Showtime Predictive Creative Performance

Since its launch in 2015, Showtime’s subscription streaming service has matured with nearly 30 million subscribers. The landscape of streaming is also entering a new era with established leaders facing growing competition from new players, such as Apple TV+, Disney+ and more. It is more critical than ever for Showtime to capture and protect their share of wallet, and understand precisely which elements of their communications are working and impacting new subscriber growth.

To promote their diverse portfolio of original series and grow their user base, Showtime develops and runs thousands of creative assets against various audience segments. In order to assess the impact of different variables, we needed a model that could account for the large number of creative variations and the ebb and flow of programming periods throughout the year. Creative variables in the model were derived from both auto and manual coding methods.

Once the coding was completed, we built a linear regression model on the full dataset to predict cost per lead based on the attributes of a creative asset. This model served as a simple empirical tool to help approximate how different creative choices could lead to a quantifiable lift in performance.

Unlike traditional reporting which is limited in predictive power, this model derives a coefficient for each variable that enables us to isolate the impact of a given creative attribute while holding all others constant.