Optimizing House Name Selection for Sister Title Prospect Mailings Using Machine Learning

Silverstar Brands (SSB) is a Direct-to-Consumer mail order company that currently mails 5 catalog titles monthly. Catalog circulation and the 12-month buyer file for each title continues to shrink, while the cost of printing and mailing catalogs continues to rise. Prospective names to mail are more difficult to find and more expensive to purchase. This project created a model to choose customers of one of SSB’s titles that are likely to purchase from a sister SSB title if mailed a catalog from that title. The dollar amount of the processing fees the customer paid, and the number of sister brand catalogs the customer received were found to be the most important variables in the model. Linear regression was the algorithm chosen due to its fast training and prediction time versus other models with similar scores. The model was not as predictive as the current mailing strategy, but insights were uncovered for further refinement of the model.