The Data Challenge
Once the segments were defined using circulation behavior, data from the Union-Tribune’s proprietary readership study and the local do-not-call list were appended to the file to help define and localize the segments. Because both data sources lacked addresses, address records were reverse-appended to the telephone numbers using a unique cross-reference directory. Once the three data sources had addresses, they were combined into one file for analysis.
The proprietary readership study provided a wealth of information and helped to localize the segments to the San Diego market. This study provided information on daily versus Sunday readership, section readership, purchasing habits and topics of interest.
The local do-not-call list was obtained from www.donotcall.gov. The recent change in telemarketing legislation made it imperative that the Union-Tribune identify which consumer segments were most and least likely to be on the Do-Not-Call Registry.
Taking the three data sets provided by the the newspaper, Integras conducted a data audit—a critical step for companies with traditional operational customer files (files focused on billing versus marketing analysis). This process enabled Integras to validate the information provided by the newspaper and to interpret internal codes. The data audit revealed data anomalies such as zero values, missing fields, problem source codes and outliers.
Once all of the information was validated and corrected for analysis—which required several weeks of interaction between Integras and the Union-Tribune’s IT department (and multiple data pulls to correct for data integrity issues)—more than 300 behavioral profiles were created using factor analysis and clustering techniques. The resulting system yielded six final segments that were both meaningful and relevant in terms of multiple behaviors.
These six segments range from Segment No. 1, which represents a customer who pays full rate for his or her subscription and has subscribed for more than three years, to Segment No. 6, which represents subscribers who tend only to purchase the newspaper at a discount and stop subscribing once the discount expires.