Home / Banking Strategies / Drowning in data, starving for insight: Starting the customer analytics journey

Drowning in data, starving for insight: Starting the customer analytics journey


Customer analytics has received a lot of attention at banks in recent years. Banks possess a plethora of data on their customers, in many cases already collected in data warehouses for regulatory compliance. And this suggests there must be value to extract from all that information. But what data to use, exactly? And to what end?

Indeed, enormous potential exists for improved customer relationships—and operating margins locked inside that data if the organization can derive insights and act on them. Here’s the good news: Customer analytics does not require moonshot levels of investment. In fact, starting with simple goals often marks the best way to start forging data-driven customer relationships. Often, opportunities lie within easy reach that offer significant returns. Once the original objectives are accomplished, organizations can apply additional levels of analytics, with increasing degrees of sophistication and investment. Here’s what the journey looks like.

Going full circle: Start with the end in mind

Having worked with hundreds of companies on customer analytics, we base our viewpoint on how best to earn a return on investments on significant experience and observation. The common principle across industries and companies engaged in profitable analytics applications is this: Begin with the final objective in mind and work backwards.

Many potential applications of analytics have already been identified and implemented in the industry. Customer acquisition, retention and yield management are the most frequently discussed, but the list of possible applications is broad across the lines of business at banks. Successful implementation requires matching analytics to operations; having an operational team member lead the project is often the best approach.

Make your own grade: Identify how to improve the business

A simple phrase describes how many businesses make operational improvements via customer analytics: Go from one-size-fits-all to personalized. A client made this statement after we had worked with him on several projects and it summarizes many of the opportunities to operationalize analytics throughout the organization.

Let’s use customer retention as an example. The process for introducing an analytical-driven retention campaign into an organization can be as simple as moving from that one-size-fits-all communication strategy to messaging driven by deviations from normal customer patterns.

In one non-banking project, we found customers had the same payment pattern over their tenure with the company. We could categorize customers as “early,” “on-time” or “late payers”—because customers typically paid their bills as soon as they received the notice, when the bill was due or when they paid all their monthly bills, which may have been after the due date on the bill from this company. We found that sending reminder notices to all customers at the bill due date proved too early for some and too late for others.

But once we identified each customer’s typical payment time, we sent reminder notices once they landed one standard deviation away (typically three days) from their usual pattern. This timing of reminder messages helped the company to react quickly when early-paying customers were late and avoid irritating customers very likely to pay later in the month. The result: This message timing adjustment reduced customer churn by 14 percent.

Skin on the digits: Design and complete the analysis

Once banks identify an opportunity to improve an operational process with analytics, they can next determine the analytical approach and the necessary data. In the above example, the approach was to calculate a payment-date distribution for each customer. The data needed for this calculation was the date of historical payment transactions by account. 

For customer acquisition analytics, econometric models can estimate the effect of certain factors on a customer’s probability of acceptance. The data necessary for these models are comprised of customer acquisition attempts—both wins and losses that have been made—with information on the factors for each offer included in the model. Data can be collected for the initial analysis using manual extracts. If the models and processes prove effective, automated data collection can be developed and the probability calculation updated for each acquisition campaign.

The buzz of A/B: Test and learn

Verify the accuracy of predictive analytics using A/B tests, applying the suggested action to a sample of customers you can observe in comparison to a control group that does not receive the suggested action. We recommend ongoing A/B tests for all analytically supported processes, and dashboards that show the predicted and actual results enable executives to stay informed about the project. 

When a bank considers a decision in which they lack data (often the case with a new product or pricing model), A/B testing offers an excellent path to support the decision without the use of predictive models. In these cases, several hypotheses can be defined, and multiple tests conducted, in advance of enterprise-wide implementation. Ongoing testing and reporting will foster improvements to the process over time.

Return of the consumer: Improving customer profitability without losing customers

Raising revenue and operating margins through targeted pricing and service offerings represents a field of analytics employed in many industries. It’s often called yield management and has grown in prevalence and sophistication in recent years. Banks can now individually target prices due to the one-to-one nature of the bank-customer relationship. While individual pricing may sound controversial, it can benefit customers as well as banks if it’s done right. Just as airlines have created service tiers and sophisticated pricing rules, banks can design customized service offers and price schedules to best meet an individual customer’s needs. For price-sensitive customers, the ability to access a limited set of services for a reduced cost offers a better option than being priced out of the bank altogether. For banks, it can improve customer retention and profitability.

From takeoff to touch down: The customer analytics journey

As banks adopt analytics throughout the enterprise, they can leverage greater sophistication and investment. But innumerable technologies and big data solutions exist, so beware of “boiling the ocean” when all you need is some warm water. To recap, executives should keep in mind the following guidelines when considering customer analytics projects:

  • Have a line of business team member lead the analytics project.
  • Do not do anything without a business case that justifies the investment.
  • Start with the end in mind—identify the decisions to be supported by analytics; select the analytics approach; collect the necessary data.
  • Define the metrics for measuring success.
  • Verify success through A/B testing and performance reporting.

By the numbers, a bright future

If the effect of applied analytics on other industries is a guide, banks have enormous opportunities to improve performance through data-driven decision-making. The amount of data banks have on their customers is vast, and the nature of the bank-customer relationship enables frequent and repeated interactions, providing many opportunities to learn about the customer’s preferences and take targeted actions. The potential for customer analytics at banks is limited only by management vision and creativity.

Matt Lindsay, Ph.D., is president of Mather Economics, an analytics firm with clients in banking and myriad industries worldwide. He is based in the greater Atlanta area.