Finding the Sweet Spot: Ad Delivery on Streaming Media

Abstract

Free ad-supported streaming of on-demand content is growing. Platforms that provide this service need to find a balance between the interest of the viewer (to increase content consumption) with the incentives of the platform (to decrease ad avoidance). Using causal machine learning methods that combine debiased machine learning with instrumental variables, we capture the causal effect of four independent levers of ad delivery on content consumption and ad avoidance. We investigate whether there exists a sweet spot in the values of these levers that balance the interest of both stakeholders. Our results show that a decrease in the number of pods (ad breaks) or length of pods results in an increase in content consumption and a decrease in ad avoidance. Similarly, an increase in the diversity of ads decreases ad avoidance but has no material impact on content consumption. However, we observe a tradeoff for spacing till the next pod. An increase in spacing results in an increase in content consumption but at the cost of an increase in ad avoidance. We discuss the theoretical mechanisms behind our findings and present implications of our results for streaming platforms.

Keywords: Streaming Platforms, Ad Avoidance, Ad Levers, Causal Inference, Debiased Machine Learning

JEL Classification: M31, M37, C14, C36, C61