Optimization-Based Decisioning: Unlocking the Hidden Complexity Premium for Large Consumer Lenders

Article excerpt

Decisioning methodology implementation based on high-scale mathematical optimization sounds challenging, but the premise of this article is that there is a bounty of benefits for those institutions taking that challenge. The authors review innovations over the past two decades in business-to-consumer decisioning, identify their limitations, and demonstrate how they contribute to, and sustain, the yield improvements possible through optimization today. Organizational prerequisites and common barriers to success also are discussed.

To paraphrase Adam Smith, the strength of large business-to-consumer (B2C) ++lenders lies in their complexity and diversity. However, in the corporate case, this complexity presents both opportunity and challenge. Many corporations today wrestle to unlock the latent potential of their product- and capital-rich consumer businesses. Invisible hands, it seems, do not manage credit lines well.

Those leading-edge consumer lenders who embrace optimization are achieving risk-adjusted growth in profitability and other strategic measures that are impossible to achieve by any other means. And within a consolidating, increasingly commoditized environment, optimal capital resource allocation uniquely promises organic growth and competitive advantage.

High-scale optimization forms the core of strategic decisioning offerings from Experian, SAS Institute, and Fair Isaac Company. SAS and Fair Isaac, originally licensees and partners of Marketswitch, have since launched sophisticated, competitive offerings to those of Experian/Marketswitch. (1) All three are achieving strategic and operating benefits for B2C corporations.

This article's summary of optimization's benefits and real-world effects will be of interest to risk managers, operating executives, marketing officers, and C-level executives. If your organization is not using optimization, it's likely that you are competing with optimization-based decisioning, whether you are aware of it or not.

Optimization's place in rational, risk-managed revenue growth is no longer a matter of conjecture, hype, or hope--it is a demonstrated fact.

Not Your Father's Regression Model

Let's first clarify what we mean by optimization. Second, we shall describe what we term the Three Generations of B2C decisioning, and articulate what constitutes state-of-the-art financial and marketing resource allocation. Third, we indicate where optimization--properly instantiated in the modern B2C corporation--may in turn provide recurring and accelerating growth opportunities for its adherents.

What It Is Not

Optimization is a precise term that is largely abused in common usage. Perhaps the best way to define it is to say what it is not.

It is not working harder, slightly better, and learning incrementally through iterative behavior. This activity-based management ethic ("We'll do more stuff better and at inevitably lower cost than our competitors") is a dead-end street for those B2C lenders who hope to break free of the competitive pack.

It is neither taught to nor typically practiced by statisticians, who, as we know, constitute the bulk of the professional community dedicated to business intelligence. Statistical modeling informs optimization by providing a crucial input to a discrete and fundamentally different mathematical process. While a statistics curriculum may include an introductory summary of optimization techniques and benefits in a course or two, optimization is properly the realm of a distinct academic and professional discipline-operations research.

In its classical mathematical formulation--a very old idea, to be sure--optimization is irrelevant to the demands of the very large-scale consumer lender. Formulating the problem does not yield an operational solution in high-scale environments. The classical, simplex-method solution to this formulation is a balky piston engine in a world requiring silent and powerful turbines. …