specific. During the festive quarter, COD orders came back at 58% while prepaid orders returned at under 15%. Its month-by-month RTO series runs 39.2% at the November 2025 festive peak, 25.6% in January 2026 and 21.0% by March 2026 for the brands that made three operational changes. The same report puts tier 2 and tier 3 cities at 66% of incremental order volume in FY26, which is exactly where COD preference is highest.

Read those together and the picture is clear. Your growth is coming disproportionately from the geographies with the worst payment mix, and the difference between a 21% RTO and a 39% RTO is not luck.

The arithmetic, worked

Take a month where you spent ₹1,00,000 on Meta and the dashboard reports 3.0× ROAS, so ₹3,00,000 of attributed order value. Assume the payment mix and loss rates below. Use your own numbers, not these; the point is the shape of the calculation.

• Attributed order value: ₹3,00,000, at a ₹300 average order value, so 1,000 orders
• Payment mix: 60% COD (600 orders, ₹1,80,000), 40% prepaid (400 orders, ₹1,20,000)
• RTO on COD at 30%: 180 orders worth ₹54,000 never deliver
• RTO plus failed delivery on prepaid at 3%: 12 orders, ₹3,600
• Post-delivery returns and refunds at 4% of the 808 orders that did deliver: about ₹9,700

Delivered and kept revenue is roughly ₹2,32,700, so your net ROAS is about 2.3×, not 3.0×. That is a 22% haircut before you have paid for anything.

Now add the costs a return actually creates, because an RTO is not a neutral non-event. You paid forward shipping, you pay reverse shipping, you pay the COD handling fee on the attempt, you pay to repack, and on perishables you may not get the unit back into saleable stock at all. At a conservative ₹120 of round-trip logistics on each of the 180 COD orders that came back, that is another ₹21,600 of pure cost with zero revenue against it. Your effective ROAS on money that reached the bank is now closer to 2.1×, and your contribution margin after cost of goods is lower again.

This is the same point Kae Capital's Sunitha Viswanathan made to Inc42 in July 2026: a 3X ROAS campaign can be sitting on a 5% real margin once cost of goods, discounts and returns come out. The number is not lying. It is answering a different question from the one you are asking.

Why food, FMCG and nutrition behave differently from fashion

Most Indian returns advice is written about apparel, and the mechanics do not transfer. Rest of World reported in February 2023 that about 14.9% of products sold online in India in 2022 were returned overall, with fashion far higher at roughly 25% to 40%. Fashion's problem is post-delivery: fit, size, colour, and customers deliberately ordering three variants intending to keep one.

Food and supplements almost never fail that way. Nobody orders three jars of ghee to try one. Your losses concentrate somewhere else:

• Refusal at the door. The order was impulsive, the buyer forgot, or a family member declines it. This is an intent problem, and it is fixable before dispatch.
• Undeliverable address. Incomplete addresses and unreachable phone numbers, disproportionately in the tier 2 and 3 pincodes now driving your volume growth.
• Perishability on the way back. This is the one that makes food different. A returned kurta goes back on the shelf at full price. A 500g jar that has spent two weeks in reverse logistics, in an Indian summer, in a carton you did not pack the second time, is not stock you can confidently resell. On dated inventory, a carton that comes back with four months of shelf life left is a markdown, not inventory.

The consequence for your reporting is that food and nutrition brands should track RTO and cancellation separately from post-delivery returns, because they have completely different fixes. Lumping them into one "returns" percentage hides which lever to pull.

What actually changes the number

Unicommerce's report attributes the move from 39% to 21% RTO to three operational decisions running together rather than a better courier: a prepaid incentive at checkout, pincode-level courier routing based on real delivery performance, and address verification before dispatch. That matches what we see. In priority order:

1. Make prepaid the easier choice, do not ban COD

Blocking COD outright removes the buyers you cannot reach any other way, particularly outside metros. Converting a share of COD intenders is a different move: a small prepaid-only discount, free shipping on prepaid, faster promised delivery, or a COD handling fee that is disclosed honestly at checkout. Every converted order moves from a category returning at 58% to one returning under 15%, on Unicommerce's festive numbers.

2. Verify intent before you dispatch

An automated confirmation on WhatsApp or SMS that asks the buyer to confirm the order and the address, with a link to switch to prepaid, does two jobs at once. It catches wrong addresses and it forces a second moment of intent on the impulsive order.

3. Route and gate by pincode, not by gut

Your own delivery history already tells you which pincodes and which couriers fail. Use it: restrict COD above a value threshold in the worst zones, or route those zones to the carrier that actually performs there.

4. Raise order value so a delivered order absorbs more

Bundles and multi-packs change the arithmetic of the entire account, because a higher average order gives every campaign more room to be profitable at the same acquisition cost. This is exactly what we did on Vediko Origins, where cancellations and returns were eating into every win: a CRO audit first, then bundles to lift order value, then spend concentrated behind the products that actually move. Sales rose 1321% in eight months, ROAS ran consistently above 4× with some campaigns past 5.6×, conversion reached 2.8% and CAC fell 60%.

5. Feed the truth back to the ad platform

If you send Meta a Purchase event at checkout with the gross order value, you are training delivery to find people who place orders. If you also send a cancellation or refund signal, or send the purchase value at the point the order is delivered and settled, you train it to find people who keep them. That is the single highest-leverage change on this list and the one most brands never make.

A worked example: what removing the crutch looks like

Bon Fiction came to us with broken tracking, Meta failing, and a business quietly dependent on manual WhatsApp orders. The rebuild started with the fundamentals: who to target, which SKUs deserved the spotlight, campaign objectives pointed at the outcome the business actually wanted, a proper full-funnel structure, and Google switched on. In five months online revenue grew over 6×, entirely on paid media, with no COD crutch propping up the numbers. Prepaid revenue is real revenue, without the return and cancellation leakage that inflates a COD-heavy top line.

The wider point is that the fix is rarely one department's job. Media, checkout and fulfilment produce the number together, which is why our CRO and web work and our media work usually ship in the same sprint, and why the food and beverage version of this looks different from the FMCG one.

Dashboard ROAS tells you how good your advertising is at generating orders. Net ROAS tells you whether your business makes money. Only one of them pays salaries.

The one row to add to your weekly report

Ad spend, attributed order value, delivered order value, returns and refunds, round-trip logistics cost on failures, and net ROAS. Six columns. If you cannot fill them in today, that is the first thing to fix, before the next creative test and well before the next budget increase.