ONDC in 2026: open protocol, closed liquidity

ONDC in 2026 has a paradox at its centre. The programme set out to do for Indian commerce what UPI did for payments — break open a market dominated by a handful of platforms and let any developer participate on equal terms. Four years in, the headline metrics look like progress: 6.5 million peak monthly transactions, 7.6 lakh sellers onboarded, 630 cities covered. But read them carefully and a different picture emerges. The open protocol is real. The liquidity is not open. One segment genuinely works. The rest is a mix of subsidy-driven volumes, inflated onboarding counts, and a structural question about what "open commerce" can actually solve.

The transaction-counting problem

Start with the number most cited: millions of monthly transactions on ONDC's open network. What that figure does not clarify is how many involve two genuinely independent participants: a buyer on one platform, a seller on another. The alternative is a single entity acting as both buyer app and seller app, routing transactions through the Beckn protocol internally.

Namma Yatri, ONDC's most-cited success, operates as both buyer app and seller app. The ride request and the driver supply are managed by the same organisation. The transaction appears in ONDC's open-network tally, but both sides of the market are controlled by one entity. That is not fraud; it is a legitimate way to build on an open protocol. But it is not the interoperability that "open network" implies to a reader of the headline figure.

Mobility works. Retail does not. There is a reason for both.

Namma Yatri's numbers are worth taking seriously: 130 million rides, over 700,000 drivers, Rs 2,000 crore in driver earnings, approximately $5 million in total capital spent. That last figure is the one worth sitting with — it is a fraction of what incumbent ride-hailing platforms spend in a single market in a year.

Mobility works on ONDC because ride-hailing is, from a commerce-infrastructure perspective, a simple problem. No inventory. No SKU catalogue. No returns. No counterfeit risk. No last-mile logistics complexity beyond the driver. A ride is a service with two parties, a location, and a price. The Beckn protocol handles that cleanly.

E-commerce is structurally different. Retail requires SKU management across thousands of sellers, category-specific logistics, returns handling, and customer service at scale. These are not protocol problems — they are operations problems. ONDC does not resolve them; it delegates them to whichever buyer app wins the consumer relationship, with no shared infrastructure and no platform absorbing the cost when things go wrong.

The "UPI of e-commerce" framing assumed commerce and payments are structurally similar. A UPI transaction has two parties and a number. An e-commerce transaction has a buyer, a seller, a product, a logistics provider, and a returns policy. That is a different class of problem, and the Beckn protocol, well-designed as it is, cannot make it simpler by being open.

The subsidy cliff

PhonePe Pincode was ONDC's most prominent retail buyer app. At peak, ONDC was directing roughly Rs 3 crore per month in buyer-side subsidies to hold retail volumes up. In late 2024, those subsidies were cut to around Rs 30 lakh per month — approximately a 90% reduction. Monthly retail orders, which had peaked at 6.5 million in October 2024, fell to around 4.3 million by early 2025, a 30-35% decline.

Pincode exited non-food categories by mid-2024, failed at quick commerce, and shut its consumer-facing operation entirely in December 2025. PhonePe now positions Pincode as a B2B tool for offline merchants. This is a significant departure from the national open-commerce buyer app it was meant to be.

The NRAI, representing over 50,000 restaurants, paused new member onboarding onto ONDC and was publicly exploring alternative arrangements. Restaurants that had joined cited erratic fulfilment, customer service gaps, and lower per-order economics than on incumbent platforms.

ONDC's FY25 financials are telling: Rs 33.4 crore in revenue against Rs 180.5 crore in expenses, a net loss of Rs 147 crore, with employee costs up 37% year-on-year. A programme sustaining transaction volume through subsidies while running a 5.4x cost-to-revenue ratio is not yet a business — it is a policy initiative looking for product-market fit.

The seller activation problem

The headline figure is 7.6 lakh sellers onboarded. The transactionally active number (sellers who are live, listing products, and receiving orders) is approximately 1.16 lakh. Roughly 85% of sellers counted as onboarded are not meaningfully active.

The gap traces to how seller apps were compensated during ONDC's growth phase: many received fees per seller onboarded, not per seller who completed a transaction. The result was aggressive sign-up campaigns with no corresponding activation. Sellers were counted, then abandoned because the orders never came.

Fewer than 1,000 sellers received more than 10 orders in 90 days, according to ONDC's own open data. A seller in Bhopal or Coimbatore who joins a national open-commerce network and receives fewer than 10 orders in three months will not maintain their listings. They will continue selling on WhatsApp Business, where the customer already is.

Open protocol, closed liquidity

ONDC is genuinely open at the protocol layer. Any developer can implement Beckn, get certified, and join the network. The specification is public. The registry is open. In that technical sense, the programme delivers on its founding premise.

But actual buyer demand has concentrated in Magicpin, Namma Yatri, and Ola. Sellers can list across the full open network, but their order flow depends on which buyer apps are actively driving traffic. Open supply, closed demand. The concentration is one layer removed from where it was in the incumbent duopoly, but the structural dependency for a small seller has not changed as much as the framing suggests.

The mechanism is straightforward. Open protocols distribute supply cheaply. Any seller app can onboard a merchant at near-zero marginal cost. But demand: customer acquisition, retention, repeat purchase. This requires investment that accrues to whoever owns the consumer relationship. In India's digital commerce market, that is still a small number of well-capitalised apps. ONDC moved the bottleneck; it did not remove it.

The quiet B2B pivot

Less visibly: Uber is exploring B2B logistics via ONDC rails (TechCrunch, May 2025). Aditya Birla Finance, Tata Capital, Bajaj Finserv, and Karnataka Bank are in ONDC's financial services cohort. Insurance pilots are live with Bajaj Allianz, Aditya Birla Health, and Kotak General Insurance.

The B2B and financial services case is structurally different from B2C retail. B2B has no consumer acquisition problem: you are connecting businesses that already know they need each other. Financial product distribution via an open discovery layer addresses real fragmentation in Indian distribution without requiring consumer subsidies to sustain the volume.

ONDC as B2B logistics infrastructure and financial services distribution rail is a different product from "India's Amazon alternative". It is a narrower thesis, less legible as a national digital initiative, and harder to present at policy events. But the structural fit is more honest, and the unit economics (where no consumer is being paid to transact) are potentially more durable.

What to watch for ONDC in 2026

A few things that will tell you where this is actually heading.

Whether any retail buyer app publishes positive per-order economics without subsidy support. Four years into the programme, that number has not been made public. The absence of the disclosure is itself a data point.

Financial services volumes: not signups or pilot announcements, but actual disbursals, policy issuances, and completed transactions. A monthly run-rate of a few hundred crore from the FS cohort by end of 2026 would be a genuine signal that the B2B pivot is substantive.

Beckn Protocol Foundation independence. The open-source protocol sits underneath ONDC, but if the programme were restructured or defunded, it is not clear the protocol ecosystem would survive independently. Namma Yatri demonstrates that Beckn works in production at scale. The institutional continuity question is a separate one.

The open-commerce thesis is not dead. It may simply not be about retail, at scale, in India, funded by the government. An open protocol for B2B logistics and financial services distribution, unglamorous and largely invisible to consumers, could still be the outcome that matters most.