Changing the Rules of the Game

There was a time not long ago when marketing decision makers regretted they didn’t have enough information.

In the world of information, especially database marketing, we have gone from scarcity to surplus almost overnight. This means we’ve exchanged one set of problems for another. Where we once were reluctant to act because we didn’t know enough, marketers are now hamstrung because they know too much, some of it contradictory and much irrelevant or incomprehensible.

But there is hope. A prospect model can identify the most important data attributes and turn this problem into an opportunity.

We all know the numbers. It costs six times more to acquire a new customer than it does to retain an existing one. Attrition affects every company, regardless of industry, products or services. Whether your company is a start-up that does nothing but prospect, or an established firm, it’s prospect or perish. Prospect modeling can statistically change your cost structure and actually be predictable-and yes, profitable.

One of the basics of database marketing is that by analyzing your customers, you can identify the ones that are the most (and least) profitable. This makes it easier to recognize the characteristics of prospects who resemble your best customers.

A prospect model will help spot these qualities and determine the importance of data attributes and their combinations. These attributes are assigned weights according to their importance, and a score is then based on the weighted attributes.

Database marketers have access to modeling software tools and companies that provide analytical services to turn this abundance of data into information and knowledge. As more and more marketing databases are built and implemented, marketers are realizing that leveraging the data and identifying opportunities can be challenging.

Because of this, modeling, data mining, neural networks and other statistical and artificial intelligence technologies have gained in importance. These tools use different methodologies to reveal hidden data patterns, trends and correlations that allow marketers to predict solutions to business problems and/or increase profitability-and in this case, identify good prospects.

There are several steps in developing a prospecting model. Two prerequisites: customers and a prospect database/universe. With all the data and lists available, you can get lost before you even start. The prospect model will allow you to build on the sources of proven prospecting data and to expand your universe.

Once you decide on the lists and sources to include in your prospect universe, analysis should be conducted to determine the files to be rented and incorporated in the database. The lists can be analyzed based on penetration and revenue per record against a customer list or house file. When the lists are identified, negotiations must be arranged with the data owner for use of that information in the database.

The prospect database is created like any other marketing database. Business rules must be developed and maintained during the process. You may decide to include your house file with the prospect database for future analysis and comparison. Once the file is complete, the prospect model can be built.

The next step is to enhance your house file with the appropriate data. Business-to-business data includes SIC code, number of employees, annual sales volume, year the business was started and more. Examples of consumer enhancements are income, age, marital status, type of dwelling and other lifestyle information.

Once the enhancement is finished, profile and segment your house file. This process uncovers hidden sub-segments of customers with similar characteristics. Customers are classified into their respective homogeneous groups to identify unique target markets within the customer file, as opposed to viewing all customers as a single entity.

Then develop the regression profile model. This model will mine the prospecting database for demographic clones, and will target each customer group separately for more precise results. Prospects will be ranked from most likely to least likely to respond. Upon completion of this step, your prospect database is scored and ranked in tiers of performance.

Now the fun-testing and validating the model-really begins. Within each of the performance tiers, randomly select the same number of prospects for the test. This test will help determine the depth to which the model can be effectively and profitably utilized.

Validation occurs when the responses mirror the projected lift by the model in the performance tiers. These responses are used to refine the model by the performance of the mailed universe, thus optimizing profitability and ensuring that the model is performing well.

The prospect model can also be used to “filter” future list rentals. Before going into the merge/purge, outside lists must be enhanced and then scored with the model. Once the outside lists are scored, the names that fall below your break-even would be suppressed.

The power of prospect modeling is truly realized with the abundance of data now available. With prospect modeling, the ultimate problems facing database marketers should be how deep to mail into the model and what offer to use. Their only regret should be that they didn’t start prospect modeling sooner.