1 Billion Transactions

Q&A with Arif Khan, Chief Digital Officer, National Payments Corporation of India.

Hosted by Santanu Paul, CEO and Co-Founder, TalentSprint

Santanu Paul: I think it is interesting that NPCI, or the National Payments Corporation of India, where you are the CDO, is such a well-kept secret. Most people don't realize how close this company is to their lives. Most of them are using NPCI products everyday but may not even realize it. Tell us more about NPCI and why this company is so important.

Arif Khan: Well, I say at home that NPCI is something that you probably feel, but you never see. And that is the role we play as a financial railroad in the country. Just to give an example, every time you go to an ATM and withdraw cash in India, it is running through our system. We run something called NFS which is the ATM network of the country and has all banks participating. That is just one example. A lot of the audience is using apps like Google Pay, Amazon Pay or PhonePe. Most of them actually run on our railroad called UPI or BHIM UPI. So whenever you are doing a person-to-person transfer or you're scanning a QR code or you're making a collect payment, it's running on that. Not only that, if you're issuing cheques, there is something called CTS behind it which is again run by NPCI. On your cars, and especially if you have bought a car recently, there is something called RFID, it's a FASTag sticker. This lets you pass through the tollgate without stopping. So that technology called ETC is run by us. NPCI, just to give a background, is a unique combination. Its owned by banks. We have 56 banks who are the parent owners of the organization. We are a not-for-profit company. So we don't return profits back to the organizations that have invested in us. People in India talk very often that banks are not very forward-looking and they do not make enough investments in innovation. I think we are a perfect example that shows otherwise. You talk of banks investing, not expecting a return, but hopefully now with whatever we are doing in digital they are getting back enough. So that's a bit about NPCI. As we speak, we work with more than 1100 banks in the country and we are owned by 56 banks and let's see how the stakeholders do.

Santanu Paul: Perfect. So as the Chief Digital Officer of NPCI, as a CDO, help people understand what a CDO really does. What does your day job look like? How stressful is it? How interesting is it?

Arif Khan: This is something that even I go back home and wonder (laughs). I keep coming up with answers for myself. One of my key jobs is to run the transformation initiatives within the organization. By transform, I mean what we have done at NPCI, what we were doing now, and where we want to go. We have aligned the organisation to what we call RUN and GROW and TRANSFORM. So I play a larger role on the grow side. Grow means we have some existing products and we want to build on top of that; some new capabilities and use cases.We want to change the technology stack over there, make it more open, scalable and all the fun things. And the other side is to transform. Transform largely has to do with more innovation to reach different segments. How can we make different payment or product flows and so on and so forth. Having said that, I focus mostly on technology and new innovation products and also run the international business for NPCI. With more and more Indians spreading across the world and more and more Indians aspiring to travel and have holidays and honeymoons, wherever they go they take payments along with them. I follow honeymooners and figure out how they spend at a destination or on location and try to decode whether we can be part of the payment process.

Santanu Paul: Excellent. That's a great example. So the whole purpose of today's conversation is to talk about One Billion Transactions. I emphasize the word Billion, which starts with a B and not an M. I remember, right after demonetization, the first time UPI came out, it did 1 Million transactions in the month of December 2016. Fast forward to today. I am told that September 2019 is going to be a very interesting month because UPI is supposedly going to breach the 1 Billion Transactions per month barrier. That's just huge. I suppose in the future it's going to get to 1 Billion per day. So that frames the whole conversation of a billion transactions. The audience would like to hear about the killer products that NPCI is building. UPI is obviously one of them. Tell us about the UPI journey and why it has taken the country by storm.

Arif Khan: I was just thinking about the UPI numbers you mentioned. Incidentally, during that time I was not with NPCI. I was on the other side. I was with a start-up which was using NPCI railroads and creating payment products. But yes, the UPI journey has been quite phenomenal. As we speak, this month will be close to a billion transactions. If you go back three or four years, we already had a great product in the form of IMPS which was doing millions of transactions. But we realized that IMPS was essentially a credit transaction where you could push money; it was used for remittance and it was doing well. What we needed to supercharge the Indian payment ecosystem is a product which could do both pull-push and debit-credit. It needed to be a product native to India and not something that we cut-paste from overseas. That's when UPI came into being. It is completely open and allows for banks to operate on both sides. It's actually the only true four-party model that exists across the globe. By design, it allows the banks to be there, so that security and trust comes into play. It's regulated by RBI, and it also allows for a lot of Fintech players to play on the edge. Google Pay, PhonePe, Startup Aggregators, and PSUs have now come into play. What that has done is, it's now gone beyond the conventional bank’s reach. Banks had a certain number of distribution points but this has enabled them to go to larger audiences deeper inside India, into different segments of India, and create some very classy experiences for consumers, and the consumers have adopted it. So I think that's the killer in the four-party model. The layering that we have done in UPI, keeping it very open and allowing innovation has helped people create different use cases. For example, whenever my son goes to a store, earlier I had to either hand him cash or a card. Now he can just go and shop and simply give my VPA ID. I get a collect request which says hey, there is a 1000 rupee request or a 500 rupee request, do you want to authorize it? I believe I have more control now. In a nutshell, UPI has opened things up and the open railroad has allowed people to innovate freely. I think that's the catalyst in growing to a billion transactions.

