Listen to streaming music - Analysis and comparison with radio

Audio streaming services have enjoyed extremely sustained growth in recent years: stream volumes on these services have risen from 28 billion in 2016 to 85 billion in 2020, and no fewer than 20 million people are now using audio streaming services. At the same time, radio is still a central medium for accessing music: in 2021, 17.5 million people will be listening to radio music programs on a daily basis. The medium's developments are less favorable, however, and year after year, the percentage of radio listeners is falling, as is the time spent by listeners on the medium. These declines are particularly marked for music programs.

These opposing trends, in a context marked by a disparity in regulations between radio providers of radio and streaming services, have naturally led the regulator to seek a better understanding of streaming practices and the consumption of users of these services, particularly in relation to radio use, which is now better known and documented. The streaming services studied in this study include some of the biggest players in the sector: Spotify, Deezer, Apple Music and YouTube.

Comparisons between radio and streaming must necessarily be handled with care, whether because of the socio-demographic differences that remain between users of the two types of service (even if streaming services now reach all age categories, the oldest populations remain underrepresented among their users) or because of their fundamentally different nature: programming orchestrated by the provider of radio in one case, on-demand access based on a vast catalogue in the other. Numerous studies have shown, however, that streaming services also have the capacity to guide their users' listens through a variety of methods: algorithmic notices, the composition of playlists managed by the operator, or the highlighting of certain titles on the home page.

The study shows that, across all the streaming services studied, the most successful titles are predominantly French-language, in contrast to radio, where English and French titles are on an equal footing. Rap also clearly dominates the list of titles most listened to on streaming services, whereas radio has a much more balanced breakdown across all stations. These first two results are directly linked, as rap is now mainly listened to in French. Furthermore, the most listened-to tracks on streaming services are highly concentrated. On some services, the 10% most listened to tracks account for up to 60% of all listens, while radio is less concentrated. Another notable difference is that tracks stay much longer in the top radio charts (12 weeks) than in those of streaming services, where rotation is faster (around 3 weeks). Titles also more frequently achieve significant success on streaming services before finding success on radio. In addition to analyzing listeners' habits, the study also looks at Spotify's track promotion practices.

This is analyzed in three ways: the creation ofplaylists, the highlighting of these playlists on the service's homepage, and the algorithmic notices used by the service. The study shows that the most popular playlists, as well as those highlighted by Spotify on its home page, are predominantly made up of English-language titles. Furthermore, while rap remains the dominant genre within the most popular playlists , the playlists highlighted present a more marked diversity of genres, with a sustained presence of pop/rock in particular. Finally, Arcom's preliminary analysis of a notice algorithm highlights the importance of track language in the algorithm's operation.

Arcom carried out this study based solely on usage data made public by streaming services. Future work by the Centre national de la musique's observatory of musical diversity, of which Arcom is a partner, may shed further light on the impact of new listening modes on musical diversity.