ONDC at three years: 288 million orders, 4% market share, and what the gap explains

In April 2026, ONDC quietly passed 288 million cumulative orders since its pilot launch in 2022. That is a real number, drawn from real transactions. It showed up in industry newsletters, ministry briefings, and a handful of venture-capital decks as evidence that India's open commerce infrastructure is finally working. The coverage was largely positive.

What most of it did not mention: ONDC holds an estimated 4.3% of India's e-commerce market. Amazon India and Flipkart, together, hold 83%.

Neither fact invalidates the other. But reading one without the other produces a misleading picture of where India's open network for digital commerce actually stands in 2026.

Where the 288 million came from

ONDC's transaction growth follows a recognisable DPI arc. A pilot in a handful of cities. A rollout across more geographies. Category expansion beyond retail. Milestone press releases at each inflection point. By January 2023, there were fewer than 600 sellers on the network. By August 2025, 167,500 retail sellers were transacting across more than 630 cities.

Food and grocery drove the largest share of volume, primarily through Magicpin, which built an ONDC-native buyer app and achieved meaningful adoption in Tier-1 markets. Mobility, financial services, specifically credit products embedded at the point of purchase, and logistics added volume on top. March 2025 saw 16.5 million transactions in a single month, and by early 2026 daily retail orders had settled around 150,000.

For a network that was processing a few hundred thousand transactions monthly in 2023, that trajectory is substantial. The question is what it means relative to the market ONDC is trying to reshape.

The four gaps the milestone number does not carry

The 288 million orders figure is accurate. These four gaps sit alongside it without cancelling it.

Market share

India's total e-commerce market is estimated at 950 million to 1 billion orders per quarter in 2025-26. ONDC's cumulative three-year total is roughly equivalent to what Amazon India processes in three to four weeks. The 4.3% market share is not a failure; it is a baseline. But it means the open network, despite real growth, is still operating at the margin of the ecosystem it aims to transform.

Seller activation vs seller onboarding

ONDC onboarded approximately 760,000 sellers and service providers by March 2025. Of those, around 167,500 were transacting regularly by August 2025, an activation rate of 22%. The gap between onboarding and activation is the harder operational problem. Onboarding requires registration and catalogue upload. Activation requires the seller to receive orders, fulfil them, manage returns, and find that the unit economics work. ONDC's network design does not solve that last part directly; it relies on seller apps and logistics partners to do it. Not all of them have.

Consumer awareness

Most Indian consumers buy from Meesho, Blinkit, Swiggy, or Amazon directly. They do not think of themselves as ONDC users, and they should not need to. The open network model works precisely when the protocol is invisible and the buyer app is the interface. But this means consumer trust in ONDC is entirely mediated by the buyer apps that sit on top of it. Many ONDC-native buyer apps are still working through UX gaps that incumbents spent years and significant capital smoothing out. Price discovery without checkout friction, reliable returns, and fast delivery are table stakes that most ONDC buyer apps have not consistently hit for all product categories.

Revenue and economics

ONDC's FY24-25 financials show ₹33.41 crore in revenue, up 127% year-on-year, against a net loss of ₹147.13 crore. The revenue growth is real and directionally good. The absolute loss is not small for a government-backed infrastructure entity that will eventually need a self-sustaining model. The core infrastructure fee that ONDC planned to introduce has faced repeated delays. A durable revenue mechanism is still a work in progress.

What ONDC has actually built

The honest assessment cuts both ways. Alongside the gaps, ONDC has delivered things that are not routine.

The Beckn protocol is the most significant. A publicly specified, open standard for discovery and transaction across categories (retail, mobility, financial services, logistics) is not an easy engineering or policy achievement. The fact that it has achieved enough adoption for cross-network transactions to work at all is meaningful. The protocol is now being referenced in digital commerce initiatives in other markets, which is a reasonable test of whether the underlying design is sound.

The structural separation of buyer apps, seller apps, and logistics is architecturally different from Amazon's model and from the super-app model of Meesho or Jio Mart. Whether that separation ultimately benefits small sellers or introduces coordination overhead that incumbents do not carry is an empirical question being answered by current market data. The architecture, at least, is in place.

The financial services layer is underreported. ONDC enables credit products to be offered at the point of purchase by multiple competing lenders, using open data rather than a single platform's proprietary buyer data. Early disbursement numbers through the network's credit stack are promising, and the policy implications (for financial inclusion and for competition in lending) are larger than most ONDC coverage suggests.

The CEO succession

Founding CEO Thampy Koshy stepped down in mid-2024 after a three-year tenure covering ONDC's entire launch arc. His departure, combined with the earlier resignation of Chief Business Officer Shireesh Joshi and the exit of non-executive chairperson R.S. Sharma in late 2023, left ONDC running under an interim executive committee for a substantial stretch. Vibhor Jain, who had been serving as COO, was appointed MD and CEO effective April 7, 2026.

Leadership transitions at infrastructure entities carry a different risk profile than at startups. The protocol is not going anywhere. But the commercial and policy decisions (how to monetise the network, which categories to prioritise, how to handle seller activation, and whether to disclose transaction quality methodology) require executive continuity. That continuity was disrupted for longer than was ideal. Jain brings genuine public-infrastructure experience, including prior roles in Aadhaar and tax-system projects. The question his tenure needs to answer is whether the COO-to-CEO path signals a priority on operational execution, and whether that is the right priority for this moment.

What 2026 needs to show

The UPI comparison will keep being made, and it is not entirely wrong. UPI is open infrastructure that required government mandate and patient capital before it achieved ubiquitous adoption. ONDC is the same type of bet. The difference is that UPI's user experience was immediately better than swiping a physical card or entering card numbers on a mobile browser. ONDC's user experience, for most product categories, is not yet immediately better than Amazon or Blinkit. Closing that gap is the precondition for the analogy to hold.

Four things would make 2026 a materially better year for ONDC than 2025:

• Seller activation past 30%. Getting from 22% to 30% or above requires solving training, device access, and unit-economics problems that the open network architecture cannot address by itself. This requires seller-side investment by participant apps, possibly with government incentives behind it.
• Transaction methodology transparency. Publishing a public methodology for how cumulative orders are counted, distinguishing arm's-length transactions from other types, would resolve the transaction quality question and strengthen the credibility of every future milestone.
• A live infrastructure fee. A working, implemented fee model that generates recurring revenue is the difference between a sustainable network and one dependent on grant cycles. The repeated delays need to end.
• One product category where ONDC buyer apps are clearly competitive. Price discovery parity plus checkout reliability plus returns handling, in one category, would change consumer behaviour faster than general awareness campaigns. Food delivery is the most likely candidate given existing volume.

ONDC has built infrastructure worth having. The honest question for 2026 is whether the execution can match the architecture, and whether the network will be transparent enough about its own metrics to know when it has.