Santanu Paul: The ultimate use case. Kids collecting money from parents by sending UPI collect requests to their virtual payment addresses! What a great use case indeed (laughs).

Arif Khan: True. There are multiple other avenues that we have opened up. Let us get a bit more into it. If you have been following SEBI IPO guidelines, then you know that IPO ASBA was a very cumbersome process.Your money gets blocked, then there is a whole lot of reconciliation. Now, SEBI has almost mandated UPI to be the payment method of choice. It's also helping retail investors. As an example, if I am buying into a mutual fund and I do a transaction at 2:30 PM in the afternoon, and 2:00 PM is the NAV cut off time, then my money is debited today but I get my portfolio credited the next day. So my question is why am I losing today's interest? Why can't you just lock my money and take the money out the next day? After all, you're guaranteed the money and my account has been locked for that. So, UPI has opened up a lot of use cases. We believe we will do more. Hopefully sooner than later, we will have standing instructions and recurring payments also. I mean that's something the regulator is considering.

Santanu Paul: Clearly, UPI is a massive universe that's opening up as we speak. Let's now turn to a couple of other NPCI products. While UPI is the poster child of what made NPCI the face of digital payments, there is also momentum around RuPay. Rupay is a different kind of use case, because we are coming from behind, trying to build a swadeshi card to compete with the MasterCard and VISA schemes. And by all indication, it's really working. So talk about the role of RuPay and how that's important as a debit or credit card.

Arif Khan: Cards have never been for the masses in India. At least not been looked at that way. It's always been an aspirational product whether it's a credit or debit card. Before NPCI stepped into the scene, there were other networks and at best there were 30-40 banks issuing cards. One thing that RuPay has done is now you have ten times more banks issuing cards. The Jan Dhan Yojana, that our Prime Minister announced, had every family opening a bank account backed by a RuPay card. So RuPay right now is a very debit heavy scheme. We have 600 million debit cards already in the market, but we also have a credit card program that is running. We've done a few million cards over there, seen some good results on credit as well as prepaid, so we will expand it. What RuPay has done is enabled an average Indian to actually carry plastic in his wallet. Yes, the activation rate probably has not yet reached the level that we wanted, but it's like a sleeping asset. It's like your computer and imagine that I've already got a piece of software sitting over there waiting for one day when you will start interacting with it. So as India becomes more and more digital, RuPay will only gain strength with the installed base of 600 million. That's one side of the story. What we also realize is that these geo corridors that India has, like with Singapore we are a favoured nation, with countries like Bhutan which are our neighbors, or with the Middle East and UAE with a lot of people of Indian origin. There's a lot of travel and tourism going both ways and there is a bunch of 8-12 countries that we have identified to see how can we lay our railroad over there. For example, most Indians who go to Singapore like to go shopping at Mustafa, like to go to Duty Free. Can we enable those transactions? On one side, we have tied up with international schemes like Discover, Diners, and JCB. In fact, China UnionPay in India runs on our railroad. On the other side, what we are seeing is these are cards for a certain segment issued by certain banks. What about the larger domestic cards we have? How can we open them up? So by doing these bilateral tie ups, we are enabling these cards too. And you need to look at it from two sides; one is the card getting used there. The other thing is it also establishes a remittance corridor. For example, if I am not mistaken, Singapore to India annual remittances are close to a billion dollars. Typically, remittance has been a very painful thing. It’s high cost. The turnaround time is not very efficient. If it gets lost somewhere it's very difficult to trace it. I think we are trying to establish a standard where we could work with these three-four corridors and make it very seamless. Then imagine you are sitting in Singapore with your mobile, or you walk into an ATM, and just do a seamless transaction.

Santanu Paul: China has shown through China UnionPay that they can really go up against the global biggies like MasterCard/VISA. I think China UnionPay and JCB are good success stories. You also talked about a partnership with JCB. Do you foresee a day where RuPay is internationally accepted, at par with MasterCard, VISA, CUP, JCB kind of brands?

