But Spotify’s chief R&D officer Gustav Söderström revealed an even more compelling tenet of the service’s growth strategy, one that draws inspiration from the likes of Uber, Google and Tesla in its ambitions to dominate the personalization market: “self-driving music.”
In Spotify’s case, “self-driving” is just a flashy substitution for “automated” and “context-aware”: using advanced, always-on data analysis to mold the discovery experience autonomously to an individual user's tastes and behavior, across multiple devices and environments.
Indeed, the amount of data Spotify collects on both ends of its “two-sided marketplace” is significant. On one end, there are now over three million artists on the service who release 20,000 new pieces of content every day on average. On the other end, the typical Spotify user clocks in 49 minutes a day on the service, increasing to 80 minutes for Premium users; multiply this by 159 million monthly active users across 65 markets, and that engagement generates three billion music-related events every day, including an average of over 330 million daily “discoveries” (i.e. listening to a certain track or artist for the very first time). In total, Spotify works with over 200 petabytes of data behind the scenes -- more than five times its nearest competitor, claimed Söderström -- to develop a superior personalization product.
Of course, one crucial platform for a "self-driving" music future will be the car: 50 million Spotify users, or around 31 percent of the company's user base, already interact with the service in the car, and Spotify is responding through deeper investments in partnerships with Google's Android Auto, Apple's CarPlay and similar audio integration platforms.
Beyond the car, a key component of Spotify’s “self-driving” music strategy at large will be a platform-agnostic approach that meets users where they are already active. According to its investor deck, not only does Spotify currently have over 250 integration partners, but 75 percent of Premium subscribers also use the service across multiple devices, and those who use Spotify on more than one device demonstrate lower churn and higher lifetime value for the company. Thus, unlike competitors Apple, Google and Amazon, Spotify aims to continue prioritizing better software over proprietary hardware as its key value proposition to the music industry (even as Spotify begins hiring for its first physical product line).
“We’re not just watching the home become software: we’re also watching the home become more heterogeneous, running different operating systems across different devices from different partners, rather than relying on a single platform,” said Söderström. “This is exactly why we made a bet on ubiquity … on doing what is best for the user, not for the company. We are solving the user’s problems by being everywhere.”
But will a “self-driving” music discovery mechanism really serve the artist community, in terms of connecting meaningfully with fans? In the past, artists have expressed fear and criticism of Spotify’s over-quantification and product-ization of users' listening habits, creating a different sort of gatekeeping mechanism in the industry; writer Liz Pelly described this phenomenon as “the automation of selling out.”
Part of the impetus behind this concern is the accelerating growth of mood-driven, lean-back listening experiences over artist-focused ones, which the “fake artist” allegations against Spotify thrust into the spotlight. A recent study by streaming analytics startup Chartmetric found that Spotify’s context-driven playlists (i.e. those defined around a specific activity, such as running, or around a time-specific event, such as Women’s History Month) have significantly higher median follower counts than content-driven playlists (those focused more on specific genres, languages or geographies). Year-over-year follower gain for context-based playlists also outpaced content-driven ones at 84.6 percent versus 62.3 percent, respectively.
"Self-driving music" sounds like the most extreme case of a context-driven music industry: one where lean-back consumers and tech platforms dictate the journey, while artists and rights holders sit shotgun but have little control over the steering wheel.
Nonetheless, Spotify insists not only that blended curation approaches thrive on its platform -- according to its investor deck, 30 percent of streams on the service are “algotorial,” or derived from curation that is a mix of algorithmic and hand-picked, such as the Time Capsule playlist -- but also that smart automation of discovery has tangibly benefited artists by widening the one-percent pool and by targeting prospective fans more effectively than terrestrial radio and other incumbent alternatives of similar scale.
According to head of creator marketplace Charlie Hellman, Spotify has seen a 28 percent growth in the number of artists it considers “top-tier,” from 16,000 to 22,000 artists, and strives to continue expanding this cohort into the “hundreds of thousands.”
Key tactics in widening this top-tier pool will include improving the Spotify for Artists data dashboard -- which around 100,000 artists (only 3 percent of total artists on the platform) currently use on a monthly basis -- as well as investing more in “dedicated research and innovation teams working on software to help musicians make their best work,” said Hellman, referencing groups like the in-house Creator Technology Research Lab and recently-acquired collaborative digital audio workstation (DAW) startup Soundtrap.
Hellman also described Spotify’s overarching vision for music as “a meritocracy where an artist’s success is determined by how they connect with fans, not with gatekeepers,” adding that Spotify wanted to be “like an Amazon seller or Airbnb host” in its relationships with artists and labels.
Both Ek and global head of creator services Troy Carter echoed Hellman’s sentiments in their own speeches. “It took literally a full year to get [Lady] Gaga’s ‘Just Dance’ on the radio,” said Carter, who formerly managed Lady Gaga as the founder and CEO of talent management firm Atom Factory. “Now, instead of fighting against industry gatekeepers, artists can break out in weeks, or even days.” Ek attributed this newfound success in the artist community to the wider, more open distribution of information: “We all know information is power, and it bridges the gap between the struggling and the successful. It re-balances the power.”
Neither the Spotify execs onstage nor the investors in the audience brought up exactly to what extent this alleged redistribution of power, in the form of more transparent information, will translate to more dollars and stabler careers for lower-tier artists. Evidence points to an inverse relationship between scale and per-stream royalty rates, which presents a challenge for niche artists who are not cut out to win in a purely consumption-driven environment. Plus, Spotify is facing several copyright infringement lawsuits, worth more than $1 billion, that could cost the company dearly before it even finds a clear path to profitability in the first place.
But at the end of the day, Spotify is betting that Wall Street investors care about more about scale and efficiency. “We want everyone in the industry to be more efficient, and to have more impact," Ek told the audience at the outset. "The more music people discover, the more they will listen, and the more artists will be successful. That, of course, will attract more people to Spotify.”