Beyond Views: Measuring and Predicting Engagement on YouTube Videos

This work studies engagement, or watching behavior in online videos. Most current work focuses on modeling views, which is known to be an inadequate measure of engagement and video quality due to different viewing behavior and spam views. More broadly, engagement has been measured in reading behavior of news and web pages, click-through in online ads, but not for videos. We study a set of engagement metrics, including watch time, percentage of video watched, and relate them to views and video properties such as length and content category. We propose a new metric, relative engagement, which is calibrated over video duration, stable over time, and strongly correlated with video quality. We predict relative engagement and watch percentage with intrinsic video and channel features, and can explain most of its variance - R2=0.79 for watch percentage - without observing any user activity. We further link daily watch time to external sharing of a video using the self-exciting Hawkes Intensity Processes. We can forecast daily watch time more accurately than daily views. We measure engagement metrics over 5.3 million YouTube videos. This dataset and benchmarks will be publicly available. This work provides new dimensions to quantify engagement in videos, and paves way towards in-depth understanding of different video verticals such as music, news, activism and games.