Arif Khan: Clearly, from an India market perspective, RuPay is already there and higher than some of these guys that you mentioned. If you look at merchant spends, we are clearly number two in India. For ATMs, we are 99.9% of the market. We certainly hope to one day reach the scale of China UnionPay but it's going to be a little more calibrated. What we need to understand from a NPCI context is RuPay and UPI will go hand in hand. They're two different payment methods that will grow. So when I spoke about RuPay and these corridors, we basically look at UPI also in those corridors. In China, AliPay and WePay are dominant, but they are not exactly interoperable. The Indian approach is to run an interoperable solution for cards as well as UPI. So we will gather our momentum in a different way. In our journey, a lot of Fintechs and startups will start playing a more pivotal role compared to what happened in China, where it's very tightly controlled. I think what we are doing is building an enabling railroad to allow many partners and innovators to come in.

Santanu Paul: That's the defining UPI principle. Let's get to a third product that I want to cover. What's happening with the national electronic toll collection system is exciting. We see signs of FASTag at highway toll booths. Could you talk about that and where does this take NPCI?

Arif Khan: The entire manual and cash-based electronic toll collection system is not only a waste of your time and effort, but also look at the fuel that is getting burned every time you are waiting five minutes at the toll booth. There is also the aggravation of not having correct change. Obviously, the grand idea is that you can just seamlessly pass through and help make it a very smooth flow of traffic. As we speak, more than 750 national highways have adopted it. The state highways are now beginning to come onboard, and by the end of this financial year a lot of states would come into play. Right now, the uptick is for commercial vehicles, followed by passenger vehicles. The way we designed it allows for new experiments and use cases. I think it will extend to parking. There are new government diktats for heavy traffic violation fines. When it comes to parking, most of us are offenders. Finding a parking spot in a city like Mumbai, Chennai, or Hyderabad is difficult. For example, every second Sunday I visit one of the hypermalls with my family. You can imagine me completely burned out after a hectic week, it is 10:30 AM on a Sunday morning, and I am driving with my kids. They are pretty excited. I'm not excited because when we finally get there, finding a parking spot is so difficult. Now imagine this, I have an GPS-based app that tells me beforehand how to navigate the parking lot and takes me directly to an open spot. Annoying problem solved!

Santanu Paul: We have talked of three NPCI products. If we had time, we could have picked many more exciting products to talk about. But these three products are representative of where NPCI is coming from, where it is today, and where it is going. In a sense, it covers the entire spectrum. Let's now turn to deep tech and the impact of new disruptive technologies, Artificial Intelligence, Machine Learning, Blockchain, Cybersecurity, IoT, and so forth. As NPCI builds out citizen scale operations across the country and internationally, what is the role of AI and Machine Learning in solving hard problems?

Arif Khan: We are, at the end of the day, a technology platform entity. So technology is at the heart of everything for us. We want to build platforms because we want to be an enabler of citizen scale payments. It’s a bit like Facebook. They are a platform and there are people building on top that. I believe that's where NPCI should get to. So clearly we need to worry about how we scale, how we make our platforms robust, how to be open. These are the key design elements that we keep reviewing and refreshing. AI and Machine Learning obviously is extremely important because we work with multiple banks. Each bank is at a different stage of their journey on how they handle fraud management. So we felt there is a need for centralized risk management so that our end consumers sleep well at night. AI plays a key role here because what it allows us to do is to learn from the mountains of transactional data that we gather and have with us. It allows us to start detecting patterns so we can determine in real time if there is a fraud pattern here, or something needs tight monitoring there. What have started a couple of AI models on scoring. It is really helping us.

Santanu Paul: Are you scoring transactions or consumers?

Arif Khan: Actually both. Right now it is at the level of transactions. The idea is because we have multiple products, it can be aggregated at other levels too, say device, geography, or segment levels. We are trying to roll up to a consumer level but obviously it has its own challenges. Right now what we're doing is taking the model score and supplying it is as an alternate score to participating banks. Many of them have their own scoring engines. They combine our score and theirs and a decision is taken. Whatever happens subsequently, whether it was right or wrong, is fed back into the model. So we keep learning. There are also smaller banks that cannot afford such tech investments and we run on this stuff on their behalf.

