The article argues that ONDC is at a crossroads: it has achieved scale on paper and policy “wins” but is struggling with messy execution, conflicting incentives, and the risk of becoming either a glorified discounting pipe or a stalled public-utility experiment, depending on what it does in the next 12–18 months

Introduction: A Vision Worth Having — An Execution Under Scrutiny

In 2022, India made a bold institutional bet. The Department for Promotion of Industry and Internal Trade (DPIIT) incubated a Section-8 non-profit whose stated purpose was nothing less than the democratisation of digital commerce — the end of the duopoly that Amazon and Flipkart had quietly, relentlessly consolidated. The Open Network for Digital Commerce (ONDC) was to be the UPI of e-commerce: open, interoperable, impossible to capture.

The premise was, and remains, intellectually sound. India’s e-commerce market had concentrated into a duopoly with platform commissions of 25–40%, effectively taxing small merchants for access to their own customers. The idea of a protocol layer — an open, interoperable network — was a legitimate and important policy response.

Four years later, the data renders a more complicated – and in places troubling – picture. ONDC has generated transaction headlines, earned Prime Ministerial commendation, and produced an impressive deck of participation numbers. Beneath that surface, however, lies a set of structural contradictions that no cumulative transaction count can resolve: a subsidy-dependent growth model in secular retreat, a retail segment losing ground when incentives are withdrawn, a governance vacuum at the top, technical infrastructure that has failed to earn builder confidence, and a strategic sprawl that has produced presence in many categories and dominance in none.

This analysis examines those contradictions. A correct diagnosis does not guarantee a functional cure. The question worth posing – calmly and without political inflection – is whether ONDC’s problems are primarily executional or foundational. The answer determines the appropriate response.

I.  The Subsidy Trap: Growth or Dependency?

Begin with the numbers that matter most. ONDC’s reported retail orders peaked at 6.5 million in October 2024. By February 2025 — just four months later — that figure had fallen to 4.6 million, a decline of nearly 30%. Retail’s share of overall ONDC transactions collapsed from 53% in May 2024 to just 31% in February 2025. By April 2025 retail orders stood at approximately 4.3 million — a 35% fall from the October peak.

What happened in between? Incentives were slashed.

ONDC had been operating an elaborate subsidy architecture: up to ₹50 per order reimbursed to buyer apps, ₹75 per order delivery subsidies, and per-seller onboarding payments. At its peak, each network participant could draw ₹3 crore per month in financial assistance. By late 2024 that cap had been cut to ₹30 lakh — a 90% reduction. The market response was immediate and instructive: PhonePe’s Pincode, once positioned as a flagship grocery delivery partner, exited the network entirely in 2024. Paytm quietly removed the ONDC shopping icon from its consumer-facing application. The National Restaurants Association of India reportedly paused new onboarding.

The correlation between incentive reduction and order collapse is not coincidental. A LocalCircles consumer survey found that only 15% of e-commerce users had placed an order on ONDC in FY25, and — critically — 54% of those who tried found the experience more cumbersome than the prior year. When discounts disappear and the product experience does not compensate, users leave.

ONDC has noted, correctly, that these incentives were drawn not from the government’s consolidated fund but from ONDC’s own corpus, raised from 19 institutional investors including SBI, NABARD, and ICICI. This is a meaningful distinction. But it does not dissolve the structural question: public institutions deployed capital to subsidise private customer acquisition costs, distort market signals, and create the illusion of demand-side traction. The network’s FY25 net loss stands at ₹147.13 crore, down from ₹195.6 crore in FY24, with total expenses of ₹180.54 crore against revenue of only ₹33.41 crore. Employee costs alone rose 37% year-on-year to ₹60.42 crore.

Demand seeding is a legitimate strategy, provided it is time-bound, targeted, and accompanied by parallel development of the structural value proposition that sustains demand after the subsidy exits. There is limited public evidence that this parallel work was executed with the rigour the strategy required.

The fundamental question for ONDC’s institutional backers is this: were these incentives an investment in demand seeding, or were they masking the absence of product-market fit? The strategic question for ONDC’s board is not whether losses are narrowing – it is whether the product has demonstrated any organic, incentive-independent value proposition. The data, at this moment, does not support that conclusion.

