Machine Learning is Transform (Music) Content Consumption

Machine Learning is Transform (Music) Content Consumption

According to a study by McKinsey, when executed correctly, personalisation can improve ROI on marketing spend by 5 – 8x and lift sales by 10%. As such, marketers now consider delivering a personalised brand experience as a top priority. In a study by the Digital Services Group, 94% of Senior Executives stated that providing personalisation is critical to reaching and retaining customers. 

The mass adoption of personalisation by marketers is in part a consequence of ‘The Netflix Effect’. This concept refers to a new form of personalisation driven by artificial intelligence and machine learning, which has been a cornerstone of the popularisation of digital-first companies like Netflix and Spotify.

By utilising this technology, these companies have moved beyond standard audience segmentation towards providing a brand experience that is truly unique. Netflix, for instance, tracks your watching habits to tailor their dashboards to each user – helping people to easily access new content that is most relevant to them.

Netflix and Spotify have helped condition people to want and expect greater personalisation from the other brands and services they interact with. In fact, 86% of consumers say that personalisation now plays a key role in their decision-making process.

The introduction of machine learning to music, online dating, news, publications, social media and other industries has considerably changed how consumers discover and consume new content.

How Pandora Transformed Music Discovery

In 2000, 80% of the music industry’s revenue came from less than 3% of the releases. Musician and composer Tim Westergren, developer Jon Kraft and tech consultant Will Glaser, founded Pandora Radio to challenge this. Their premise was simple, to transform how people discovered new music.

Their method to achieve this was to develop an online radio platform that made recommendations based upon shared musical attributes of artists and songs users’ already listened to. This initially involved manually cataloguing thousands of songs by 450 unique attributes – including instrumentation, voice, melody, harmony, form and rhythm. Using this information, Pandora could create unique stations based upon the artist, song or genre a user selected. The platform would then continue to monitor user behaviour (thumbs up/ downs, skips, bookmarks, searches and volume changes) to enhance and fine-tune recommendations.

Rather than merely regurgitating the Top 50 Chart, Pandora learns as much as possible about a user in order to create the experience that is most likely to appeal to them.

A Consumer-Centric Music Industry

The algorithm-led approach to connecting people to new music that Pandora innovated has now been adopted by a plethora of digital music services (Spotify, Deezer, Google Play and Apple Music).

Where many of these services differed, is that their model was based on, on-demand music – a feature that Pandora didn’t originally offer. This meant that music fans could listen to any music in the service library as often as they liked. This enabled the likes of Spotify and Apple Music to learn more about users listening habits – allowing them to push more recommendations and unique AI-curated playlists.

Consequently, on-demand orientated music services have risen in popularity, with the likes of Pandora having to adapt their original business model to meet new consumer demands. 

This rise of Spotify has demonstrated a major consumer shift. Rather than being pushed artists, people want to have a more active role in discovering new music. Whilst chart music still has major relevance, people are increasingly more likely to listen to it via Spotify (where they have an experience that is fully tailored to them) than radio.

More Diversified Content Consumption

In an article on The Baffler, journalist Liz Pelly criticised Spotify for diluting the variety of music that people listen to and putting too heavy an emphasis on ‘chillout’ playlists. However, there is evidence to suggest that Spotify and Apple Music have actually helped diversify music fans

According to a report by Girls Gen Z Digital media company Sweety High, 97% of Generation Z females say they listen to at least five musical genres on a regular basis. With more options to find undiscovered music, Gen Z is embracing a more diversified range of music, utilising new and traditional media channels for discovery and consumption.

Final Say 

Adopting machine-learning practices within the way you share and distribute content can be an effective way to build a stronger relationship with an audience and help boost retention. 

One of the most effective practices that both Apple Music and Spotify adopt is being habitual in the way they create playlists. Users know that every Monday Spotify will update their personal Discover Weekly playlists with new music. Consequently, it becomes part of the users’ Monday ritual to go return to Spotify to discover their new recommendations.

Within a year of launching Discover Weekly, Spotify had generated 5 billion streams from the algorithm-curated playlists. Leveraging machine-learning and user data are now crucial to maintaining loyalty and driving people to return to the service.