Santanu Paul: The way I see it, when you get to a billion transactions a day, even if 1 transaction among 100,000 transactions is fraudulent, that will mean 10,000 transactions a day which are fraudulent. And you don't want to be stopping a transaction to manually inspect whether it is fraudulent or not, you want to detect it on the fly and prevent it on the fly as best as possible. So this whole thing seems to be ripe for AI and Machine Learning. In fact, your models have to get extremely good at saying whether it is a good or bad transaction with a high degree of certainty. It should not be that good transactions are getting stopped and bad transactions are going through. That would be terrible for customer experience as well as the balance sheet. Right?

Arif Khan: There is always that fine balance between risk and opportunity. You need to think about all this from a customer experience point of view. Imagine you as a customer, especially a new customer, or any customer for that matter, who is carrying out digital payments. You mess up their hard-earned money, and for sure that customer is not coming back. Then imagine the flip side of the coin where you know you have enough money in your bank account, or you have enough credit worthiness, and for some odd reason the engine has blocked that transaction. And that is typical in our country. Most banks decide there is a velocity, or they say that in a day they will allow upto two lakhs or whatever. But the question I keep asking myself is how are you judging their limit? If they are high on creditworthiness then why are you saying that they should only spend this much in a day. So that's where the entire data profile about what is the category, where are they spending, how are they spending, that entire learning becomes important. I don't see any better way than using AI to solve this problem. We need to come up with different models for different segments. That's the only way we can balance the risk versus opportunity.

Santanu Paul: Absolutely. All good things in life involve trade-offs. So let's talk about the other big favorite technology that everybody is intrigued by, confounded by, and talks about all the time, namely Blockchain and Distributed Ledger Technologies (DLT). 80% of use cases that are talked about in Blockchain usually relate to Fintech. In other words, FinTech is THE domain of Blockchain. How does NPCI plan to use it, and how does it relate to end customers?

Arif Khan: Blockchain is the domain of the bold, I guess (laughs). Wherever there is an exchange of money there are two parties, there are banks, there's money coming in and money going out. So banks are always in a situation of a plus and a minus, owing each other money. The robustness of the ecosystem is built around the belief that there is a guarantee of settlement, that the owed monies will be exchanged at the right time in the right way. That's the crucial thing for the ecosystem to sustain itself. Now making sure that credit liquidity is there, and the cash supply is there for that, requires a lot of funding. I mean it's cash lying idle in one way or another. The simple way to look at it is, whenever cash is locked up, it is losing value, or there is a cost to it. So one thing we're trying to solve is how to reduce settlement time, because that is typically T+1 day or more. Until that time your positions are open. What a Distributor Ledger does, by the way it works, is that all information is available at all nodes, so the net-off settlement could happen very locally, then and there. This is the way to free up liquidity and gain efficiency. In an emerging country like India, the key thing is to bring down the cost of payments. You can't push people towards digital and get them to pay for it. You really need to make the cost as minimal as possible. Also, it helps increase the robustness of the ecosystem. Imagine for some odd reason one bank going kaboom. This technology would help us detect those faster and contain the damage to the overall ecosystem.

The second is, and I keep saying this again and again, a payment system is as good as the trust users have in it. I am sure all of you have faced challenges with refunds. You buy a movie ticket online or buy groceries, you buy something on one of these ecommerce sites, you buy through a card, UPI or whatever, and you have a transaction failure. You do not get the product, you cannot consume the service, but you are out of pocket with your money. If we want the next 300 million Indian consumers adopt digital, and this is their first or second experience, they are not going to talk digital again.

What we need is proper lifecycle management, whether it's a refund or a dispute or a chargeback. Let's take an example of chargeback. You are the consumer who complains to your bank, your bank complains through the network to the other bank, the other bank reaches out to the merchant, collects some information, and push it back into the network. The entire journey could take two days, three days, four days and it goes out of hand. But if all the information is centrally available, the first bank by itself could solve it within minutes. And if they can automate it, it can become real time. 90% of disputes and chargebacks, as per the study we did, could get resolved in less than fifteen minutes. This can inject great confidence in the ecosystem from a consumer point of view. Banks benefit immensely too. The quantum of money and the operational staff managing reconciliation is usually massive. In spite of all the automation, a good 10 to 12% of all transactions go into manual reconciliation where someone has to judge and try to course correct it. And course correction always hurts that one odd unlucky customer. So banks have a great upside too!

Santanu Paul: I think you made a couple of very interesting points that DLT technology will actually help lower friction, lower cost, and simultaneously increase overall system hygiene and transparency so we actually get to know where the bad apples or high risk actors in the payment ecosystem might be, or where are they going to come from. That's like an early warning system.