II.  The Breadth Illusion: Everywhere, Dominant Nowhere

From its earliest phases, ONDC pursued a strikingly broad sectoral agenda: grocery, food delivery, fashion, electronics, mobility, financial services, agriculture, education, healthcare, and more. By 2024, the network was simultaneously entering quick commerce, sachetised insurance, and in-game shopping integrations. The logic appeared to be that maximum surface area would accelerate network effects.

The outcome has been the opposite. Network effects require critical mass in a single use case before they compound across adjacent ones. The most successful digital public infrastructure globally — UPI being the canonical domestic example — succeeded precisely because it solved one problem very well before expanding. UPI solved person-to-person and person-to-merchant payments with singular focus. ONDC has no equivalent vertical dominance to show.

In food delivery, Swiggy and Zomato’s advantages in logistics density, consumer trust, and operational excellence remain unassailed. In grocery, quick commerce platforms — Blinkit, Zepto, Swiggy Instamart — commanded a GMV of ₹33,000 crore by FY25, growing at approximately 40% annually, and controlling two-thirds of all e-grocery orders. Fashion and personal care, categories that anchor conventional e-commerce volumes, remain marginal on the network. Electronics has failed to attract serious large-brand participation. PhonePe’s Pincode — perhaps ONDC’s most visible retail champion — exited non-food categories by mid-2024, failed to gain competitive footing in quick commerce, and shut down its entire consumer-facing operation in December 2025.

The multiplication of verticals has not been accompanied by demonstrable dominance in any single category. The honest question ONDC’s board must confront is whether the decision to pursue simultaneous expansion across ten verticals was driven by strategic logic or by institutional pressure to demonstrate visible momentum to funding stakeholders. Breadth, as a strategy, confuses presence with traction.

III.  A Tale of Two Counts: The Seller Metric Problem

Senior ONDC leadership has consistently cited seller count as a headline achievement. As of April 2025, the network claimed 7.6 lakh sellers and service providers. Prime Minister Modi’s commendation cited 700,000 sellers operating across 600 cities.

Set against this: as of December 2025, the government itself confirmed that 1.16 lakh retail sellers were live on ONDC across 630 cities. That means approximately 85% of ‘onboarded’ sellers were not active in any meaningful transactional sense. This is not a rounding error. It is a structural one.

The incentive architecture that paid seller apps for onboarding – regardless of whether those sellers subsequently transacted – created a predictable outcome: seller apps onboarded aggressively to capture per-seller payments, while genuine marketplace liquidity remained concentrated in a tiny fraction of the registered base. Seller churn is high, catalogue quality is inconsistent, and repeated consumer frustration has damaged platform credibility.

Compounding the problem is where what little “active” seller presence does exist is concentrated. The bulk of any transactionally live sellers on ONDC are not in the core e-commerce verticals — grocery, fashion, or electronics — that the initiative was built to democratise. They are in mobility, logistics, and transportation. Auto-rickshaw drivers, cab aggregators, and delivery agents have been counted within the seller universe, inflating headline figures while contributing nothing to the e-commerce supply-side liquidity that small kirana owners, apparel brands, or electronics retailers would represent. ONDC’s own Network Observability dashboard makes this uncomfortable reality legible: fewer than 1,000 sellers across the entire network received more than 1–10 orders in the preceding 90-day period. The remainder — of a claimed base running into lakhs — either parted ways after onboarding, were never meaningfully integrated in the first place, or are mobility and logistics participants whose inclusion obscures rather than represents e-commerce market development.

Registered seller count and active seller count are distinct measures, and the difference between them determines actual liquidity on the network. Without transparent publication of active seller metrics — segregated by category, transaction frequency, and order value — the registration figures function as vanity metrics rather than liquidity indicators. When leadership presents ‘lakh of sellers’ as evidence of ecosystem vibrancy, they are presenting a denominator without the numerator that gives it meaning. Vanity metrics, when substituted for outcome metrics at the leadership level, are not merely misleading — they actively delay the corrective responses the organisation needs.

IV.  Transaction Integrity: When Buyer and Seller Are the Same Entity

Perhaps the most conceptually fraught issue ONDC has never adequately addressed publicly is this: how does the network count — and characterise — transactions where the buyer-side application and the seller-side application are controlled by the same entity?

Magicpin is simultaneously registered as both a buyer app and a seller app on the ONDC network. Namma Yatri — ONDC’s mobility showcase and one of its fastest-growing ‘success stories’ — functions as both the demand and supply aggregation platform within its rides ecosystem. In mobility particularly, Namma Yatri operates as an inventory seller node, with no meaningful interoperability offered to external buyer apps.

