RTO and COD in Indian D2C: Your ROAS Is Wrong Until Delivery Confirms

Your Meta dashboard says the campaign did 3.2x. Your Shopify store agrees: ₹32,00,000 in orders against ₹10,00,000 of spend.

Eleven days later, 1,400 of those orders have come back undelivered. The customer never paid — they never had to, because it was cash on delivery — and you paid the courier twice, once out and once back. The campaign's real return is not 3.2x. It is closer to 2.2x on revenue and, once the reverse logistics are counted, meaningfully below breakeven.

This is not an edge case in Indian D2C. It is the default condition, and almost no reporting stack accounts for it.

Why this is structurally different

Three things happen between the dashboard number and your bank account, and each one is invisible in standard reporting:

1. RTO removes a fifth to a third of orders outright.
2. Reverse logistics charges you for the privilege, on orders that produced nothing.
3. Remittance lag means even successful COD revenue arrives one to three weeks after the order, which is a working-capital problem separate from the profitability one.

The three numbers you actually need

Most Indian D2C reporting has one number: order-time ROAS. Here are the three that replace it.

Delivered ROAS  = Delivered order revenue ÷ Ad spend
Realised ROAS   = (Delivered revenue − RTO costs) ÷ Ad spend
Contribution ROAS = (Delivered contribution margin − RTO costs) ÷ Ad spend

Each strips one more layer of fiction. The third has a threshold — below 1.0, the campaign consumed more than the gross profit it produced.

The uncomfortable part: a team optimising to "keep ROAS above 3x" would call this campaign a marginal success and scale it.

Computing delivered ROAS

The join is straightforward once the courier status data is in the same place as the orders. The structural point is that an order's outcome is not known on the day it was placed, so any query grouping by order date must handle immaturity explicitly.

Reducing RTO, in order of effect

RTO is not weather. A large share of it is addressable, and the interventions have very different effort-to-return profiles.

That last point is where reporting turns into money. RTO is not evenly distributed:

The patterns this surfaces are consistent across most Indian D2C businesses: COD RTO rates are multiples of prepaid rates, low-AOV orders return far more often than high-AOV ones, and a minority of pincodes account for a disproportionate share of the cost. Campaign-level variation is often as large as the COD-versus-prepaid gap — which means creative and audience choices are quietly determining your RTO rate, and nobody is looking.

What this does to your targets

Everything downstream shifts, and shifts in the same direction.

Frequently asked questions

What RTO rate is normal?

It varies widely by category, price point and geography, so treat any single published figure with suspicion. What is consistent is the shape: COD rates are a large multiple of prepaid rates, low-AOV orders return more than high-AOV ones, and rates concentrate heavily in specific pincodes. Measure your own by segment — the segment variation is far more actionable than any benchmark.

Should I just stop offering COD?

Rarely the right answer. COD is a large share of the addressable market in India, and removing it removes those customers entirely rather than converting them to prepaid. The better move is to price the difference: a prepaid discount that is smaller than your per-order RTO cost is profitable on every order it converts, and costs you nothing on the ones it does not.

How do I handle orders still in transit?

Exclude them from settled reporting with a maturity filter, and show them separately as in-flight. Never blend a settled series with an immature one — that is what makes recent days look like a collapse and triggers panic-pausing of working campaigns.

Does RTO affect prepaid orders too?

Yes, but far less, and the economics differ: a prepaid RTO becomes a refund, so you return the money and eat the freight both ways, but the customer had committed something and self-selected as more likely to accept. Track them separately — blending prepaid and COD RTO rates hides the single most important segmentation in the dataset.

Where does the RTO cost belong — marketing or operations?

In contribution margin, which means it lands in both. It is a variable cost per order, so it reduces the margin every ROAS target is derived from. Assigning it to operations and leaving marketing targets unchanged is exactly how a business runs campaigns that operations then loses money delivering.

The summary

• A COD order is a promise, not revenue. Order-time ROAS measures intentions; delivery decides whether they became money.
• Three numbers replace it: delivered ROAS (was it delivered), realised ROAS (net of RTO costs), contribution ROAS (net of everything, with a 1.0 threshold).
• RTO costs are paid twice — forward and reverse freight — plus handling and a write-off on returned stock, on orders that produced nothing.
• Always use a maturity filter. Reporting unsettled days makes working campaigns look broken.
• RTO concentrates by payment method, AOV band, pincode and campaign — that last one means your creative choices are setting your RTO rate.
• Re-derive breakeven ROAS on the delivery rate. At 45% margin and 22% RTO the bar rises roughly 40%, and every platform bid target should move with it.
• Feed delivery outcomes back through the conversion APIs, so the algorithm optimises for customers who accept parcels rather than ones who click.