How Banks Can Make Use of Data-Driven Customer Insight

By Mohammad Nasar and Christopher J. Sifter, PMP
| 11/29/2018
As competition in the financial services industry continues to intensify, banks are ramping up their use of machine learning and artificial intelligence (AI) as part of their efforts to attract and retain customers, increase customer engagement, and improve the overall customer experience. 

A growing number of banks are finding that today’s advanced analytics and data intelligence capabilities can give them deeper insight into their customers’ behaviors and expectations. Armed with this understanding, they are able to do a better job developing products and services that address their customers’ needs and deliver these in a way that reinforces customer satisfaction and loyalty.

Driving the demand for greater customer insight

As new types of financial services organizations compete for their business, today’s bank customers are becoming more discerning and have heightened expectations regarding convenience and accessibility to banking services. At the same time, customers also are becoming more knowledgeable and technologically adept, demonstrating increased affinity for digital technology in general and the online management of financial transactions in particular. These trends, coupled with continuing sensitivity to fees and charges, have provided an opening for many financial technology organizations to fill a perceived gap in certain banking services. 

Banks are responding by taking a hard look at how they manage the customer experience, with increasing focus on mobile and digital transactions. Meanwhile, although regulatory pressures are easing somewhat, compliance continues to demand considerable resources and add to the pressures on traditional banking organizations.

In this environment, a growing number of banks have come to regard analytics, AI, and machine learning as increasingly important tools in their efforts to operate more efficiently while still attracting and retaining customers. For example, when bank executives participating in a recent Crowe webinar were asked how they view machine learning and improved customer insight as an organizational priority, almost nine out of 10 gave the issue some degree of priority – only 11 percent said it was not a high priority. 

Banks’ organizational priorities
exhibit 1
Online survey of Crowe webinar participants, Sept. 20, 2018. Numbers might not add to 100 because of rounding.
Yet even as the banking industry in general appears to recognize the need for new technical capabilities to improve customer insight, many banks are still struggling to embrace the technology. Nearly half (47 percent) of the respondents in the survey expressed interest, but acknowledged there are other pressing needs that are competing for bank resources and attention. Only 23 percent of the participants said they were embracing machine learning and customer insight as part of their day-to-day business. 

Data science concepts

The first step in bridging the gap between technology’s promise and its current level of adoption is to establish a clear understanding of the underlying data science concepts. More specifically, it is important to establish the distinction between AI and machine learning.

Those two terms trigger widely varying perceptions among various audiences, resulting in a significant lack of clarity. For summary purposes, it is adequate to say that AI involves the use of machines that have been programmed to perform specific “smart” tasks. These tasks can range from the automated processing of transactions to the identification of certain defined customer or transactional characteristics to allow for useful and effective customer segmentation. 

Machine learning, on the other hand, is a subset of AI. This form of AI involves the application of self-adaptive algorithms, which a computer can use to build predictive models about likely customer actions or behaviors. Rather than relying on human actors to identify patterns, predict their significance, and prescribe an appropriate response, machine learning automates the identification, prediction, and prescription processes. 

A similar distinction should be made between programming and learning. Programming involves spelling out a complex series of instructions regarding how a machine is to react to various conditions and situations. Machine learning, on the other hand, involves continuous, automated self-improvement, so that the machine adapts its approach to react to new conditions and situations.

Historically, business intelligence has focused on recording and reporting what had happened in the past regarding a customer, an account, or various critical business metrics. The goal was to develop dashboards that helped managers visualize and understand what had happened, so they could develop strategies for improving performance. 

Today’s business intelligence systems move beyond such descriptive analysis and advance to the level of predictive analysis – that is, suggesting what is likely to happen in the future based on what has been observed so far. Further, machine learning advances the process into the realm of prescriptive analysis, in which the system responds by recommending actions designed to increase or decrease the likelihood of something happening – such as increasing customer engagement or decreasing customer churn.

In banking, these capabilities can be used to develop more accurate and detailed customer segmentation models, which organize the customer base into various categories based on similarities that might not have been identified in the past. These models then can be used to estimate and predict future behaviors, such as identifying customers who are likely to repay a loan early, open or close a deposit account, use electronic banking, visit a branch, add to a credit line, or engage in various other activities. Such detailed segmentation can enable a bank to tailor marketing and sales strategies that are more precisely aligned to each specific customer.

Data science applications in banking

Beyond marketing alone, today’s most successful banks are finding new ways to apply these deeper customer insights for a variety of purposes. For instance, many banks are using the insights gained from AI and machine learning to fine-tune their product offerings, in some cases unbundling product features so customers can personalize their accounts with features that are of particular value to them. 

