Using predictive modeling for circulation that sticks

One of the best times to think about retention is before you ever approach a customer. Randomly selecting consumers within your market to approach for a subscription sale is a recipe for churn. Today, the preferred approach is to use some sort of database modeling and analysis to profile the most desirable customers.

This modeling can come in many forms (see the related Close-ups), but at its core the strategy is the same:

• Identify a pool of current customers. Identify the characteristics that they share.

• Then go after non-customers who look like the model. For example, if your longer- term subscribers tend to be homeowners in certain zip codes at a certain income level, then targeting non-subscribers that match the same criteria will be more likely to succeed than a random selection of consumers within the market.

The data collection and analysis can be extremely sophisticated, or it can be fairly simple and straightforward. If your paper has not done this type of modeling before, begin by keeping it simple. You can probably extract basic geographic and certain demographic data from your own customer files. Then you can purchase lists of consumers who match these criteria from virtually any list broker.

Some newspapers turn this concept around. They profile the customers that have quit the paper. The idea is to find out characteristics that are common among people who have not proven to be ideal long-term prospects. The newspapers then apply this model on their current subscriber base to identify those most likely to quit. This pool of customers becomes the "at risk" database, and garners added attention from the retention team. You could call this "reverse modeling." The strategy is identifying a group of people for special treatment based on the characteristics they share with others.

More sophisticated modeling options are available from outside market-research vendors and list brokers. These models are more complex but can yield a greater range of prospect groups for you to approach. Using these models you can carefully segment your audience by gender, age, lifestyle, purchasing habits, and many other criteria. If your newspaper is adding a golf section, for example, you can use data to identify people who’ve purchased golf clubs in the last 12 months, or people who subscribe to golf magazines.

Some of the attached close-up articles provide real-world examples of newspapers using modeling strategies. But as an industry we have barely scratched the surface. These approaches are not only effective, but cost effective. Much of the “cost” is in the form of intellectual capital. That’s just a fancy way of saying “thinking.” And thinking is free.