Algorithmic Recommendations and Earned Media: Investigating Product Echo Chambers on YouTube

Abstract

`Echo chambers' in digital media, where a person is exposed to similar content through algorithmic recommendations, are often thought to increase digital polarization. Firms, however, may aim to replicate such echo chambers, to ensure that consumers are repeatedly exposed to their content. We investigate the viability of such a strategy in the context of the largest US charitable organizations on YouTube. Charities have limited resources and could benefit significantly from augmenting earned media impressions. Across two studies, we find that an algorithm recommends a video on a different topic not associated with the focal charity about 45\% of the time. The algorithm frequently steers users to popular videos that are unrelated to the focal charity. This holds irrespective of whether individuals are logged into their YouTube accounts or not as well as independent of the sequence in which users view these videos. Moreover, we show that as a user follows a chain of recommendations provided by the platform to second, third, fourth and fifth recommendation, it becomes increasingly likely that the algorithm moves away from focal charity videos. Our results suggest that it is unlikely that organizations like charities can leverage echo chambers to generate earned media and that attempting to do so could side-track interested users.

Keywords: Echo Chambers, Charities, YouTube, Algorithmic Recommendations