In an age where fintech startups are growing at a rapid pace, Pune-based EarlySalary is creating ripples in the lending market. With their mobile-first platform, smart risk scoring system and other unique offerings, they are ready to take on the competition head-on. Founded by Akshay Mehrotra and Ashish Goyal, the company is also looking to introduce a fast and easy way to get a loan or buy products and pay later.

Underlying their unique offerings is the extensive use of analytics and data science. And Balakrishnan Narayanan, who is a seasoned analytics professional with over a decade of extensive experience in banking and financial services, is heading this crucial department. He has been a part of leading global banks contributing to the field of advanced analytics and also consulting engagement with fortune 500 companies across geographies. 

Analytics India Magazine got in touch with Bala, as he is fondly called by his team members, to understand how is he driving data analytics at EarlySalary and the challenges that come with spearheading an ambitious startup. With a team of more than 25 people, he is on a mission to help the business evolve with data. 

Here are the interview excerpts:

Analytics India Magazine: How are you driving the analytics functions at EarlySalary?

Balakrishnan Narayanan: Analytics is a culture within Earlysalary, it is a way of life. All our analytical solutions are derived keeping the end customers in mind. Our goals are aligned with business objectives and priorities. We have pioneered the real-time underwriting process using non-traditional data. Our focus has always been to set a high standard within the risk and credit function using analytics which is critical for any lending business. Other avenues where analytics plays a decisive role is customer analytics, enabling us to understand our customers’ behaviour to remain service-oriented and ensure customer delight. Innovation and hunting for new-age data science solutions help us remain ahead of time.

AIM: What are some of the challenges you face while leading the analytics functions of one of the leading fintech companies in India?

BN: As a startup ‘speed’ is very critical, we don’t want to miss out on an opportunity because we are not ready with a solution or offering. We have always been aggressive when it comes to growth at the same time we have managed our risk well within the set margin. Currently, we process about 90,000 loans a month and we aim to grow this by 10x. While that’s a very positive sign, however, this comes with a lot of challenges. More data to store and consume means we need to upgrade our existing infrastructure to handle that volume in real-time. Tools and techniques that we use today need to be upgraded keeping future in mind. Further, the team needs to be upskilled so that they can start using the latest analytics platform and algorithm.

AIM: What are some of the challenges you face while setting up the data science team at your organisation?

BN: Analytics is still evolving so are the analytics tools. Once you adopt a tool and build capability around it, migrating to newer tool becomes a very tedious task as there are millions of customers who might get impacted. Hence, we need to have a futuristic view while selecting the right technology stack. Another major challenge is getting the right talent to work with you. Though there is a huge demand for experienced professionals, it is more of a candidate-driven market with a Demand and Supply gap. Especially, fresh university graduates or with a couple of years’ experience with some training into data science have a very high expectation from the job without being ready for the same.

AIM: What are the major avenues that you are tapping with analytics at EarlySalary?

BN: However, overwhelming the word fintech could be we are still in the business of lending or ‘risk’ to say. Our focus would still be on building one of the best-in-class credit and risk platforms. New-to-credit (NTC) customers are our prime target segment. These customers typically do not have any historical loan or repayment behaviour on the bureau and would be difficult to underwrite using the traditional methodology. Hence, we continuously strive to enhance our capability to understand the customer segment better by identifying the newer data source. We do not shy away from trying or getting any kind of new data to experiment with. We have embraced the taboo word “garbage in garbage out”. As an analytics practitioner, I believe it is our prime responsibility to get into newer and unknown avenues to get relevant data than worrying about the quality of the data. It is our data engineers who transform and enrich the quality of the data which could be consumed for further insights.

As an analytics practitioner, I believe it is our prime responsibility to get into newer and unknown avenues to get relevant data than worrying about the quality of the data.

Balakrishnan Narayanan

AIM: How has analytics adoption helped you further your goals at EarlySalary? Please highlight with use cases.

BN: Our Leadership team understands the importance of analytics hence right from day one of Earlysalary’s existence we have relied on analytics. That’s how we have reached a mark of 1 million loans in a short span of fewer than 4 years of our existence. Our AI/ML-enabled real-time underwriting platform churns a large amount of structured and unstructured data. Our algorithm looks at over 3000 features of a customer before making the decision to approve or reject the customers’ application. Imagine the traditional way where underwriters were physically analysing these data points while approving a loan, it would have been practically impossible.

AIM: What are some of the goals in analytics that you aim to accomplish this year? What is your roadmap for the coming year?

BN: Fraud management is one of the important areas where all the fintech lenders are struggling with. It is easy to build a model to predict the ability and willingness of the customer to repay the loan, it is equally challenging to predict intentions. We have strong ML algorithms and processes in place to identify these maligned intentions. However, we would want to further enhance our learnings from the past to build a stronger foolproof system that would outsmart the fraudsters. The other area that we are working on is process automation using embedded analytics. We have seen the success of RPA in the past and it also boosts our customer experience. One of the other critical tasks would be migrating from the traditional applications to a new-age platform that helps setting up the base for future expansion.