Predictive Data: It's The Real-Time Thing
* Predictive Models. Although much has been made of the sophistication and utility of predictive models, the fact is they normally generate response rates in the 2 percent to 5 percent range. While this may be good compared to random samples or even intuitive selection, it's nothing to crow about. The model predictions are wrong more than nine times out of 10 because they look at everyone's behavior through the same mirror. Today, however, there are true alternatives that don't rely on historic patterns of other customers, but rather on the real-time behaviors of each customer. And when it comes to marketing data, this is no time for anything other than the real-time variety.
That's because with new technologies accelerating the speed of commerce, a marketer who is slow to use real-time data will fast disappear if windows of opportunity keep closing at an ever quickening rate.
That's not to say historical purchase information has no value. As with any history, it has important lessons to teach, especially about understanding customer behavior over longer periods of time. Historical data, however, has been shown to have very little predictive value in anticipating such behaviors as attrition or defection. Understanding the behavior of your customers in the future is, more than ever, dependent upon knowing what they are doing today.
Fortunately, with today's increasingly affordable and sophisticated database technologies and marketing applications, it's possible to get real-time views of purchasing behavior and their implications through an invaluable tool:
* Behavior maps. Behavior maps provide an invaluable customer cartography. They allow marketers to track the most recent purchasing activities and state-of-mind of their clients and thereby project a far more accurate picture of future purchasing behavior. The most sophisticated of these behavior mapping technologies can even interpret nonactivity.
The combination of these two capabilities constitutes a breakthrough in the field. It's one that can drastically increase a marketer's likelihood of reaching potential customers when they are most likely to buy. Just as important, the behavior map can help that marketer package an offer to which a customer is most likely to respond, and produce the individual behavioral insights that are most advantageous to the marketer.
Why, for example, should a video marketer offer a "Fifth Movie Free!" promotion when a customer has a pattern of buying five movies at a time anyway?
Because each person is a unique individual with individual tastes, behavior maps finally allow marketers to understand and address those individual's needs with the proverbial right product at the right price at the right time.
Behavior maps don't rely on assessment of other customer purchase patterns, which may or may not have any relevance to the customer you may be looking at. Instead, it relies on assessments of that particular customer's behavior and makes offers on the basis of the customer's past interactions coupled with their current purchasing activity or lack thereof.
The result is a revenue-generating shift away from the relatively fuzzy "all-for-one!" marketing widely used today and toward the one-to-one marketing ideal behavior maps make possible. This better informed marketing is created by capturing the uniquely different paths of behavior each customer establishes and projecting it onto a behavior map. These paths reveal a customer's specific needs and/or inclinations. And they help to answer many vexing questions that have baffled marketers.
What is the likelihood of an immediate purchase? Are sales accelerating or decelerating? Is the customer more price-sensitive than interested in convenience? Does a customer's behavior paths indicate a growing or shrinking share of wallet?
A behavior map can answer all those questions and more, but more importantly, at its best, it can answer them accurately for a specific person rather than for some vague cohort that shares a common profile.
Can it raise the chances for merchandising success beyond the paltry 2 percent to 3 percent response rates that most direct marketers are used to? Can it exceed the higher effectiveness of live sales people? When marketers understand individual purchase patterns, they learn what circumstances, if any, are necessary to spur sales. They also find out which offers may be exercises in futility -- before they waste a lot of time and money in the discovery process. In either case, behavior mapping finally delivers the systems intelligence and marketing insights needed to generate more revenues, reduce marketing costs and very likely, both.
Jeff Caplan is vice president of marketing and business development at Verbind Inc., a marketing software developer with offices in Boston and Massachusetts.