For predictive analytics to work, two different species must cooperate in harmony: the business leader and the quant. In order to function together, they each have to adapt. On the one hand, the quant needs to attain a business-oriented vantage. And on the other, the business leader must navigate a very alien world indeed. Deal and Pilcher’s new book, “Mining Your Own Business,” helps with that second bit.
Note: This article is excerpted from Eric Siegel’s foreword to the recently released book, “Mining Your Own Business: A Primer for Executives on Understanding and Employing Data Mining and Predictive Analytics,” by Jeff Deal and Gerhard Pilcher and was originally published in Analytics magazine.
Bridging the Divide
Bridging this gargantuan divide is worth the effort. Take for example a tax fraud detection story worth 10 digits (covered in the book’s Introduction). Elder Research, Inc. (ERI), the consultancy that spawned the book, delivered predictive models to the IRS that increased the agency’s identification of a certain type of tax fraud by a factor of 25. This saved the feds billions (with a “b”).
This success exemplifies a widely applicable paradigm. Across commercial and government sectors, predictive targeting achieves a multiplicative improvement to broad scale operations (albeit often a single-digit multiplier rather than that whopping 25-fold improvement). In addition to deciding which tax returns to audit, predictive models determine which customers to contact for marketing, which debtors to approve for increased credit limits, which patients to clinically screen, which employees to woo away from quitting, which persons of interest to investigate, and which equipment to inspect for impending failure.
No Longer Business as Usual
Thus data science earns its status as hot, lucrative and sexy. This is the Information Age’s latest evolutionary step, technology that taps data to drive decisions more effectively. It’s the very act of scientifically optimizing resource allocation for … just about all processes. Various outlets have dubbed data scientist as the best, most in-demand and even “sexiest” job. And if you haven’t heard, data is the new oil. Industry research forecasts that demand will continue to grow and estimates the global predictive analytics market could reach as high as $9 billion by 2020.
But to capture this value, you must construct a durable bridge across the quant/business culture gap. The core technology – which learns from data to predict – is only half of the trick. Deploying it is more than just a technical process – it’s an organizational process. Existing business operations must change by way of implementing analytics. It’s no longer business as usual; science now drives the enterprise’s primary decisions and actions en masse. In this sense, data science is intrinsically revolutionary.
Deployment is Key to Success
As a result, the greatest pitfall that hinders analytics is to not properly plan for its deployment. For each analytics initiative, it’s critical to build a pathway from the get-go that will lead to integration. This requires bridging the cultural gap. It takes the socialization of buy-in: Line of business staff must agree to make big changes. To that end, they must learn what a predictive model does for them and they must be willing to put their faith in it.
That doesn’t always work out. With refreshing frankness, this book reveals an Elder Research study of their own early client projects that showed a full third of projects fail to attain business results, despite 90 percent attaining technical (analytical) success. The difference is often whether the organization actually implements the fruits of analysis.
Overcoming Resistance to Change
No guts, no glory. With inertia, resistance to change, and a lack of confidence as primary impediments, there’s no more eye-catching antidote than Dr. John Elder’s legendary willingness to put his money where his mouth is. Before founding Elder Research, John once invested all his own personal assets into his own predictive stock market trading system, in response to hesitancy on the part of his client to move forward (I recount this story in detail in the book “Predictive Analytics”). And in a story from Chapter 6 of the book at hand, “Mining Your Own Business,” John doubled down against a major credit card company, betting Elder Research could beat their established analytical methods to model credit risk. If Elder Research failed, they’d cut their service fee in half; but if they won, the cost would double (yet, in the latter case, the company would gain enormously from the improved predictive model). This tactic served handily to move the project forward.
Don’t worry. When inertia hinders progress, you don’t necessarily need to take the dramatic approach of wagering your own money. There are other options among the established best practices for managing predictive analytics initiatives. It’s largely about educating the organization and opening discussions to understand the concerns of skeptics.
Value in Prediction
Unfortunately, convolution and the appearance of arcane complexity threaten to extinguish a newcomer’s excitement about the potential value. This might leave the person feeling compelled only by the pressure that comes from hype: “Everyone’s doing it!” Let’s nip that in the bud right now. Predictive analytics’ value is simple and concrete: It helps run operations more effectively by way of predicting behavior, i.e., the outcome for each individual consumer, employee, healthcare patient or suspect. These predictions are each just numbers, aka scores or probabilities. Since they directly drive decisions, they’re by definition the most actionable deliverable you can get from analytics. One need only learn a limited bit about the (fascinating) “rocket science” that generates these predictive scores to integrate them and realize their value.
Regrettably, today’s tremendous data hype does not always relay this value proposition or any specific value proposition at all. The pervasive buzzwords big data and data science enthusiastically remind us there is value to be had, but do not refer to any particular technology or approach. These terms are general catch-alls for “doing smart things with data.” They really have no agreed upon definition beyond that, although they do allude to a vital cultural movement lead by thoughtful data wonks. Big data is nothing more than a grammatically incorrect way to say “a lot of data” (like saying “big water” instead of “a lot of water”). Data science is a redundant term, since all science involves data; it’s like saying, “book librarian.” In one section, the authors of this book delve deeper with solid coverage of the extensive taxonomy of terms and technology.
In a field propelled largely by data nerds, it may come as no surprise that most books serve the hands-on quant. Those books dive into the technical practice. After all, for a quant, the technology and software tools are much more tangible and easy to define than the more elusive, “human” arena of organizational processes and project management. As a natural-born geek, I know from personal experience.
This book is different. Jeff Deal and Gerhard Pilcher wrote it to serve the much neglected other side of the coin: you, the business leader. It delivers the two ingredients you need for success: 1) an understanding of the technology so you can speak the quant’s language, and 2) a guide to analytics management best practices, including how to build your analytics team and avert the most costly pitfalls.