Big Impact with Big Data​: Navi, Using Data Science to Transform Lending

Data is now known as the oil of the 21st century, and Navi, the trend-setting digital lending app, is achieving some major goals in the BFSI space using data science to offer speed, user convenience, and improved customer journey.

Let’s look at what went behind building an infrastructure that is fast, flexible, and scalable to achieve the unthinkable in loan underwriting; shortening what traditionally took about a month to just a few milliseconds.

The end game for us was to enhance the customer experience in the loan application process. In traditional banking, a loan application requires the applicant to provide multiple supporting documents, and the application itself takes weeks or sometimes months to process due to slow underwriting.

Also, many applicants require small-ticket loans under 5 lacs, which is not worth the operational costs for many traditional financial institutions.

By automating underwriting with AI and ML, we were successful in radically changing the risk assessment process, making it fast and cost-effective for us, and easy, affordable, and accessible for customers to apply for loans.

Analyse and Predict to Optimize

The robust, predictive ML models at Navi assess applicants based on important variables like their history, loan history, banking transaction data, demographics, income, etc., to gain deeper visibility into their past financial behaviours. Predictive analytics on the payment history of the applicant helps us predict their future repayment behaviour.

Our ML models are good at uncovering anomalous patterns present in an applicant’s dataset and past fraudulent activities such as application fraud or information fraud. With data analytics, we can explore and mitigate risks.

Geared for the Future with a Tech-First Approach

As Navi grows in popularity, especially post-Covid, as we predict small scale businesses will turn to Navi for quick loans, the data points to look at will also increase. We are confident of handling the influx of data as the scalability of our platform is something we’ve addressed since the beginning and are constantly working to enhance. Our tech-first approach to all problems, present, and future, allows us to embrace growth.

With moratoriums on loans during Covid, the bureau history of applicants who defer on their EMI payments is not a true representation. Such factors are also considered while modeling. Additionally, we focus on better customer segmentation and solve for each segment separately to build refined ML systems.

The data science team is working on ML models that can consume large datasets, learn from them, and make accurate predictions by compiling and analyzing newer datasets. With multiple teams looking at the same data, we are also ensuring the data serves as a single source of truth to avoid implications as data grows.

Conclusion

I’ll end by reiterating that a tech-first approach at Navi has allowed us to build ML systems with unparalleled predictive power and speed, and use data science to deliver on Navi’s promise of instant, paperless and accessible loans.

“Data science is revolutionizing every industry today. Why not the BFSI space? With a tech-first approach that is now synonymous with Navi, we took on the challenge head on with a data-driven mind-set to build ML models that transformed credit underwriting. We’re confident our ML models will support Navi’s hypergrowth with massive-scale analytics.”
— Prathyush Potluri, Data Scientist