Commercial audience retention of television programs : measurement and prediction

Program audience ratings are typically used as a reference in placing and pricing television advertisements. However, the discrepancy in audience size between a program and its commercials impairs the reliability of the program ratings. This study proposes a new metric, commercial audience retention (CAR), to measure a program’s capability of retaining its audience when a commercial break occurs and develops a model to predict a program’s CAR. The CAR metric and prediction model are tested and validated using a sufficient dataset with consumer TV viewing and program broadcast records for 1 year. We find that some factors that influence program ratings or commercial avoidance have no significant effect or have different effects on CAR. Our empirical results may be of value for advertisers and TV stations in purchasing and pricing of commercial airtime. A real-world application of the CAR metric for an advertising company is offered as an illustration.