Spotify + The Machine: Using Machine Learning to Create Value and Competitive Advantage

Spotify’s use of machine learning is central to its strategy.

The Issue of Music Discovery

How many songs exist today? Though there’s no consensus, the order of magnitude is estimated to be in the hundreds of millions. Added to this stock are the thousands of songs released each year. Due to this sheer volume of music, listeners are challenged to discover music they like. Additionally, some listeners don’t know exactly why they like a particular song and may even prefer a broad range of genres. This is what makes stumbling upon a song or getting a recommendation from a friend so exciting. Song discovery has historically been aided by subjective sources such as DJs. In the 2000s, music streaming platforms such as Pandora relied on manual curation or tagging to drive their song recommendations.1 Though better than discovering songs by pure luck, discovery aided by manual curation and tagging is ultimately tough to scale and can’t provide truly individualized recommendations.

 

Launched in 2008, Spotify is the world’s largest music streaming service with 159 million monthly active users across 61 countries.2 At the time of the company’s initial public offering (IPO) in April 2018, Spotify generated €4 billion in revenue and was growing 45% annually. Music streaming services have experienced outsized growth compared to the music industry overall (see Figure 1). Accompanying this rapid growth is intensifying competition as Pandora, Apple Music, Tidal, SoundCloud, Amazon, and Google all fight to attract new subscribers.

 

Figure 1

Source: IFPI, “Global Music Report 2018: Annual State of the Industry”, https://www.ifpi.org/downloads/GMR2018.pdf, accessed November 2018.

 

Core to Spotify’s strategy for winning in this crowded market is its ability to provide personalized recommendations and help users discover new music, which is enabled by its investments in machine learning. In its IPO prospectus, the company highlighted this strategy stating that it will, “continue to invest in our artificial intelligence and machine learning capabilities to deepen the personalized experience that we offer to all of our Users” and that “this personalized experience is a key competitive advantage.” Given Spotify’s deep pool of data (200 petabytes compared to Netflix’s 60 petabytes)2, the company is well-poised to create competitive advantage and provide users with a continually improving service.

 

Spotify’s Machine Learning Strategy

Spotify’s strategy has consistently focused on machine learning. First, its machine-generated, personalized playlists such as Discover Weekly and Release Radar account for 31% of all listening on the platform compared to less than 20% two years ago. The company employs three types of machine learning to enhance its recommendation engine: collaborative filtering, natural language processing (NLP), and raw audio models1. Through collaborative filtering, Spotify provides recommendations to users based on the preferences of users with similar tastes. With NLP, the company scours articles, blogs, and song metadata to generate “tags” associated with each song and compares those tags with those of other songs. The company also analyzes which artists or songs are frequently mentioned along with the song in question to refine the pool of song recommendations. Through raw audio processing, Spotify is able to identify commonalities between songs through their musical elements (e.g. tempo, time signature, key). While collaborative filtering and NLP allow Spotify to point users to popular songs they may enjoy, raw audio processing allows the company to make predictive suggestions for songs with very little user awareness.

 

Second, Spotify has bolstered its strategy through several acquisitions. In 2014, Spotify acquired Echo Nest at a $100 million valuation3 strengthening its music recommendation capabilities. In March 2017, Spotify purchased Sonalytic which develops audio feature detection technology.4 In May 2017, Spotify acquired Niland, a startup which provides more accurate music search and recommendation.5

 

Finally, Spotify is exploring the use of machine learning to help artists compose songs. To do this, Spotify hired François Pachet in the summer of 2017 to be the Director of the company’s Creator Technology Research Lab. Though Pachet views machine learning as a complement to the artists’ creative process, one could envision a future where Spotify uses its machine learning capabilities to compose its own original content based on its users’ preferences.

 

Looking Ahead

It is clear that Spotify is taking deliberate steps to improve its value proposition through investments in machine learning. Collectively, “machine learning” and “artificial intelligence” are mentioned 15 times in the company’s IPO prospectus2; an indication of the technology’s importance to the company. The company should (i) continue to hire top data scientists to ensure that its recommendation engine remains best-in-class and (ii) expand its base of users and artists rapidly to widen the data set which feeds its recommendation engine.


Though it seems like Spotify is the market leader today in part due to its use of machine learning, how else can the company ensure that no other company sprouts up with a better algorithm to make personalized recommendations? How will Spotify, given its market clout, shape artists’ process of new music creation?

 

Word count: 788

 

 

References

  1. Sophia Ciocca, “How Does Spotify Know You So Well?”, Medium, October 10, 2017, https://medium.com/s/story/spotifys-discover-weekly-how-machine-learning-finds-your-new-music-19a41ab76efe, accessed November 2018.
  2. Spotify Technology S.A., February 28, 2018 Form F-1. https://www.sec.gov/Archives/edgar/data/1639920/000119312518063434/d494294df1.htm#rom494294_4, accessed November 2018.
  3. Ingrid Lunden, “Spotify Acquired Music Tech Company The Echo Nest In A $100M Deal”, TechCrunch, March 7, 2014, https://techcrunch.com/2014/03/07/spotify-echo-nest-100m/, accessed November 2018.
  4. Hugh McIntyre, “Spotify Has Acquired U.K. Music Startup Sonalytic”, March 7, 2017, Forbes, https://www.forbes.com/sites/hughmcintyre/2017/03/07/spotify-has-acquired-u-k-musicla-startup-sonalytic/#a4c0c3f6fcbe, accessed November 2018.
  5. Jon Russell, “Spotify Buys AI Startup Niland to Develop its Music Personalization and Recommendations”, TechCrunch, March 18, 2017, https://techcrunch.com/2017/05/18/spotify-buys-ai-startup-niland/, accessed November 2018.

