An Interview with Credolab, an Alternative Credit Score Startup, Optimizing Consumer Loan Approval

With many overlooked communities and credit-worthy people getting the cold shoulder during loan applications, Credolab is set to change that.

(Image Credit: credolab.com)

An Interview with Michele Tucci, Chief Product Officer of Credolab.

About Credolab

Credolab is one of Singapore's hottest fintech companies that has developed a bank-grade digital credit score backed by alternative metadata for a wide range of lenders and consumers. Credolab uses anonymous metadata to optimize and assess an applicant’s financial health and status to help financial institutions decide whether or not to grant a loan, a credit card, or any other financial services to the potential candidate.

What was the inspiration behind Credolab?

The inspiration was from Peter Barcak, our CEO and founder. He was a chief risk officer for a bank in Ukraine, and he was challenged with the lack of data to score people properly. He tested a few data sources while still in Ukraine and realized the smartphone data was quite predictive, even more than telco data. He decided to move to Singapore and established Credolab to address that very problem. There is the lack of data to score the unscorable, especially in emerging markets. This was back in 2016. If you Fast forward to today, scoring the unscorable is a problem even in the US. According to FICO, 53 million individuals in the US do not qualify for mainstream financial services. They have a FICO score below 500. According to other sources, there are 62 million of these people. So what Peter realized, back in 2016, was that the percentage of smartphone users was increasing, and people were generating data by virtue of using their smartphones. So we decided to develop an embedded finance solution that sits within the mobile app of the lender and accesses only first-party and privacy consented data that we use for credit risk assessment.

Is Credolab’s alternative credit score an alternate score to FICO, or does it supplement it? Does it allow for more or less candidates to be approved from loans?

It is an alternative but also a complementary tool to FICO. From a statistical point of view, the two types of data (credit scores) have a very low correlation. The interest of lenders and banks is to use data that are orthogonal to each other. So if there is a low correlation, and the data is also predictive, every bank should be encouraged to use as many data sources as possible. So we compliment the FICO score and supplement it with a different analysis of the same individual. We focus more on the behavioral side. We look at the willingness of customers to repay a loan. At the same time, FICO traditionally looks at the ability of somebody to repay and also looks at historical data as a predictor for future behavior.

When you look at the sales funnel of a lender, you have different points where the lender pulls different data, different data sources, and different scores. Some verifications and assessments need to be done. FICO usually is pooled deck origination, loan origination, or card origination. The lender has two choices, a soft pool or a hard pool. You can look at Credolab similar to a soft pool in terms of assessing pre-qualifying customers. So the Credolab mobile score could be used in the front of a funnel to filter out delinquent customers and those based on the analysis of the digital footprint that are very likely not to repay. So once this very first assessment is done, and perhaps even after the KYC is done and the verification of the identity is verified, the lender begins the hard pull from FICO. So our system optimizes the funnel and also optimizes the way costs are allocated. Our system may also approve more customers for loans by reducing unit economics, which is how much money they spend to underwrite every single application.

So today, without Credolab, lenders pools a lot more soft pools from FICO, however some of these candidates should be avoided because they will be rejected further along the process because of low FICO scores. Using our mobile score could also optimize the funnel, to help approve more creditworthy customers but also overlooked customers as well.

How do you use anonymous metadata to score customers, and are the scores accurate representations of financial health and status?

We have two components of our technology. One is the way we access the data, and the second one is how we extract the information value from the same data. So the way we access data is through a mobile SDK software development kit. So imagine this as a small library of code that sits inside the mobile app of the lender and accesses data only when the lender warns. The lender is in full control of the data workflow. We access data only after receiving the user’s privacy consent and after receiving the operating system’s permissions. So there are at least three levels of protection, and if that wasn't enough, we added a fourth layer of protection of data privacy. We only access metadata, which is data about other data, so we don't read any personally identifiable information. Once we have access to the metadata, we apply the second part of our technology, the data modeling pipeline, rooted in machine learning algorithms. We extract value from this data, and we do all of this in real-time, so there is no lag. It's an embedded scoring solution that delivers value without adding friction to the process.

What is your response to customers that are worried about privacy concerns?

So we are very, very transparent about it. We clearly mention what data we access, why we access data, and how we process data. In doing this, we comply with the GDPR in Europe but also the CCPA in California. Other states in the US are not as evolved as California is in

terms of privacy protection. We also have a prominent disclosure within the app itself, so when the customer goes through the application form of a loan or a credit card through the app before our technology accesses data, we disclose that we are about to access data, so it is contextual. We don't access data persistently, and we don't work in the background continuously checking what you are doing with your mobile. We access data one time only upon loan application, and that's it. If the customer doesn't apply for a new loan, the SDK sits there without doing anything. So you can look at it this as a fifth layer of protection.

What are the future ambitions and vision for Credolabs?

As a company, we have come a long way. Today we are in 29 countries from the US to Australia, from Mexico to Nigeria, and everything in between. We have grown from being vertically focused on retail banks and consumer finance to buy now pay later, digital lending, and super-apps. We are an embedded scoring technology company, and we can participate in the mobile onboarding journey of any company with a large customer base and is moving into finance. We are able to monetize data assets that e-commerce portals already have and help bring them to their full potential. We are able to monetize relationships with existing customers that are not necessarily buying every day from your e-commerce platform but are willing to borrow money from you. Because of the existing relationship with the e-commerce portal, there are more chances to be approved through the e-commerce portal rather than a traditional bank. A trend that we also see in the US is of overlooked communities such as the African-American communities. We saw an increase in the number of digital lenders focused on these communities that are traditionally leftover from mainstream financial services. But guess what? They do have a smartphone! So without introducing any bias into the process, any rash racial profiling, we can still offer these overlooked communities a chance to receive credit at fair terms without needing to go to subprime lenders or payday lenders.

We plan to continue to focus on buy now pay later with a focus of helping overlooked communities. The reality is because of the high percentage of smartphone users and increased number of individuals applying for loans digitally, we can offer a score for 100% of our customers and not just the hit rate of the Credit Bureau or Experian boost score. Our advantages are many. We have a high hit rate, high predictive power, no bias in the model, with a low correlation with FICO score. With expansion into the United States a big priority, lenders should line up and test our alternative score!

---

About Michele Tucci

Michele Tucci is the Chief Product Officer of Credolab. With over 20 years of experience in the FinTech space, Michele has worked for companies such as Captial One, Mastercard, Jusp, Mint Payments, and FinstarLabs. Michele is originally from Italy but has worked all over the world, currently residing in Singapore.

https://www.linkedin.com/in/micheletucci/