This article first appeared in Digital Edge, The Edge Malaysia Weekly on March 29, 2021 - April 4, 2021

Malaysia’s general election in May 2018 was seen as a close one and Dylan Tan wanted to play his part. Malaysians all around the world were trying to come home to vote, but some did not have the financial means to travel back.

So Tan publicly tweeted his credit card details and told those living abroad to buy plane tickets home, maxing out his RM20,000 credit limit. For those who did use his card, he told them that they could pay him back over the next three months. 

“Vish and I had an agreement. I was going to max out my credit card, but if we lost the money, we were going to share the cost,” says Tan, referring to his friend and now business partner Vishvesh Suriyanarayanan.

Amazingly, Tan saw almost all of his money returned. This trust exercise showed that a little trust goes a long way with Malaysians, and sometimes, all that is needed is a bit of empathy when dealing with financial difficulties.

Taking this experience into consideration, the pair co-founded and developed Malaysia’s buy-now-pay-later (BNPL) fintech platform, Split. The platform was initially focused on flight purchases and other travel services, but when the pandemic hit in 2020, the company needed other revenue streams and pivoted to offer e-commerce payments.

“It was already at the back of our minds when we started that we needed to venture into the e-commerce space eventually so we kept that in mind when we designed the back-end system. When the pandemic hit, it worked in our favour because we took less than a week to tweak the system to accommodate the shift to e-commerce,” says Vishvesh.

So, how does Split work? It’s simple. At checkout, a customer will be given the option to pay with Split. Split will then pay the store the full amount upfront while the customer pays Split back in up to three monthly interest-free instalments by linking their debit or credit card or via an FPX bank transfer. At bricks-and-mortar stores, a QR code will be displayed for payment with Split.

Since then, Split’s customer count has grown exponentially, with the company taking on payment orders of between RM100 and RM2,000 on behalf of more than 350 merchants nationwide. Split is also supported by all major e-commerce platforms and offers application programming interface (API) integrations as well. Even smaller businesses, such as those that operate on Instagram, can sign up as a merchant with Split. The duo’s goal now is to get more people on board.

As the BNPL industry is new, it is still unregulated in Malaysia but Tan says they have been continually engaging with Bank Negara Malaysia and other regulatory bodies to ensure transparency. 

“We don’t lend money to consumers, nor do we charge them any late fees and interest. We’re just providing a different way for consumers to make their purchases, therefore, we don’t fall under any regulatory purview right now,” he says.

“But around the world, we have seen BNPL companies come under scrutiny because of excessive fees levied on consumers, so we just cut out all consumer fees and made it as consumer-friendly as possible. And that has worked for us.”

Tan says the company gains its revenue by charging merchants a success fee between 4% and 8%, depending on sales from partner stores. While it may seem like merchants are on the losing end, Tan explains that it is of value to merchants as Split is opening up a merchant’s customer base by providing a consumer segment that they previously wouldn’t have access to.

When someone applies for a credit card, a bank typically carries out a credit risk assessment before approving their application. This is to ensure that the applicant will be able to service their repayments after using the card. The traditional credit risk assessment can take days and would require people to submit a slew of documents and financial statements, which may deter some from applying altogether.

When developing Split, Tan and Vishvesh had to come up with their own credit risk assessment method as there was no comprehensive credit scoring system available at the click of a button.

“So imagine us providing our BNPL service using traditional assessment methods and you, as a consumer, wanting to buy a yoga mat. You will have to wait two days for your purchase decision to come about and that is a poor user experience,” Vishvesh explains.

“We knew we needed to build our own credit risk assessment model from the ground up. I know this may seem like rocket science, but it’s actually not. We just rely on informed decisions and we keep updating those decisions as we get more data.”

What does this mean? Let’s say you have to decide between two transactions. The first one is a customer buying premium fitness apparel in Bangsar. The second is a customer is buying a second-hand budget smartphone in a rural town. Vishvesh explains that, based on preliminary data, it can be concluded that the first customer probably has a medium to high income and thus, has a lot more disposable income to invest in their fitness. 

“So we take this principle and apply it to the hundreds of variables that we have. We look at your behaviour on the website, age, your spend amount, the time at which you made the purchase and so on. And with these parameters, we are actually reasonably able to predict whether a person is going to pay us back,” Vishvesh says.

Tan says in Split’s credit assessment model, every consumer who comes onto the platform is assigned a score based on certain parameters. On the other hand, merchants also have a score to indicate their risk threshold. This means that high-risk merchants will have high thresholds. These thresholds then determine whether a customer can purchase a product or service, based on the score assigned to them.

Over time, as a consumer uses Split and builds a relationship with the system by paying instalments on time and providing positive interactions, their score will be upgraded, qualifying them for more purchases.

“So there are cases where a consumer cannot make a purchase with Split because we need to be very responsible with how we offer our services. We see BNPL as a credit stepping stone for everyday Malaysians so we try to encourage good repayment behaviour and we incentivise with the gamifying experience, where consumers have access to more purchases if we have a good experience with them,” Tan explains.

“In the early days, it was just the two of us doing this weekly but today, at scale, we’re using machine learning (ML) models to do it, which has significantly improved our repayment ability. And despite not charging any consumer fees, we now operate as a positive-margin business.”

He adds that they are also looking at the underbanked community, who may not have access to comprehensive banking services but do have basic bank accounts. This alternative credit risk assessment will also allow more people in this community to access various products and services that may have been inaccessible to them before, which is one of Split’s core missions.

“The goal is to have zero delinquency, but we still have a long way to go,” says Tan.

Fraud is a common problem faced by every payment provider, says Vishvesh. One of the main types of fraud is when a customer’s payment method is being accessed without their permission, so very early on, Split worked to nip it in the bud.

“We took a stance that we would not let a transaction go through unless the bank account holder approves it. So when you sign up for a Split account, you would give us your personal contact details and before we charge the first instalment, we will put you through a three-step secure authorisation process,” Vishvesh explains.

“We use a lot of fraud tools that global financial institutions are using and we also signed up with three major fraud prevention providers to fortify our fraud prevention.”

Tan and Vishvesh acknowledge that the pandemic placed a financial strain on many people. So when a small handful of consumers did not respond to payment reminder emails, they went down to the source to find out what was going on. In one case, a consumer could not service his repayment as he was let go and had to work as a ride-hailing driver to make ends meet.

“We understand situations like this and what we do is we work on an alternative plan for them, kind of like instalments on instalments, if you will. The whole reason is to help them instead of pressuring them financially,” Vishvesh says.

“Once we have the data, we retrain our system as well. Our goal is to have a predictive model but we will need thousands of occurrences before that can happen. So for now, we take the human approach where we empower our customer service team to work out flexible payment plans with consumers, without charging them any extra fees.”

Tan says the whole system is automated right down to this point, where human interaction is pivotal. “Our philosophy is to take a very human approach to recognise that at the other end of the transaction is another human being. They have a job, they have their personal lives and they have their families. So when it comes to financial hardship, let’s just empathise with them and help them get out of it. We don’t want to punish them for financial hardship, that’s not in our interest.”

Tan foresees the BNPL market taking off well locally, as not only does it service the underbanked, but it may also be the preferred choice for certain transactions. Countries such as Australia, the US and even parts of Europe that offer BNPL services have seen a shift from credit cards to BNPL.

“Many consumers realise that they are paying annual and interest fees on credit cards, while BNPL offers better services. I expect to see BNPL emerge as one of the more popular ways of making payments in the future,” says Tan.

Save by subscribing to us for your print and/or digital copy.

P/S: The Edge is also available on Apple's AppStore and Androids' Google Play.