Harnessing data to serve the underbanked

A wave of innovation is addressing market inefficiencies that make it difficult for banks to accurately assess the credit risk of this segment of Americans.

Americans with low or no credit are ill-served by many financial institutions. A common assumption is that this group of 100 million doesn’t want to pay their bills, or that they are irresponsible. This view is both hasty and wrong.

The underbanked, or those who lack access to traditional financial services, is made up of many different types of people, including retirees, entrepreneurs, small business owners, students, young adults and immigrants.

Data and machine learning have catalyzed a wave of innovation to address the market inefficiencies that make it difficult for banks and credit unions to accurately assess the credit risk presented this segment of Americans.

Sign up for the free BAI Banking Strategies newsletter and get industry insights delivered to your inbox.

By clicking the Subscribe button, you acknowledge that you have read our Privacy Policy and Terms of Use and agree to be bound by them.

Financial institutions are working to harness new data sources in service of more accurate credit underwriting. These include proprietary data sets generated by systematic tests that focus on learning about specific underserved segments, as well as bank account data that complement and augment traditional bureau data.

Machine learning and causal inference techniques have increased the accuracy and reliability of credit scoring, and innovative companies have also invested in making their models more explainable and fairer – two facets of credit underwriting that will only increase in importance in years to come.

Additionally, innovations in data science have quickened the pace of model building, validation, review and deployment. This results in better software quality, greater model accuracy and faster solution deployment. These improvements compound and enhance financial inclusion by giving us a more accurate picture of someone’s risk. They also allow companies to reimagine the business processes for marketing, acquisitions, fraud and operations.

It is easier than ever to launch a new product or service, but harder than ever to keep it simple for the user.

Solving this conundrum boils down to a focus on understanding customer needs and replacing complexity with experiences that are simple and intuitive. The best companies in this space use high-quality, proprietary data and algorithms across multiple facets of their products and services to analyze sequences of user interactions, which in turn allows them to personalize and simplify the user experience.

In the digital world, speed isn’t everything – it’s the only thing. Whether it is a service query, money transfer, credit application or cash advance, a real-time action that solves customer needs can be a memorable experience that keeps current customers and brings new customers in the door.

For the underserved, waiting is a source of anxiety because a delayed payment can often mean an unpaid bill or an extra fee that they may not be able to afford.

To increase the speed of decision-making, many data disciplines must come together. It starts with robust data product management and governance that ensures high quality data is accessible and able to be analyzed, a modern data stack that enables real-time streaming and computation, and a decisioning stack to deploy and monitor models and policies.

Because customer preferences are changing rapidly and the demand for new products and services is accelerating, financial companies need to act as learning powerhouses that can use data to identify a customer need and home in on product-market fit through tight cycles of testing and feedback.

Simply having great data isn’t enough, however; what matters is how you use it. For example, in order to innovate and manage risk in products and services, data companies should not be siloed in their own team, but rather integrated throughout the business. Data should be widely available and trusted throughout the organization and fostered as a core competency among business and product teams.

The factor that differentiates data leadership is fostering trust among the most important stakeholders. First, employees need to trust the provenance, accuracy and quality of data in order to make decisions assertively and confidently. Second, customers need to be able to trust that their privacy is being protected and that their data is being managed securely and used fairly to benefit them. And third, regulators and partners need to be able to trust and verify data and model governance.

For a data company to be great at serving the underbanked, they must enshrine trust in the data principles that govern all aspects of data collection and use.

Shane Holdaway is chief executive officer at Mission Lane.