Today, Spotify is the number one platform for on-demand music with over 150 million users. With advanced knowledge of how to use big data, artificial intelligence (AI), and machine learning, they found their marketing-key to success that pleases both artists and listeners (Voogt, Yau, van Doorne, & Erickson, 2018). With over 40 million songs, 2 billion playlists and about 3 million artists, Spotify is a Mecca for music lovers but can also be overwhelming at times. The data-driven company logs around 1 Terabyte of data every day (Johnson & Newett, 2015). How do they manage with all this data? In an effort to turn its mountains of data into practical benefits for both their users and musicians, two features stand out from the rest: Discover Weekly and Spotify for Artists.
Connecting the Dots
With ‘Discover Weekly’, Spotify comes up with a personalized playlist of 30 songs every Monday for each individual user. Spotify combines three different approaches in this feature: collaborative filtering, NLP and audio analysis (Johnson & Newett, 2015).
This strategy includes individual’s listening habits and their music profile, (including listening history, favorite songs and playlists). Further, Spotify identifies specific genres, so called micro-genres, and creates clusters of artists. This data is compared to other users’ music profiles, or professionally curated playlists using a machine learning algorithm. Based on patterns, Spotify suggests songs that certain users listened to that are unfamiliar to other users. This principal is similar to the very familiar ‘Others also liked…’ suggestion, you might know from other sites (Pasick, 2015; Swinney, 2018).
Natural Language Processing
‘Text’ such as the lyrics, news articles or blog posts are collected on the web and compared to each other. Are certain artists mentioned together? How do lyrics differ or overlap? By answering these kind of questions, Spotify decides which songs to recommend (Pasick, 2015; Swinney, 2018).
Audio analysis is an AI based tool that basically analyses the sound of a song to identify its unique structure which then enables the tool to find songs that are comparable in structure. By understanding which audio metrics pertain to individual users, Spotify is able to finetune their recommendations (Pasick, 2015; Swinney, 2018).
Spotify for Artists
For artists, Spotify collects many valuable digital data that enables them to track their music performance. For instance, they can see which playlists translate to new fans, which songs are listened to most, but also even more detailed insights like the numbers of local listeners and followers on specific locations (Voogt et al., 2018). Not only can artists use Spotify’s digital data to acquire some knowledge due to these sorts of insights, they can also use these insights to target their preferred audience. This can lead artists and record labels to make more informed decision, for instance where to perform next. For example, by a recently added tool (‘Artist’s Pick’), artists can ‘pin’ a certain song or playlist to their page and add some text to it. Here, they often promote new songs or albums, or direct people to the locations they are performing and promote their tour dates. What makes this tool especially interesting is that Spotify does not simply provide a platform to share music on, but that they offer artists a communication tool to connect with their audience – even if they are not releasing any new music (Spotify for Artists, 2018).
So, All is Good?
Spotify is able to understand how to use data in a way that humanizes its use. By combining many different tools like NLP, collaborative filtering, audio analysis and personalization, Spotify is able to market itself as more than just a music streaming platform. From a business perspective, Spotify has grown a lot in market value and its users seem to be satisfied (Iqbal, 2019). In this sense, it seems that the cause of creating a better music platform justifies the method of collecting the data of its users and using it the way they do.
However, people may wonder whether Spotify only uses their data to predict music taste, mood and Grammy winners (Mahdawi, 2018; Solomon, 2015). Rumor has it, that Spotify might use their data to influence the music industry as it is creating an AI that can compose music for artists, that speaks to a specific audience (Titlow, 2017). This could lead to Spotify competing with major record labels and reshaping the music landscape with an unfair advantage some might say. On the user side, Spotify can use the data points stored on their users to target ads more successfully (Mahdawi, 2018).
It is also worth noting that Spotify has also received a lot of critique from artists about how Spotify’s success is largely based on their art, but they do not receive the benefits (i.e. royalties) they believe they deserve. Moreover, no concrete legislation exists yet that states how much an artist should get paid per streamed strong (Ijsendoorn, 2018). A major criticism point that actually sparked the launch of ‘Spotify for Artists’. This initiative allows independent artists to upload music to their Spotify account directly without any financial constraints, making Spotify more attractive (Payne, 2018). However, there is still plenty of room for improvement – especially in regards to the royalty distribution.
The following video explains the Spotify royalty system thoroughly:
Facchetti, C. (2017). How to hide what you’re listening to on Spotify – Tech Advisor. Retrieved March 6, 2019, from https://www.techadvisor.co.uk/how-to/internet/how-hide-what-youre-listening-on-spotify-3663111/
Iqbal, M. (2019). Spotify Usage and Revenue Statistics (2018) – Business of Apps. Retrieved March 6, 2019, from http://www.businessofapps.com/data/spotify-statistics/
Ijsendoorn, Mark. (2018, January 3). Dit is waarom Spotify zulke problemen heeft met royalty’s betalen. Retrieved February 20, 2019, from
Johnson, C., & Newett, E. (2015). From Idea to Execution: Spotify’s Discover Weekly. Retrieved February 28, 2019, from https://www.slideshare.net/MrChrisJohnson/from idea-to-execution-spotifys-discover-weekly/3-Spotify_in_Numbers_Started_in
Mahdawi, A. (2018). Spotify can tell if you’re sad. Here’s why that should scare you. Retrieved March 6, 2019, from https://www.theguardian.com/commentisfree/2018/sep/16/spotify-can-tell-if-youre-sad-heres-why-that-should-scare-you
Pasick, A. (2015). The magic that makes Spotify’s Discover Weekly playlists so damn good. Retrieved February 28, 2019, from https://qz.com/571007/the-magic-that-makes-spotifys discover-weekly-playlists-so-damn-good/
Payne, O. (2018, September 20). Spotify Will Now Upload Independent Artists To Upload Music Directly To The Platform. Retrieved February 20, 2019, from https://www.forbes.com/sites/ogdenpayne/2018/09/20/spotify-will-now-allow-independent-artists-to-upload-music-directly-to-the-platform/#359a03f26d45
Smith, C. (2019). 72 Amazing Spotify Statistics (December 2018) | By the Numbers. Retrieved March 5, 2019, from https://expandedramblings.com/index.php/spotify-statistics/
Solomon, D. (2015). Spotify Has Predicted The Grammy Winners Via An Algorithm. Retrieved March 6, 2019, from https://www.fastcompany.com/3041943/spotify-has-predicted-the grammy-winners-via-an-algorithm
Spotify for Artists (2018, May, 23). Engaging your audience . Retrieved from: https://www.youtube.com/watch?time_continue=166&v=xJ0yE88ldL4
Swinney, J. (2018). How Data Is Creating Better Customer Experiences at Spotify – Credera. Retrieved February 28, 2019, from https://www.credera.com/blog/technology solutions/data-creating-better-customer-experiences-spotify/
Titlow, J. P. (2017). Why Did Spotify Hire This Expert In Music-Making AI? Retrieved March 6, 2019, from https://www.fastcompany.com/40439000/why-did-spotify-hire-this-expert in-music-making-ai
Voogt, B., Yau, J., van Doorne, T., & Erickson, N. (2018). Spotify Marketing Strategy: a Step by-Step Guide for Artists & Managers. Retrieved February 28, 2019, from https://heroic.academy/spotify-marketing-strategy/