• How did the Oakland Athletics manage to win 103 games in 2002 on a budget of $40 million, while the mighty Yankees won the same number of games with a payroll of $125 million?
• How did the Boston Red Sox reverse an 86-year-old “curse” and win the 2004 World Series, then repeat as champions 3 years later?
Readers of Michael Lewis’ best seller Moneyball: The Art of Winning an Unfair Game are familiar with these tales of “sabermetrics” and know that the power of analyzing data can give any organization an edge, whether the game is baseball or direct marketing. (There is also a new film version of the book starring Brad Pitt due out in September 2011, which promises to show data geeks in a whole new light.)
As kids, we may have played with that Magic 8-Ball toy introduced for Boomer kids in the 50′s, which promised to forecast the future based on any question we asked it. In direct marketing to the 50+ audience (or any market segment, for that matter) we have a similar toy called Predictive Models. While they don’t even pretend to rely on magic, they do rely on statistics, which also hold a certain amount of mystery and intrigue for a number of us. While there are many different methods (regression, decision trees, neural networks) used in building predictive models, they are all based on the same basic principle – you can make better decisions about future results by analyzing and looking for correlations in data collected from past results of a similar nature.
How to find a little “Marketing Magic.” In marketing terms, if you want to know who will respond to an upcoming mail campaign, look for patterns in the data of past campaign results, to determine what differentiates between people who responded and those who did not.
Predictive Models aren’t perfect – they will not guarantee you perfect results and exactly identify who will respond. What they WILL do is give you an edge over business-as-usual approaches and improve your (if you’ll pardon another baseball reference) overall batting average. And predictive models DO work! A look beyond the sports world to the consumer marketing world will show that predictive models are being used everywhere, and they are being used because they have been demonstrated to bring a strategic advantage to companies that use them, and help improve the customer experience for many of these same companies. Most people cross paths with predictive models every day, and may not even be aware of it!
• Banks have been using predictive modeling for decades. Models built on credit (or FICO) scores help banks determine how likely you are to meet repayment obligations on loans. (The scores also fuel response-enhancing models for a number of other products.)
• Amazon.com and Netflix employ models all the time. Looking for a recommendation for a good summer read or a film for Friday night? Online providers use predictive models to make recommendations for books and films based on your demographic profile and past purchase or viewing history, comparing the history of others with similar profiles.
• Brick and mortar retailers are there, too. Many stores will often send out discounts and coupons for products that have been purchased frequently by people in your age or income band, knowing that someone like you might be the next person to make a similar purchase.
So, how about you? Predictive models may seem to work “like magic,” but their application is actually very practical. And practically speaking, they only work when a team of statisticians, database managers and strategically thinking marketers work together to communicate goals, and understand the challenges to build and implement winning models. It will not be an overnight transformation to migrate to these tools if your organization is not already using them. But if you want to be an organization that needs to compete and win against established, powerhouse competitors like the New York Yankees, having a data-aware and analytics-driven culture is a good opening to a long, successful season.