The ONDC conundrum: Where protocols win… and where they don’t…

India’s ONDC (Open Network for Digital Commerce) has been the most ambitious decentralized alternative to the centralized e-commerce marketplace model pioneered by Amazon and others.

Yet, it has (thus far) struggled to fully deliver on its promise to reorganise the commerce ecosystem through a protocol-based approach.

To many, this is even more confounding when compared with the massive success of UPI, another seemingly similar protocol for orchestrating payments, which has transformed financial services (to the extent of digitizing roadside begging) across India.

The key to unlocking ONDC’s potential may well lie in unpacking the economics of protocols and open networks.

This is an ambitious, far-sweeping analysis… and there’s a lot to unpack here… a lot that runs counter to current conventional wisdom on the economics of open networks.

This analysis is not a comment on the commendable efforts behind an ambitious initiative like ONDC as much as it is on the erroneous ways in which media, investors, and analysts interpret the mechanics at play.

There are three parts to this deep-dive:

1. The fallacy of the unbundling + interoperability argument
2. Factors that determine whether protocols win or lose
3. Determining which players win and which ones lose in a protocol’s ecosystem

Let’s give this a spin!

 

A quick primer on ONDC

If you know about ONDC, feel free to move on to the next section.

But if you don’t, a picture is worth a thousand (or 200) words… so I’ll skip those 200 words and drop this picture here instead:

Source

Essentially, ONDC provides a protocol-based organization of an open network of buyer and seller side interfaces and services instead of one centralized platform that bundles all those services together.

Instead of one single platform managing buyers and sellers, buyer-side agents can onboard buyers, seller-side agents can onboard sellers, and the protocol can settle transactions.

Protocols = Unbundling + Interoperability?

At the heart of protocol economics lies unbundling. I’ve written about this before in Unbundling the Unbundlers:

Marketplaces bundled demand-side onboarding, decision support systems, and tools with supply-side onboarding, decision support systems, and tools. Protocols unbundle the two.

For instance, a marketplace like Ebay bundles seller onboarding, seller analytics, buyer onboarding, buyer decision support, search functionalities, and exchange infrastructure. All these components will be unbundled.

As a result, we move away from central market-maker architectures to decentralised agent-based architectures, all coordinated by a shared, common protocol.

All of those points broadly hold true.

In fact, this unbundling playbook played out to perfection for India’s protocol economy poster-child, the Unified Payments Interface (UPI). ONDC is often compared and contrasted with UPI so it’s worth getting to know UPI as well.

UPI provides protocol-based orchestration for inter-bank peer-to-peer transactions, using a payments identifier (UPI ID) linked to your mobile phone number. UPI is a great example of unbundling (of the custodianship of money linked to an account from the payments linked to an ID) and interoperability (using a unified identifier that works across banks).

The logical fallacy of the unbundling argument

Analysts and pundits explaining value creation with ONDC rely on the following core argument:

ONDC creates value through a combination of unbundling and interoperability.

The proposed path to success is:

Step 1: Unbundling (of the commerce value chain)

Step 2: Interoperability (as specialized players emerge)

Step 3: Success (as interoperable players deliver the end-to-end value chain)

There are three fundamental flaws with this argument.

And these three factors explain why unbundling works in favour of UPI, but almost works against ONDC.

To make all this a tad more dramatic, I’d like to now propose the three laws of protocol economics.

The first law of protocol economics

First, unbundling – in general – increases coordination costs.

This brings us to the first law of protocol economics:

A protocol creates value through unbundling to the extent that it can absorb and resolve the coordination costs originating from unbundling.

Unbundling and interoperability do not magically lead to success.

In fact, unbundling + interoperability can create a nightmare if the coordination costs aren’y sufficiently resolved.

The lego comic picture below – churned out by Midjourney – illustrates what such a coordination nightmare would look like.

This is why protocols are not always the right solution.

The second law of protocol economics

What, then, determines how high these coordination costs are?

This brings us to my second law of protocol economics:

The more the number of steps in the value chain you unbundle, the higher the coordination costs.

In the case of UPI, unbundling is valuable because it’s largely occurring across two layers only: the sending bank and the receiving bank. The coordination costs are minimal and are sufficiently resolved by the protocol.

However, ONDC unbundles the commerce value chain across many more layers (seller onboarding, inventory aggregation, search, fulfilment, returns, dispute etc.).

In this case, unbundling increases coordination costs to an extent that can’t be resolved by the protocol alone.

Coordinating across seven specialized layers of the value chain is orders of magnitude more complex than coordinating across only two layers (as with UPI).

This coordination failure translates into a poor consumer experience. The buyer may effectively find inventory but may have a poor fulfilment experience. The fulfilment may work well but the returns may not. The returns may work well but a dispute raised may not be sufficiently resolved.

With a centralized platform, a single entity coordinating across these steps successfully absorbs the coordination costs. With a protocol, every additional layer unbundled creates a non-linear increase in coordination costs.

The third law of protocol economics

There’s a final reason that ONDC struggles where UPI is successful.

And with that, the third law:

A protocol’s ability to absorb coordination costs is inversely proportional to the degree of variability in use cases it needs to directly coordinate.

Payments is horizontal and has a narrow degree of variability. The payment flow for buying a toothbrush is similar to the payment flow for buying plumbing services. The commerce flow for the two are vastly different.

The use case that UPI supports in payments is very well suited to protocol-based mediation. This is also why even in platforms, it is so difficult to unseat payments players, while vertical competitors emerge all the time to compete with Ebay and Amazon.

The criterion ‘directly coordinate’ is crucial here. Yes, payments are embedded across a wide variety of use cases, but the payments flow itself is relatively narrow with low variability and high standardization.

Commerce extends across a variety of varied use cases. Vertical use cases require specialized mediation. This is why vertical marketplaces repeatedly come up and compete with Ebay and Amazon, despite their horizontal dominance.

This combination of vertical specificity (high variability) and high coordination costs makes ONDC especially poorly suited as a protocol-only solution to commerce.

The protocol effectiveness spectrum

Bringing all this together, we realize that protocol effectiveness may lie across a spectrum – from weak to strong.

Here, I’m specifically referring to a protocol’s effectiveness at coordinating market activity.

The lower the variability in use cases and the lower the coordination costs, the stronger the protocol’s effectiveness at mediation.

The higher the variability in use cases and the higher the coordination costs, the weaker the protocol’s effectiveness at mediation.

 

This explains why ONDC struggles while UPI wins.

UPI sits at the ‘strong’ end of this spectrum.

ONDC, conversely, is stuck at the weak end of the spectrum.

Solving for protocol effectiveness

Well, how do we solve for protocol effectiveness?

You can either work to reduce coordination costs…

Or reduce the variability in use cases supported.

Moving along both axes helps us ascend the protocol effectiveness spectrum.

 

Let’s look at both options in turn as potential solutions to the ONDC conundrum.