Effects of music recommendation systems in the streaming era

BY BOLUWATIFE DANIEL-ADEBAYO

In the early 2000s, streaming platforms were introduced as a new way of accessing music. They were to act as a liaison between the artist and their audience. According to the IFPI Global Music Report of 2025, the music industry generated $29.6 billion (U.S Dollars) in revenue in 2024 and 69% of that revenue was generated from streaming platforms and services like; Spotify, Apple Music, Deezer, Tidal.

Streaming platforms and services have also acted as an agent of music discovery. This is with the help of a successful intensive curation system involving both human inputs and algorithms. To put more context, algorithmic curation is the process of organizing, selecting and presenting subsets of a corpus of information for consumption. This system of operation provides a personalized engagement for its consumers but an over reliance on user data. This raises a critical concern for artistic autonomy and cultural diversity.

The algorithms help to simplify the process of personalizing music by creating playlists with the help of a data feedback loop. Playlists have become so instrumental that commercial radio programmers usually play trending songs on Spotify and Apple playlists over promoting songs themselves. As a result, the algorithms are not just deciding, filtering and selecting what to expose the public to but are disciplining the visibility of an artist within the platform.

A good example of an algorithmic curated playlist is the Spotify Discover Weekly playlist (SDW) and New Music Friday playlist (NMF). The SDW playlist curates 30 songs weekly for each subscriber whilst the New Music Friday is a global playlist. Research showed that when a song was added to a global playlist, the streaming numbers of the song greatly increased and when the song was removed the song lost numbers.

Although algorithms democratize discovery by exposing niche artists through playlists like Discover Weekly, yet, this exposure favours commercially niche acts with major label backing over independent cultural distinct artists. For independent artists with niche genres, these streaming platforms heralds a threat of invisibility for them. Without a major international label backing, they could be finding it hard to reach their target audience.

Artists have thus had to adapt their creative processes and promotional strategies to fit algorithmic logics. The detriment of this is that, music is now being treated as data by these tools and artists are pressured to think and act as software developers. This could somewhat yield positive results for the industry as there will be more creativity in the production of music.

The opaqueness of the inner workings of the algorithms would force artists to try new methods of optimization in order to benefit from the algorithm. By so doing, the artists will be forced to create new promotional techniques and be innovative in the production of music. Artists would inadvertently be forced to use sonic optimization or infrastructural optimization through contracts to prep their music for discovery. Sonic optimization involves making musical features of a song meet perceived parameters of a platform’s algorithm i.e. mood – based playlists. Infrastructural optimization on the other hand involves incentivizing repeated listening to trigger the algorithms and influence algorithmic placements on playlists thereby manipulating digital platforms.

This functional and mood driven consumption could lead to a homogenization of musical styles with artists releasing and producing the same style of music creating music to fit playlists that will make them discoverable. This capitalist shift towards a playlist centred approach will inadvertently push artists away from their authenticity in exchange for commercial gains. The long run effect of this is a cultural and ethical devaluation of the music being produced and distributed.

Moving forward, there is a need to find a balance to push the creative boundaries and allow for equality in the industry. This would require prioritizing artistry over corporate profit. By removing unfair contractual clauses, record labels would be hindered from enforcing the need for optimization. One such clause is that which pertains to the delivery of technically satisfactory masters. The labels could refuse the release of masters that are not optimized on the grounds that they are not technically satisfactory.

Also, Recommendation Systems (RS) framework should be transparent and open. The introduction of a framework that mandates platforms to be transparent to mitigate RS biases. This involves outlining main parameters used and allowing users to modify these parameters on their own terms and condition. The effectiveness of RS is critically dependent on the data input it trains on. Thus, it is important that the data collected is processed appropriately.

Algorithms significantly impact artistic autonomy by creating pressures for musicians to optimize their creative output for platform visibility and algorithmic discoverability. While this can lead to concerns about taste homogenization and the commodification of music as data, user agency and human curation still play a role in shaping the evolving digital music landscape. Addressing these biases requires a holistic approach that integrates technical, ethical, and policy-based interventions to foster equity and inclusivity.

Daniel-Adebayo, is a legal practitioner with specialisation in information and technology law. He sent this piece from Lagos.