The road to digital sophistication in financial services

Not too long ago while attending a tech conference in Las Vegas, I struck up a conversation with my tablemate who said he worked in the financial services industry as a CDO. He got a puzzled look on his face when I started talking about collateralized debt obligations. He then quickly clarified that he served as his company’s chief digital officer.

A CDO’s main job is to drive the organization’s digital transformation strategy to make the organization more profitable and efficient. In other words, the CDO is responsible for the adoption of digital technologies at the enterprise level to spur innovation, as well as to streamline and expedite day-to-day operations.

We talked at length about his vision for the company. He said the biggest hurdle he faced had nothing to do with technology, but rather with culture and change management. It’s a conversation that has stayed with me, and I want to share some of the insights I gleaned. In part one, I’ll discuss the strategies to achieve digital sophistication and get CDOs unstuck in financial services. In a separate blog, I’ll explore the challenges and opportunities in health and life sciences. I take this approach mainly because most of the examples of transformation I’ve worked on at CGI have been in these two industries.

Digital transformation challenges in financial services

The financial services industry has come a long way from the days when offering customers ATM access was the height of digital transformation. There is still, however, a long way to go. Based on our most recent CGI Client Global Insights report, only 17% of financial services organizations say their digital strategy is producing results.

What are the barriers that CDOs (and others charged with transformation) have to overcome? I’ll examine some of the more popular technologies driving digital initiatives in the financial services sector and discuss how organizations can get “unstuck” in trying to leverage each of these technologies.

Robotic process automation (RPA)

RPA uses artificial intelligence to automate mundane, repetitive and manual tasks. RPA implementations are frequently “stuck” due to employee resistance, a lack of skilled resources, high procurement costs, unclear use cases, and high maintenance costs.

The best way to succeed in RPA implementations is to apply a phased approach. Prioritize low and labor-intensive tasks first, while simultaneously empowering employees to focus on tasks that require human intelligence (e.g., approving a loan application or responding to customer questions). RPA isn’t about replacing employees; rather; it’s about making their jobs easier and more efficient.

Data visualization

Data visualization helps executives visually represent and organize large amounts of data for more informed decision-making. The recent proliferation of data visualization tools is a boon for the financial services industry.

Despite their tremendous usefulness, data visualizations frequently become “stuck” due to a lack of technical resources who can understand the raw data, normalize the data and represent the data in an easy to understand format. In addition, complex data visualizations are hard to understand by the majority of employees who mostly use simple spreadsheets and graphs for their day-to-day data management.

As an alternative, the best way to address these bottlenecks is to bring data scientists and business stakeholders together earlier in the process to discuss the right business questions and the right data for answering them. Data scientists can use various development tools to collect, clean and prepare the data, as well as develop intuitive and easy to understand data visualizations. The business owners can iterate as they delve deeper into the data.

Cloud computing

Cloud computing has radically revolutionized the financial services industry. Gone are the days when companies needed to prepare a robust plan for their IT infrastructure in advance of any application development project. Upfront infrastructure setup time resulted in a very slow start for the application development process and ultimately resulted in longer application release cycles.

Cloud computing is “stuck” sometimes due to security issues, migration complexity, performance issues, compliance requirements, skills shortage and cost management. Ensuring the cloud service provider uses state-of- the-art authentication and authorization mechanisms and complies with industry regulations and standards can alleviate these concerns. The standards must define the process for recovering data in case of a catastrophic failure. The cloud provider should have real-time performance monitoring available to mitigate any performance issues. In addition, automating the management of cloud systems can address the shortage of skilled cloud resources.

Despite their challenges, the top cloud services have taken center stage for application data storage and business logic execution. As a result, the burden of infrastructure maintenance is taken off the application development teams, enabling them to focus more on coding the application business logic.

Artificial intelligence (AI) and machine learning

Financial institutions increasingly use AI and machine learning for a wide range of business functions, including fraud detection, data security, customer support voice recognition, document analysis, risk assessment, mobile check deposit, decision-making, and many other real-world applications.

Machine learning algorithms use rule-based approaches to make decisions rather than relying on hand written conditions. This means machine learning programs can adapt to new conditions quickly without requiring changes to the underlying application code.

Machine learning projects tend to become “stuck” due to unrealistic expectations, skill shortages, complex models, high infrastructure costs, huge data requirements, and hidden program logic.

Unrealistic expectations about what a machine learning model can or cannot do are resolved gradually over time as additional problems of increasing complexity are solved by the machines. RPA and in-house training can address skill shortages. In addition, moving to the cloud can lower infrastructure costs.

Cloud computing also helps by providing the best machine learning infrastructure available (GPUs) to achieve significant performance improvements. The mystery around how machine learning models make decisions doesn’t actually impact their effectiveness. The focus should be on measuring how good or bad the decision is. The huge amount of data required for the models can be collected using tools that automatically scrape data and perform data munging and normalization for easy import to the machine learning models.

In conclusion, the financial services industry should follow the examples of the IT, entertainment and telecom industries to drive the sophistication of their digital transformation journey. Sure enough, each small step taken in the digitalization direction will lead to grand new possibilities.