Three “Musts” To Get Value from Data Science Innovations in Finance

All financial institutions are on the lookout for breakthroughs that will give them a competitive advantage. The field of data science and analytics provides fertile ground for this search, thanks to the increasing growth of ‘big data’ and machine learning methods, such as Oracle digital banking experience (OBDX) into the mainstream industry. According to Gartner, by the end of 2024, 75 percent of organizations will have made the transition from piloting to operationalizing artificial intelligence.

Most management teams recognize the value of innovation in a broad sense, and business theorists have spent a great deal of time and effort developing best practices for promoting creativity inside organizations. But our experience working with customers on data-enablement projects has persuaded us that specific considerations must be made in the data domain to maximize the possible return on one’s investment.

To be effective, data science must be directed towards strategic objectives.

The conventional wisdom holds that most great innovations emerge organically – in the case of data science, from a data scientist discovering ‘signal’ in a dataset that can be used to develop a predictive model, which can then be used to automate or support decisions – rather than through scientific discovery.

On the other hand, the underlying assumption is that the data scientist begins their search in the most appropriate location. As a result, data science professionals are in short supply, and organizations must decide where to put their attention. Traditionally ‘quant’-oriented front-office trading teams, for example, are common; however, they are rarely seen working with back-office functions such as finance or information technology operations, which are often viewed as procedurally-driven and therefore unappealing targets for data science work in general. In spite of this, such tasks account for some of the most significant cost drivers in the financial services industry. It is possible that the application of machine learning to optimize seemingly-mundane processes in IT servicing, financial reporting, or other similar areas will have a far greater impact on a company’s bottom line than marginal improvements in market or customer behavior analytics in some instances.

The critical thing to remember is that banks should visibly connect the deployment of data scientists on exploratory work with the goals inherent in the company’s overall digital banking platform strategy. The strategy must be intelligently targeted to do this; for example, a strategic directive to “cut expenses” or “increase income” is meaningless unless it is considered in the context of how banks should accomplish this. Data analysis, on the other hand, may be instrumental in assessing prospective benefits and guiding prioritization decisions. Having established this foundation for success, data product owners should be pushed to demonstrate that their teams’ emphasis is an intentional pursuit of the outcomes that have been identified to be of the greatest strategic value.

Banks must integrate data science into the operations of the company.

The strong agreement currently is that data scientists, analysts, and engineers perform at their highest productivity levels when integrated into close-knit multi-disciplinary groups.

Benefits received as a result of such integration are both direct and indirect. Improvements in communication can result in immediate benefits for data scientists and their business customers. Data scientists working side by side with their business customers can be far more responsive to business needs, reducing the friction that can occur when data is transferred between siloed business and analytics teams. The DataOps concept (which is slowly gaining in popularity) embodies this approach, leaning significantly on the advantages of Agile and DevOps methodologies in software engineering, which have been shown to be effective.

Indirect advantages occur as a result of the greater sensitivity that data scientists working in embedded teams gain to the business environment and priority setting processes. This contributes significantly to bridging the gap between the top-down strategic direction outlined above (and allowing for fine-tuning of that direction).

Data science products must be put into production as soon as possible.

A statistical model in the form of Python code or a Jupyter notebook adds absolutely nothing to a company’s bottom line. When the code is put in a calculating engine that feeds automated scoring or decision processes, and when that code is presented to a user who acts on the insights it gives, the value accrues. Improvements in business decision-making as a consequence of this process represent genuine company value in the form of decreased risk and enhanced financial performance.

Consequently, the efficiency with which an organization can convert an idea for a data science project into code that can be performed on an operational system – known as the ‘move from concept to cash’ process – is an essential indicator of the commercial productivity of data science. This extends beyond the discipline of data science to include data and applications architecture, governance, technology delivery, and business process engineering and necessitates the involvement of a much larger pool of stakeholders within the organization.

We have found that, although bigger organizations have access to a greater and deeper pool of data science experts, they are less flexible when it comes to operationalizing the insights that data science gives than smaller, less mature organizations in our experience. Naturally, forces are at work to close the gap in both directions: data science is becoming more democratized through the provision of more managed offerings (for example, by public cloud providers), and larger organizations are increasingly seeking our assistance in implementing more agile ways of working (for example, by public cloud providers).

The ability to critically analyze how data science projects proceed from original idea to production implementation and aim to minimize any needless friction from that journey is vital for an organization wanting to develop in this environment.

Data science is becoming a more critical part of the company’s strategy. Still, it’s vital for executives to avoid oversimplifying the issue by just recruiting the best data scientists and letting them get on with it. A practical data science approach and constant attention to digital banking solutions are essential to achieving a net return on investment in the field. As the rising digitalization of financial services continues, those who recognize this will have a huge competitive edge.