Der Weg zum Ruhm - Nutzung von Kuratorennetzwerken zur Vorhersage des Erfolgs von Human Brands



On access-based music streaming services users are confronted with seemingly unlimited choice and content owners such as artists and music labels compete for the users´ limited time. To help consumers navigate, curated playlists have emerged as a new tool to influence demand. Despite their importance, the role of playlist curators in shaping demand is not well understood. On the one hand, curators base their decisions to include a song on their playlists on information about new songs before they become available and they may thus influence the trajectories of these songs. On the other hand, curators may select songs that are already successful and, hence, merely react to a song's positive trajectory. In this research, we develop a methodology to determine the importance of playlist curators using a combination of machine learning methods and network analysis techniques. The model is trained on weekly curation information for a sample of 11,000 playlists from a major global streaming service over a period of more than one year. By placing each playlist in a latent space, we determine their centrality in a dynamic network of playlists at a given point in time. We then relate the aggregate weekly centrality scores to streaming data on the song-level for a sample of 182,000 songs. Our findings reveal that playlist centrality is the key determinant of a song’s success and that playlist centrality is relatively more important than the number of playlist followers since large playlists tend to list songs after they have achieved a certain level of success. We further find that songs particularly benefit from influential playlists in an early stage of the product lifecycle and that the playlist effect is stronger for independent and long-tail artists, who are generally less likely to be discovered via other channels due to limited marketing budgets. The results are relevant for content publishers, whose investments in new talent are both costly and risky.
Tatsächlicher Beginn/ -es Ende1/09/2030/12/22


  • Music streaming
  • Digital distribution channels
  • Deep learning
  • Content curation
  • Playlists
  • Influencer marketing
  • Music industry
  • Network analysis