India COD & RTO Benchmarks (2026)

Ask how much of Indian ecommerce is cash on delivery and you'll hear anything from under 30% to over 75%. Ask for the RTO rate on COD orders and the answers run from about 20% to almost 60%. Most of these numbers aren't wrong. They measure different things on different platforms in different years. This page collects the figures we could trace to a source, says exactly what each one measures and how old it is, and explains why they disagree.

Two rules applied throughout. Every figure links to where it was published. Where we did arithmetic on a source's numbers, the result is marked Derived. Where a widely repeated number couldn't be traced to an original publication, it sits in a separate list at the end and is not used anywhere else on the page.

Key numbers

Different platforms, different definitions. Read the "What it measures" column before quoting any of these side by side.

1COD share of orders

The best data is a public filing. Meesho's IPO prospectus discloses its COD share of shipped orders for five periods, so the trend can be read off directly. Meesho is a value-commerce marketplace and not representative of all Indian ecommerce. Its numbers sit at the high end of every range below.

Where COD lives: the Tier 2+ context

Every source above agrees, qualitatively, that COD is heavier outside the metros. We found no recent public dataset that gives COD share by city tier with a stated method, so treat any precise tier-level COD percentage you see with caution. What is well documented is where order growth is coming from. That matters because new and smaller-town shoppers are the ones who default to COD:

Shoppers clearly use UPI. Yet COD still accounts for most orders on mass-market platforms. The usual explanation is a trust gap: paying before the parcel arrives is a different decision from paying for something you can see. Bain's finding that more COD shoppers now pay by UPI at the doorstep fits that explanation, though it doesn't prove it. To a courier, an order paid by UPI at the door is still a COD order. It can still be refused.

2RTO: COD vs prepaid

RTO (return to origin) is an order that was shipped but never delivered, so it goes back to the seller. It's different from a customer return, where the product was delivered and later sent back. Some of the figures below measure RTO strictly. Others measure "returns", which mix the two. The second column says which.

3RTO by city, zone, speed and order value

Shipway's FY2025 report is the most detailed public breakdown we found. It covers COD ("non-prepaid") orders. City figures include only cities with at least 5,000 such orders in the year.

The gap between zones and between delivery speeds is as large as the gap between cities. This suggests geography acts partly through transit time. A customer who waits five days is more likely to have bought elsewhere, changed their mind or stopped answering the courier. The report shows correlation, not cause.

4The festive spike

Treat the Unicommerce festive figures as a direction, not a precise level. The "58%" has been read in two ways. Unicommerce calls them orders that "came back". Some vendor blogs restate it as "58% of COD orders end up as RTO". Both readings agree on the direction: COD returns rise sharply in the festive quarter, far more than prepaid returns do.

5RTO by category

This is the most-searched cut and the worst-sourced. We could not find a public category-level RTO dataset with a stated sample and method. Here is what can be attributed:

The direction is consistent: fashion, footwear and ethnic wear sit highest, and electronics and grocery lowest. We'd treat the specific ranges as practitioner estimates, not measured benchmarks. If you run a logistics or checkout platform and can publish category RTO with a sample size, tell us and we'll add it with credit.

6What one RTO costs

No public source publishes a market-wide average cost per RTO. What can be sourced are the components.

The fully loaded cost depends on things only the brand knows: packaging, the share of returned units that can't be resold, working capital tied up in transit, and above all the acquisition cost already spent on an order that never converts. Our RTO calculator starts from ₹250 per RTO. That's our own default assumption, not an industry benchmark, and you can replace it with your own numbers.

7NDR and fake COD orders

NDR (non-delivery report) is the courier's flag that a delivery attempt failed, for reasons like "customer not available", "refused" or "address incomplete". Unresolved NDRs become RTOs. We found no public NDR benchmark with a disclosed sample. ClickPost, a logistics software vendor, gives a range of 20–40% by category (Sep 2026) without a cited dataset. Shipway's finding that RTO climbs from 22% to 35% as the first attempt slips past five days is the closest measured evidence that NDR handling speed matters.

Fake or fraudulent COD orders are a common talking point. We could not trace any of the percentages in circulation to a published dataset (see the next section). The best primary signal is indirect. Meesho's prospectus says it may restrict COD for "repeat offenders" or charge a convenience fee to consumers with high RTO rates. A platform wouldn't build that unless a small group of buyers caused a large share of failed deliveries, but the filing gives no figure for it.

8Why the numbers disagree

1. Different denominators. "COD share" can mean share of orders, of shipments, of GMV, or of consumers whose last purchase was COD (the IIMA 65%). Shipments exclude orders cancelled before dispatch, and value shares fall below volume shares because COD skews to lower order values. "RTO rate" can be measured against all shipped orders, COD orders only, or "returns" that also include customer-initiated returns (Unicommerce's 20.9%).
2. Platform mix. A value marketplace (Meesho, 72–77% COD), a D2C checkout network (GoKwik, about 60%), all-ecommerce shipments (Redseer 2020, 28%) and quick commerce (lower COD reliance than e-retail, per Bain) are different populations. A single "India COD share" blends them, and the weights change with each source's coverage.
3. D2C vs marketplace. Marketplaces showed about 26.3% returns and D2C sites 6.2% in Unicommerce's FY2023 index (the marketplace figure's wording is ambiguous; see section 2). Marketplace return policies and buyer mix differ from brand sites. Benchmarks from one shouldn't be applied to the other.
4. Year and season. COD share dropped during COVID (41% to 28% of shipments in 2020), then rose again with Tier 2+ growth, and is now falling on Meesho (88.7% in FY2023 to 72% in 2025). Festive quarters push COD returns far above annual averages (Unicommerce's 58% festive figure against a 20.9% annual one, though the years differ).
5. Selection bias in vendor data. Most RTO figures come from companies that sell RTO-reduction, checkout or shipping software. Their samples are their customers, often brands that signed up because RTO was a problem. Their RTO figures may also already reflect their own tools. That doesn't make the numbers wrong, but they aren't random samples of the market.
6. Citation chains. Many figures that look independent trace back to a single original, and some to none we could find. "26% COD vs under 2% prepaid" appears across dozens of blogs. The traceable original is Shipway's FY2025 report, though some blogs attribute it to GoKwik. "60–65% COD" traces to a 2024 ET Prime piece we couldn't access. The more often a number is repeated, the more important it is to find where it started.

9Widely repeated, not verified

These figures circulate widely in industry blogs and decks. We tried to trace each one to a primary publication and couldn't, so we haven't used them above. If you know the original source, send it and we'll move the figure into the tables.

10Methodology

Limitations: several sources are vendors with a commercial interest in RTO being seen as high, us included. Corrections and new sources: hello@callfox.in.

Cite this page

FAQ

Using these numbers

Your own RTO rate, split by payment method, pin-code zone and month, will tell you more than any benchmark here. If you want to see what a given RTO rate costs at your order volume, the RTO calculator takes your inputs. If you're curious how pre-dispatch confirmation calls work as one RTO lever among several, here's how Callfox does it.