Previous:

Turning Feelings into Data: Applying Natural Language Processing to Employee Sentiment

Next:

Airbnb: Utilizing Machine Learning to Optimize Travel

9 thoughts on “Spotify + The Machine: Using Machine Learning to Create Value and Competitive Advantage

  1. Spotify appears to be at the cutting edge of bridging art and science. In your article, you raise an interesting point regarding potential competitive threats. However, given the volume of data that Spotify has collected, is it reasonable to view this data bank as a stand-alone asset? I also wonder whether Spotify is deploying its capital most effectively in its quest to push the applications of machine learning. For example, do they generate more value by 1) assessing the validity of their existing tags (e.g., generated through NLP), or 2) investing in new forms of data collection and processing (e.g., beyond NLP or raw audio processing) to come up with new ways to tag songs? Finally, it feels like Spotify still relies on its people in order to test the validity of its tags and collaborative filtering. Do you see a world in which Spotify’s machine learning algorithms no longer need human validation/testing?

  2. I love that Spotify uses their Machine Learning capability to improve user experience and is focused on the customer, not just data mining for record companies, marketing firms, etc. I think to keep their competitive advantage they’re going to have to continue to be aggressive in their M&A strategy to find new technology before it goes to a competitor. I think they also need to stay true to their listener base on stay focused on the experience of their customer. I also think they need to be careful not to allow artists to “game the system” with their inputs into Spotify. I drew parallels to our Big Data case with Gap, and how information started to develop the same “fashion” at every store. The consumer is not a good predictor of what they’ll want in the future, so I would encourage Spotify to focus on developing and promoting smaller artists who can set the “trends” for us. I think this is a big risk to record companies as well to continue to sign creative, original artists. What would Spotify be like if everyone wrote music to optimize for number of “Discovery Weekly” playlists it could penetrate?

  3. Really interesting that Spotify is investing in machine learning capabilities to compose music, l looked into this as well. Even though Pachet has framed this effort as a complement to artists, do you think the company might face any backlash for attempting to replace the artists it depends on? Also, there are a number of other companies working to use machine learning to compose music. In particular, Google is researching this as well through its project Magenta (https://magenta.tensorflow.org/). Do you think Spotify’s data collection is a big enough competitive advantage to be the leader in machine learning generated music? I am skeptical as I imagine Google has a ton of similar data through YouTube. If Google is at an advantage here, then it seems particularly risky for Spotify to alienate artists – something they should watch out for!

  4. Great Article! As someone who is loves music but very bad at remembering artists and song names I find Spotify extremely helpful. Spotify has helped me discover artists that I would have never found on my own and has recommend more artists that I enjoy than not. In response to your open question I think it is critical for Spotify to retain the largest customer base to make their recommendation algorithm better than competitors. The model is only as good as the data it collects and if customers are not listening to songs on the Spotify platform then the model will not be able to make beneficial recommendations. One way it can continue to attract customers to their platform is to have exclusive contracts with artists – guaranteeing content will only be found on Spotify.

  5. I think your last point of consideration, another way to look at it is how big is the barrier to entry for Spotify? They have created a really strong algorithm to learn consumer preferences, but that can be copied over time or perhaps even be outperformed. Players like Apple are late entrants into the market and already proving to be serious players.(1) I think new entrants is a fear and therefore playlists and other stickiness factors are imperative if Spotify wants to stick around.

    (1) https://www.digitaltrends.com/music/why-is-apple-music-beating-spotify-in-us-market/

  6. Great article! I’m a Spotify user and fan, and I’ve often wondered how they use data and algorithms to recommend new music that I usually like. I’m curious to get your thoughts on the competitive landscape and use of ML/AI in generating recommendations. Additionally, do you think it makes sense for Spotify to partner with other companies/streaming services that make recommendations based on ML? For example, could it be beneficial for Spotify to partner with Netflix or Amazon? Are there things about music recommendations that could tell us about individual preference in other areas of life?

  7. Nice piece about a topic I’ve been curious about! The recommendation engine at Spotify, namely with respect to the Discover Weekly playlist, has been a great source for new music and consistently offers songs I like. Somewhat related to what Ian was asking, I’m very curious how Spotify can use its insight to provide value to artists. I understand Netflix has used its ML and user data heavily to create original content; if not possible to create music or hire artists themselves, I’m wondering if Spotify could at least provide insights. Maybe they can develop a new revenue stream by supplying music labels with music insights? One concern I’ve had is if the learning algorithms and listener grouping will ultimately make more unique, original music less available. In other words, will we just listen to music that other groups of folks are also listening to? How do we remain open to new music that others may not have found yet?

  8. AP – love the title of your piece. Something I find interesting about this is that the quality of Spotify’s recommendations directly impacts their bottom line. The record labels charge Spotify on a per-stream basis, meaning that Spotify wants to deliver quality recommendations on its Discover playlists rather than have users listen to multiple songs for 10 seconds each and skip them before finding something they want to listen to.

  9. As eluded to by the other comments, Spotify is a very popular application that uses machine learning to create music playlists. These playlists coincides with the demands of their user. This function has made my life a lot easier.

    One thing that I am intrigue about is where Spotify will go next in regards to product offerings (i.e, Will Spotify be able to create vacation suggestions based on someone’s background, profile and traveling history?)

Leave a comment