HOW LOUDNESS, SONG NEGATIVITY AND PLAYLIST PERSONALIZATION CAN INCREASE SPOTIFY’S CUSTOMER RETENTION

Abstract

In this paper, we consider the music streaming service provider, Spotify. We consider 14 attributes of a song (13 of which provided by Spotify, mostly subjective, a few objective) and analyze the relationship between these 14 attributes/ "variables" and a song's position in the top 50 songs in the United States Top 50 playlist. Specifically, we first examine, for each of the 14 variables, whether the top 25 songs have a different mean from the mean of the bottom 25 songs. Then, we analyze, using linear multiple regression analysis followed by linear stepwise regression analysis, the relationship between position of the song on the playlist and the values of the 14 variables.