Machine learning and intelligent segmentation are also very helpful in addressing the issue of customer attrition. Applying machine learning models to a broad array of customer data can help banks predict the likelihood that a customer relationship is at risk. By understanding the internal and external drivers that are at work in the relationship, banks can interact with customers in a more meaningful way and apply their resources to those situations in which they are more likely to be able to affect the outcome. 

For example, AI can help a bank recognize patterns that indicate a customer is at risk because of a bad experience or the lack of specific product offerings that meet his or her unique needs. Machine learning then can be applied to develop a proactive intervention strategy to counteract or neutralize the negative sentiment. 

On the other hand, when the patterns indicate a customer is likely to leave because of a geographic relocation or other external factors beyond the bank’s control, machine learning analytics can make the determination to allocate resources to other relationships that the bank is more able to affect.

Data science technology also commonly is used to develop the customer insights that are needed for compliance with anti-money laundering and other regulatory requirements. In addition, AI can be helpful in developing automated responses to transaction monitoring and know-your-customer alerts, thus helping to reduce the compliance and operational costs associated with these programs. 

Critical requirements for developing deeper customer insights

The foundation for developing such customer insights is the ability to access a complete “360-degree” view of the customer. In the past, such a view would encompass every aspect of the customer’s relationship with the bank, but today it is merely the starting point. 

In order to effectively apply AI and machine learning capabilities, banks need to move beyond a view that is strictly internally focused, and instead develop a full picture of customers’ external relationships as well. To accomplish this, banks must draw on a wide range of external information sources that enable them to understand their customers’ relationships with other financial services entities, as well as other businesses, social organizations, and networks. 

Some of the most fundamental challenges banks must address in order to develop better customer insight revolve around the data itself. Making reliable predictions that drive sound decision-making requires data that is reliable and consistent. Trusted, accessible data is the essential fuel that drives the entire effort. 

Most of today’s banks are relatively mature in terms of their IT infrastructure and the ways that they handle the rollout of new software applications, including configuration management and management controls. But the same levels of scrutiny and control often are not applied to the data itself. 

This is where the issue of data governance becomes crucially important. The objective of data governance is to see to it that data is managed as a business-critical organizational asset, comparable to an organizational utility. 

Effective data governance must address the entire data management life cycle – from the collection and management of data to the protection, delivery, and ultimate use of data. In that sense, it is essential that banks recognize data governance is not just an IT problem, but rather requires an organizationwide effort encompassing people, processes, and technology. 

Another critical concept that must be understood involves the business intelligence technology itself. It can be tempting to look for a breakthrough software application or a powerful new platform that will enable banks to begin taking advantage of today’s AI and machine learning capabilities. 

In reality, however, such technology is not available as a single off-the-shelf installation. While there are platforms to help accelerate the process, developing the types of customer insights and analytics being discussed here involves a process, not a purchase. The applications themselves must be built and assembled from various components – and the nature and configuration of those components will vary from one organization to the next. No two banks are the same, so there is no single combination of applications that will be right for every institution.

An accelerated path toward improved insight from data analytics

Although advanced analytic capabilities cannot be achieved with a single application, this does not mean that the implementation of AI and machine learning must be an extended or drawn-out process. Banks can begin to reap some of the benefits of this technology in a relatively short period of time, provided they approach the effort in a logical, disciplined, and structured manner.

Such an approach will begin with data discovery – that is, understanding the data that is available, as well as the linkages, data quality concerns, and existing data governance processes that are already in place. With that baseline established, the initiative can move into exploratory data analysis, including descriptive statistics and the identification of data trends and initial insights.

This analysis provides the information needed for model design and development, including the application of preliminary algorithms and model testing. These models then are validated, refined, and finalized before being put into operation. 

Such an active and disciplined implementation effort can begin producing tangible results in a period of several months, rather than several years. This is essential in view of today’s fast-changing technology – not to mention today’s rapidly evolving competitive environment. 

The application of advanced analytics is taking many of today’s most successful banks into new approaches and new capabilities that affect virtually every area of their operations, including customer relationship management, business intelligence, strategic planning, marketing, sales portfolio management, risk management, and compliance. As the competition from alternative providers of financial services continues to intensify, the number of banks making use of machine learning and AI technology is virtually certain to increase – along with the demand for more advanced tools and technological capabilities.

Contact us

people
Mohammad Nasar
Chris Sifter
Christopher Sifter
Principal