The philosophical architecture of ONDC rests on the premise that separation of buyer apps from seller apps — with any buyer app able to discover any seller app’s inventory — creates genuine market competition and consumer choice. When a single entity controls both sides of the transaction, that architecture collapses into something functionally indistinguishable from a conventional closed platform, except one that is counted as an ‘open network’ transaction for reporting purposes.

If total ONDC transaction counts are substantially composed of same-entity-on-both-sides activity, then the growth narrative being presented to policymakers, to the press, and to the market is measuring something other than what the initiative set out to achieve.

This deserves transparent disclosure and an honest public accounting. A public infrastructure network funded by public sector institutions owes its stakeholders rigorous transparency on this distinction.

V.  The DigiHaat Contradiction: When the Umpire Picks Up the Bat

In 2025, DPIIT — the very ministry that incubates ONDC — launched DigiHaat: a buyer-side application designed to bring artisans, farmers, and small producers onto the digital commerce grid. DigiHaat is not merely a government initiative adjacent to ONDC. It is, by design, ONDC’s own marketplace presence — a buyer app operated directly by the entity that claims to be a neutral protocol infrastructure provider.

The government has set an ambitious target of bringing 10 crore small producers onto DigiHaat. Ride-hailing service Namma Yatri has been integrated into the platform. Digital marketing agencies are being empanelled to run performance marketing and influencer campaigns for DigiHaat products.

The stated rationale is inclusion: reaching rural and underserved seller communities that private buyer apps have not prioritised. This is a genuinely important policy objective. The mechanism chosen to achieve it, however, creates a structural contradiction that ONDC’s leadership has not publicly addressed.

ONDC was established on a foundational principle of open, neutral infrastructure — a protocol layer that neither favours nor operates as a market participant. The moment ONDC becomes a buyer app operator, it is no longer purely infrastructure. It is a competitor to the private buyer apps it is simultaneously trying to attract and retain. This directly affects the incentive calculations of every private buyer app operator deciding whether to invest further in the network.

The governing body should always stay neutral and should not get into the business itself. This is not accidental. It is a deliberate governance choice that preserves the neutrality essential to the long-term credibility of public infrastructure. The architecture of trust in a network economy requires that the standard-setter and the player remain distinct. When the umpire picks up a bat, the game’s legitimacy comes into question regardless of how elegantly the innings are justified.

VI.  The Documentation Problem: Infrastructure You Cannot Build On

ONDC’s technical architecture is built atop the Beckn protocol — a legitimate, well-conceived open specification. However, the execution layer — the documentation, tooling, and API stability that builders actually interact with — has attracted consistent and serious criticism from the developer community.

Integrations are routinely cited as six-to-ten-month efforts. Network participants and system integrators have reported frequent changes to protocol specifications without adequate notice, ambiguous or incomplete documentation requiring reverse-engineering for implementation, and a lack of backward compatibility that invalidates months of prior integration work. Several early-stage startups that invested significant technical resources into ONDC integrations have quietly withdrawn without public announcement.

As on the date of writing this article, ONDC is in the process of rolling out something called the Payment Protection Framework, which replaces the earlier Reconciliation and Settlement Framework (RSF 2.0). The RSF 2.0 had been made mandatory for smaller and newer network participants, many of whom spent three to four months of development efforts to comply, only for ONDC to scrap the whole system in less than a year. What makes this more disconcerting is that ONDC’s own buyer app, Digihaat, never implemented the RSF solution. The same was true for larger players like Paytm and Ola, who were not mandated to follow the same rules. Smaller players — the very ones who had a real chance to demonstrate meaningful innovation and network impact — were forced into costly and unnecessary rebuilds, losing lakhs in sunk costs as a result.

Meanwhile, ONDC has begun introducing new frameworks for an eB2B model under the DigiDukaan brand, as well as a hyperlocal “Inventory Less” discovery model. Both of these initiatives — while conceptually interesting — demand extensive new development cycles, months of engineering work, and fresh compliance efforts. Understandably, few buyer or seller apps are now willing to invest that kind of effort, given the current lack of tangible demand and traction across the ONDC network.

This matters for a reason that deserves to be stated plainly: ONDC is not a product. It is infrastructure. Infrastructure derives its value not from what the operator does, but from what third parties build on top of it. If the cost and uncertainty of building on ONDC exceeds the perceived benefit, builders will seek alternatives — and the logical consequence is the very platform consolidation that ONDC was created to disrupt.

