CW Innovation Awards: How DBS is transforming credit processes

Phased approach

DBS deployed the CAP platform in phases, starting with China and Taiwan in the middle of 2020, followed by Singapore later in the year. The project was delivered by 13 cross-functional agile teams working in lockstep across Singapore, Hong Kong, China, Taiwan, and India, using virtual collaboration tools and techniques.

Though the team faced challenges due to Covid-19 containment measures, it was able to deploy the CAP platform on time and within budget. During the pandemic, the platform proved its worth by digitising credit processes, enabling its relationship managers to cope better with the growing demand for corporate loans.

In subsequent phases, the platform, hosted on Pivotal Cloud Foundry, provided enhanced workflow capabilities and advanced features such as term sheets and optical character recognition (OCR) for financial spreading, as well as support for large financial institutions.

“We digitalised and simplified end-to-end credit processing and set the foundation for advanced credit risk management using on-demand cloud native design, analytics and machine learning,” said Jimmy Ng, group CIO of DBS.

With an IT staff strength of 17,000, DBS did much of the development work in-house. This includes building user interfaces, workflows and reports, using technologies such as Angular, JavaScript, Web Components, SpringBoot, Java, Kotlin and Docker. Other capabilities such as covenant data storage and application programming interfaces (APIs) were co-developed with Infosys. 

Time and cost savings

By simplifying credit processes and enabling collaborative digital workflows, the CAP platform has slashed the time that relationship managers spend on preparing and monitoring credit memos from 20 days to just three days. Today, they spend more time working on deals and serving clients directly instead of on administrative tasks.

And by using a modern technology stack, DBS has reaped infrastructure cost savings. Unlike traditional monolithic applications, with the cloud-native components in CAP, the bank has achieved “blue-green” deployment capabilities, so that changes can be deployed in production anytime without disruption to users.

Moreover, the platform auto-scales to cope with increasing load and auto-heals in case of failures, putting CAP at the forefront of DBS’s efforts to use technology to build transformational solutions.

And by making the platform’s APIs available on its internal marketplace of reusable software assets, DBS has been able to drive cost efficiency and reduce the time to deliver new products and services. It also led to savings of about 130 working months due to reuse and seamless integration.

The roll-out of CAP has also unlocked heaps of structured data that provides the foundation for tapping data intelligence and machine learning. Using structured data, 40% of the data and analysis is now machine generated, freeing up time for credit analysts and relationship managers to take on more complex work and client engagement activities.

Data-driven innovation

DBS stands out among its peers for its data-driven approach to innovation, underpinned by its 4D methodology (discover, define, develop, deliver) in problem-solving.

The first step to generate user data – structured and unstructured – was intensive and most important, underpinning the entire ideation process. This step began by classifying the users of the system into various personas.

For example, the personas identified in the CAP project were institutional banking group relationship managers, credit relationship managers, credit controllers and supporting business and support units.

Interviews with the identified personas were conducted in a design lab where they were not only asked questions, but also asked to perform tasks and activities. Various hands-on activities and group exercises were also conducted, all intended to generate data for the next step – synthesis.

During synthesis, the data gathered in the previous step was organised in digital format which could then be sorted filtered, and queried. For example, a recorded interview was reviewed by several members of the team and each member recorded the key observation in the system. 

Each observation became a data point. For example, “I don’t like logging into multiple systems” mentioned in the interview was entered as a pain point in the system with all the relevant context around the comment. After the data was entered digitally, it was classified, weighted and scored.

The data was then organised and bucketed into various prisms such as heat maps to identify meaningful patterns, key themes and strong trends, which were then translated into user experience (UX) designs to solve the problem.

DBS’s ideation process culminated with the creation of UX designs even before they were picked up for development. These designs were further refined by conducting user sessions on the working designs during several stages of development.

Soumya Ghoshal, managing director, middle office technology, technology and operations at DBS Bank, said CAP has been one of the largest exercises the bank has undertaken in agile development.

“We took a two-in-a-box approach, which was business and technology teams working seamlessly, hand in hand, and united by one single goal,” Ghoshal said. “The whole agile cadence was focused on creating a platform that was flexible, adaptable and robust, something that will stand the test of time.”