What Marketers Can Learn About Predictive Modeling From the US Men's Soccer Coach

For some time, digital marketers have seemed prepared to award the Ballon d'Or (which recognizes the European football player of the year) to Big Data for its purported ability to address the world's marketing challenges.

The reality, however, may be quite different.

Big Data may be creating a rigid view of the marketplace, which may be doing more harm than good. As the world immerses itself in the FIFA World Cup, the US Men's National Team provides an apt analogy.

Everyone seemed shocked when US Men's National Team coach Jurgen Klinsmann cut fan favorite Landon Donovan from the squad heading to Brazil for World Cup 2014.

Donovan has been called "the best soccer player in United States history." He is the US all-time leading goal scorer with 57, has the second-most caps in team history, and appeared in three World Cups.

It's hard to imagine US soccer without Donovan.

But let's take a step back… As a coach, Klinsmann has the authority—and the responsibility—to build the best team possible. Taken literally, that would include selecting players that have the best stats in their respective positions. By taking that approach, Klinsmann would end up with a pretty terrific starting eleven and one that would almost certainly include Donovan—on paper.

However, there's no accounting for chemistry with this approach, and so the question is whether a team that meets the key the performance metrics will perform as well as one that also considers the intangibles.

That is what happens when online marketers rely on selections based on cookies. They're able to assemble an audience that meets a limited set of very specific performance metrics: age, gender, household income, intent, etc. On paper, those marketers get an audience that looks great but one that doesn't necessarily perform well in the marketplace.

This isn't to say that the data isn't useful or important, but its application in this black-and-white manner limits its value.

The better approach, on the pitch or online, is to use data to build predictive models rather than select strict parameters in a serial manner. By weighting various performance metrics, managers and marketers alike can vastly expand the pool of qualified team members or audiences. The reality is that a player with 37 goals and 20 assists would probably do just fine. On the other hand, if a marketer were to set household income at $150,000, he or she would miss out on plenty of targets making $149,000.

By using weighted models, a broader range of qualified candidates can be added to the consideration set. More variables, such as intangible assets, can be factored in, and better results can be delivered.

Big Data Can Be Hypnotizing

Unfortunately, for too many marketers — and fans — Big Data can be mesmerizing.  And in some ways, why those folks are mesmerized is understandable. The data itself is amazing... but its use in most situations is still rudimentary and largely ineffective.

Treating data-driven marketing as a set of binary switches falls woefully short of its potential, just as assembling a set of high-performance all-star players without any recognition of the intangibles could create a fragile under-performing team with few chances to move beyond the group stage.

A well-modeled audience—one that is statistically weighted and indexed — provides marketers with real value over the long term. It gives them holistic and valid points of reference for understanding their customers and finding qualified prospects again and again and again.

So, was Klinsmann wrong in cutting Donovan? That remains to be seen. But building a high-performance team is about more than pulling from stats on paper. Marketers need to have that same mentality as they build their audiences.

MarketingProfs article: http://www.marketingprofs.com/opinions/2014/25396/what-marketers-can-learn-about-predictive-modeling-from-the-us-mens-soccer-coach