A protocol designed for open commerce must treat backward compatibility and specification stability as first-order design principles. The absence of this discipline reflects an organisation whose protocol decisions are driven by committee negotiation rather than practitioner accountability. Each developer or startup that disengages after a costly, failed integration is a signal to the rest of the ecosystem. Signals accumulate.

VII.  The Buyer App Concentration Problem

In the early framing of ONDC, the buyer app layer was conceived as a site of competition and innovation — a place where dozens of applications would vie to offer the best consumer experience. Instead, ONDC found itself dependent on a small set of VC-funded applications that had their own strategic incentives, unit economics pressures, and product priorities.

The consequences were foreseeable. PhonePe’s Pincode entered ONDC as a major buyer app in April 2023, exited non-food categories by mid-2024, and shut down entirely in December 2025. Paytm pulled back ONDC visibility. Meesho, one of India’s largest platforms for small sellers, has not meaningfully deepened its integration.

What was largely unexplored — and remains so — is an alternative model. The internet’s retail history offers instructive precedent: individual entrepreneurs routinely build storefronts on top of open supply infrastructure, earning margins by combining discovery, curation, and last-mile customer relationships. A model that enabled lakhs of small retailers across the world, local community influencers, or category specialists to operate as micro buyer apps — essentially a dropshipping-and-discovery model built on ONDC’s supply layer — was never seriously developed or incentivised.

The structural consequence is that ONDC’s demand side remains precariously concentrated in a handful of large applications, each of which has demonstrated that when the economics don’t work, they will exit. This is not a problem of implementation alone. It reflects a fundamental gap in strategic imagination at the network design level.

VIII.  Governance, Leadership Churn, and the Question of Institutional Credibility

Between December 2024 and April 2026, ONDC lost four senior executives. The most consequential departure was that of founding CEO Thampy Koshy, who had led the organisation since its inception in February 2022. Chief Business Officer Shireesh Joshi had exited the month prior. The non-executive chairperson and at least one founding independent director also departed within the same six-month window. As of the time of writing, there are various other leaders of Vice President and above levels exiting the organisation or have already left.

These departures span operational, strategic, and board-level functions simultaneously — which typically signals not individual circumstance, but something systemic about the organisation’s direction, culture, or institutional positioning.

Leadership continuity is not a cosmetic concern in an organisation of this nature. ONDC operates at the intersection of protocol governance, stakeholder diplomacy, government relations, and market development. A governance vacuum at the CEO level creates decision paralysis precisely when course-correction decisions are most urgent.

More structurally concerning is the background composition of ONDC’s senior hires — drawn predominantly from the management consulting, payments and financial services sector. India’s DPI experience in payments is genuinely world-class. However, payments, management consulting infrastructure and e-commerce infrastructure are fundamentally different problem spaces. The unit economics, fulfilment complexity, catalogue management challenges, and category-specific dynamics of e-commerce require domain expertise that the payments or consulting world does not automatically confer. The absence of senior hires with deep operational e-commerce backgrounds may be a structural factor contributing to ONDC’s inability to gain meaningful traction in core categories.

There is also an open question about conflicts of interest: when individuals who have direct business interests in how network rules are structured move into roles that shape those rules or have close contacts taking up those roles, governance integrity is at risk, regardless of individual intentions. Regulatory capture does not require malfeasance. It requires only structural proximity. The government has a responsibility — to the institutions that capitalised ONDC, to the small sellers whose livelihoods depend on its success, and to the public — to ensure ONDC’s governance reflects independence, domain competence, and demonstrable accountability.

IX.  The Market That Moved On

Perhaps the most uncomfortable structural reality for ONDC is not what it has failed to do, but what the market has done while it was deliberating.

Quick commerce — defined by 10–30 minute delivery windows enabled by proximity warehousing — has emerged as the dominant growth vector in Indian consumer e-commerce. Blinkit, Zepto, and Swiggy Instamart are collectively processing millions of daily orders through proprietary dark-store infrastructure that a protocol layer cannot replicate without equivalent physical investment. Blinkit commands over 50% of the quick commerce market. The category operated at a GMV of ₹33,000 crore in FY25, growing at 40% annually. ONDC is now belatedly attempting to enter quick commerce, but the window for establishing protocol-layer primacy in that category may have already closed.

