In my last post, I mentioned four FICO clients that have taken significant steps toward their customer centricity goals. Doing right by customers—not only in terms of conduct but in more sensible and tailored allocation of products and limits, and better communication—is leading to improvements in such key performance indicators (KPIs) as product utilization, customer satisfaction and compliance risk exposure. These upticks are, in turn, leading to improvements in market share and various financial performance measures.
The first client—which I'll refer to as Bank A—is a regional division of a global banking group, and a leader within its markets for net income, profitability, efficiency and fees-to-expenses ratio. Some of this success comes from making customer-level decisions, including setting global credit limits.
Even so, the bank was leaving profit “on the table” at originations, as it was making exposure decisions based only on customer monthly income. Exposure allocation across products was all about controlling losses. Revenue criteria weren’t being taken into account at all.
To make more profitable decisions, the bank needed a scientific way to balance risk with potential reward in four packages of its retail credit products. Once the decision is made that a customer qualifies for a package, how could the bank not only set the most profitable overall credit limit, but distribute that exposure across component products in the most profitable way?
Decision strategy optimization provided the answers. The first step was modeling the decision. Decision models help you deal with decision complexity because they make explicit the relationships between all important factors. Models capture these relationships in mathematical equations and tie them to an overall objective like maximizing profit.
The combinatorial possibilities in decisions spanning multiple products can reach dimensions that are difficult to manage. One way to simplify these complex decisions is to break them into parts. As shown below, we performed a two-part optimization, with both aimed at maximizing customer-level profit.
I break down this decision model in great detail in our new Insights white paper “Customer Centricity: Four Bank Success Stories” (No 78; login required), which I invite you to read. Because diagrams like these can only summarize what’s happening, it’s worth pointing out that there are deeper components also handling a piece of this complex decision. These include action-effect models, which, in my view, provide the real core of power for decision strategy optimization. They predict customer reactions to each of the possible bank actions and the impacts of these reactions on KPIs like revenue.
Action-effect modeling is key to performing reliable simulations before deploying a new decision strategy. In fact, when you’ve optimized a decision strategy based on carefully constructed action-effect models, using a simulation tool is almost like having an interactive map with levers you can adjust to explore and understand performance drivers and trade-offs. “What if?” scenarios are reliable because they’re rooted in the data-driven predictions about behavior and results.
Bank A explored a range of optimal operating points, shown below. This range—which is often referred to as an efficient frontier—demonstrates how the optimal decision strategy “moves” as the bank loosens or tightens the constraint of global exposure.
The blue dot directly above the gray BAU dot shows that strategy optimization raises profit by 27% with no change in global exposure. The green dot shows that profit goes up even more when global exposure is allowed to rise to 110% of BAU. The inset chart shows that one of the ways the optimal strategy is raising profit is by doing a better job of balancing risk and reward based on measures like net income.
Did it work? Yes, quite spectacularly. Optimal product limits encouraged improved spend, balance and payment activity. Bank A reached all of its portfolio objectives while increasing overall profit. Results were so strong, in fact, that the bank was able to achieve project payback in six months and a one-year return on investment of 6 to 1.