Predictable Response.

• Author: Collins, Steve
• Position: Ways to boost response to direct marketing campaigns - Statistical Data Included

You can take the guesswork out of your direct-mail marketing by using a technique known as 'predictive modeling.'

This approach boosts your return-on- investment and saves valuable marketing dollars.

Have you had a direct mail program in which you were not happy about the response rate? Don't junk the product or redesign your mailing package. The problem may be in the way you selected your target audience. At least 50 percent of the success of a direct mail campaign is attributable to effective audience selection.

Many direct mail campaigns that produce marginal responses are the result of either "intuitive" or "monolithic profiling" audience selection. Using your intuition to select a target audience is not always a bad thing. But, just because you "feel" like a certain group of customers will be responsive to your offer does not ensure success. Using a "monolithic" or generalized profile of your targeted customer may be slightly more scientific, but it can still fail to fully exploit opportunities for selecting the very best customer candidates.

It may be true, for example, that, on average, customers are aged 25 to 44, have incomes of $40,000 or more and are married. But, this monolithic approach tends to create a narrow classification system and ignores the interaction between variables or their relative importance in determining who should be mailed. You may find that customers who are age 55 plus, have incomes of $75,000 or more, and are single are also good candidates.

Predictive modeling gives your direct mail programs the performance boost they need by taking the guesswork out of audience selection. Predictive models are simply mathematical equations that define the characteristics of the target audience that you want to replicate. They are based on observing current and historical customer behavior or studying people who reacted to a prior marketing campaign. Effective predictive models are built using multivariate statistical techniques that take into account the importance of examining how variables work together in determining who would be a good candidate for selection.

Predictive models can be useful for solving a number of marketing problems:

* Who will be most responsive to my mailing?

* Which customers are likely to generate the most revenue?

* What is the next most likely product a customer will buy?

* Which accounts could possibly close within the next several months?

These are all scenarios in which we, as marketers, can use predictive models to make the best decisions.

Consider the following example of how a bank marketer can use modeling to enhance the performance of a home equity line of credit campaign.

Prior to building and implementing a predictive model for audience selection, the bank used a set of basic...