ONDC’s model — hyperlocal delivery from kirana, electronics, and stationery stores, managed through buyer-seller app intermediation — is a structurally slower, logistically more complex alternative to the dark-store model that consumers have already shown a preference for. Pursuing scale through a mechanism that cannot match the speed, reliability, or assortment depth of quick commerce incumbents is not an ambition. It is misaligned with observable consumer behaviour.

Meanwhile, Amazon and Flipkart collectively still command 83% of the Indian e-commerce market, with ONDC’s penetration remaining at approximately <1%. After four years of operation and hundreds of crores in institutional capital, that statistic should anchor every board meeting.

The self-reinforcing stagnation loop is difficult to break: without liquidity on the supply side, no amount of buyer-side development creates durable demand. Without durable demand, supply-side investment rationale does not improve.

X.  Two Rational Paths Forward

The evidence does not support preserving ONDC in its current form, with its current leadership structure, executing its current strategy. That would be the most expensive option available — not in rupees alone, but in opportunity cost, institutional credibility, and the continued distortion of the private market it was meant to enable.

Two rational paths deserve serious consideration.

Path One: Structural Reconstruction

ONDC retains its institutional mandate but undergoes a fundamental reset. This requires:

• New leadership drawn exclusively from practitioners with deep e-commerce domain experience — not payments, not government consulting — insulated from conflicts of interest with any network participant
• A concentrated vertical strategy rather than the current multi-sector sprawl; establish genuine dominance in one category before expansion
• Protocol governance modelled on organisations like the IETF — open, transparent, versioned, and stability-committed, with backwards compatibility as a first-order principle
• A buyer-app strategy that explicitly enables small, individual, and community operators rather than depending on venture-funded aggregators with misaligned incentives
• Relinquishment or transfer of the DigiHaat operator role to an independent body
• Full transparency on transaction counting methodology, particularly with respect to same-entity buyer-seller configurations

Path Two: Principled Exit to the Open Standard

If the political will for structural reconstruction does not exist, there is a second rational choice: ONDC exits as an operating organisation and the Beckn Protocol — the underlying open standard — is returned to independent stewardship through the Beckn Foundation or an equivalent neutral body. This would preserve the genuine intellectual contribution of the ONDC experiment — a working, field-tested protocol for interoperable commerce — without continuing to channel institutional capital into a governance structure that has demonstrated material limitations.

Private participants who believe in the protocol’s potential can build on it without the overhang of a government body that simultaneously sets rules and plays the game. The government’s role in digital public infrastructure should be to set standards, enforce interoperability mandates, and prevent monopolistic consolidation through regulation — not to operate within the market it is trying to shape.

Conclusion: The Principle That Must Not Be Compromised

The government’s role is to enable markets — not operate within them

There is an argument — made with some force by ONDC’s defenders — that the initiative is still young, that network effects take time, and that the UPI comparison is unfair because payments are structurally simpler than commerce. These are not unreasonable points in isolation.

But they do not answer the structural questions raised here. They do not explain why retail orders collapsed when subsidies were withdrawn. They do not explain why >75% of ‘onboarded’ sellers are not transactionally active. They do not explain the DigiHaat participation. They do not explain senior executive departures on a regular basis. They do not address the transaction counting methodology. And they do not explain why four years and hundreds of crores in institutional capital have not moved the market share needle beyond 1%.

The government of India has created extraordinary digital public infrastructure — Aadhaar, UPI, GSTN — by defining clear problems, building minimal viable protocols, and then stepping back to let the market do what markets do. The lesson from each of those successes is consistent: the state’s highest contribution is to enable, not to operate.

ONDC was conceived as India’s answer to platform monopoly in digital commerce. The intent was right. The ambition was appropriate. The execution has been, by the evidence available, insufficient to the task — and in some dimensions, actively counterproductive. India’s small sellers, regional brands, and digitally excluded producers deserve infrastructure that actually works — not one that works on paper, works in pilots, or works while the subsidies last.

ONDC must decide, with clarity and urgency, which side of that line it stands on. The current answer — that it stands on both simultaneously — is the one answer the market has consistently, and correctly, refused to reward.

This analysis draws on published financial disclosures, government statements, industry surveys, and business press coverage through April 2026. The observations and conclusions represent an independent analytical assessment. The author has no financial relationship with any party named or referenced.