Arif Khan: What we are trying to do now is a POC. With our volumes it is very important that we test for scalability. One of the benchmarks that I have given to the team is to try for a bare minimum of 10,000 TPS. I don't think there are many use cases that exist in the real world where this is true. We want to thoroughly test the technology also. And obviously, the business case too. While we can reduce the opex, what does it do to the overall capex? We have to balance the economics.

Santanu Paul: So one thing that intrigues me quite a bit is that people say India is a very interesting country because the head lives in the 21st century and the tail lives in the 19th century. We have sophisticated users who are doing 5-7 digital payment transactions a day, and then there are people who have never done one. They might be doing their first one today or tomorrow. You talked about the fact that a first or second time user having a bad experience is actually a lost cause and it hurts the cause of inclusion. How do we accelerate digital payment inclusion and bring more and more people into the system quickly? Because the whole aspiration of a less cash society lies in bringing people who are today using cash into a world of digital payments. What's NPCI role in inclusion and bringing more people onboard?

Arif Khan: At one level, NPCI is about cutting edge innovation and staying ahead of the curve. But clearly India also needs someone to reach out to the unorganized sector, which is 70% of our economy. We have in UPI and AEPS two very powerful solutions for inclusion. But there is so much more to be done. If you are familiar with Bombay and you go to the junction of ChurchGate and VT, there are these guys who sell books on different genres and topics. They actually run a book rental also, and they recognize you. You can buy a book, you can come back after one year, they have codes and all those things. Now my question is that how do we make them digital? Can I give them a very simple QR code to accept the payment? I think those are the use cases that we need to really solve for. You go to Byculla or Bandra East and there is a taxi man colony. Those cab drivers need to send their money back to their hometowns, which maybe in Uttar Pradesh, in Muzaffarnagar or wherever. How do they transmit the money? Are they going to someone? I heard that for every two or three thousand rupees they transfer, they pay around three hundred rupees. Can we not find a digital way to do that? In fact, our innovation teams are now trying to solve for these things.

Santanu Paul: You have told me the story earlier that one of your favorite use cases is how to get vada pav sellers in small towns accept digital payments. Have you solved that yet?

Arif Khan: We may think we have, but the merchant doesn't believe so (laughs). There's a place called Karjat not very far from Bombay. If you stop at the main Karjat junction and ask anyone "Vada pav kaha milega?", they will say "Satyu vada pav jao". You go there and it's behind a small house. I think he has got eight or nine people working for him. The average customer is buying five or six vada pav, and my guess is this guy does sales of about 2000-3000 rupees per day. It's not only the customer paying him 15 rupees for a vada pav, he's paying his workers, his helpers, and his suppliers. And they in turn may be sending money back home. How do we help them with digital payments? Certainly not by giving them an app that operates only in English! We have so many languages, Marathi, Tamil and so forth. For example, our BHIM app supports more than nine languages already. We want to make sure that in the coming months we go to twenty languages. When you look at India from a demographic point of view, it's like a million communities and a thousand needs. How do we cater to these microsegments? Give them the right language options? Can we not have something very intuitive based on voice or gesture? Of course there will be a question whether they have handsets, but we will let someone else solve that problem.

Santanu Paul: You were referring to startups along the way and I know that the startup ecosystem wants to participate in the NPCI platform story. And I know that you are doing a lot of work trying to create an interoperable platform with and for Fintech startups. So tell us a little bit about the kind of initiatives that you have started to run to help Fintech startups come on board and become part of the bigger success story of NPCI and India.

Arif Khan: Sure. But let me admit we should do a lot more for startups than we have done so far. We are stepping on the pedal now. We have created an API accelerator. What we realize is that there are a lot of startups who want to try to do a lot of different use cases, they have a good understanding of UPI, they have enough material, so we are giving them the API. We are also offering offline consulting to them, helping them to create the right product. In fact, not only is this an API accelerator, it may even come with some co-working space. So imagine, there is this bunch of very young guys (in fact, this actually happened to me in a very remote corner of India), smart dudes that understand the technology very well. But they don't have capital and they don't have a place at home to really work. They are staying in small places. So what we're saying is let us have a physical space, nothing very fancy, where some of our guys are available to help creative innovators and entrepreneurs co-create with us. In parallel, we have started running theme based hackathons, to try and solve for genuinely interesting problems. For example, can we have that app that runs purely based on gestures? Or how to make it easy for a rural person in a village to manage their pin number which is in English? A whole lot of such efforts are underway. I think you will hear a lot from us in the coming months. I think by this financial year end we should be there.

Santanu Paul: Fantastic. Thank you so much Arif for such a wide ranging and candid interview.

Watch the entire interview here https://www.youtube.com/watch?v=Bs93R57t2bw