`1 Where Is Spotify?`

Daniel Ek, Spotify’s cofounder and CEO, has been called the most powerful person in the music business,¹ as well as one of “the most influential minds in tech.”² In 2017, he made it on Time magazine’s list of the world’s “100 most influential people” for having “helped transform the way people listen to music and the way artists interact with fans.”³ Based on what the press reports about Spotify, it would seem that Daniel Ek rules an empire and is destined to implement a personal and unique vision of technology and music. While Spotify certainly is shaping the way music is attended to, the habitual attribution of agency and power to this company and its CEO raises more questions than it answers. What kind of autonomy does Spotify possess as a corporate enterprise? What makes Spotify a Swedish success story, given the financial losses the company has sustained each year of its existence? Who owns and rules over music streaming?

Spotify’s story is often told as a narrative of how the rise of streaming caused the decline of piracy, thus saving the music industry from crisis: “From the very beginning, our vision was to offer a legal music service, as good or better than the pirate sites, giving users access to all music in the world, for free.” That was how Ek put it in 2009.⁴ Today, he would omit the last two words. As this chapter will demonstrate, however, history did not follow a linear path guided by entrepreneurial vision. Rather, we argue that Spotify is a shape-shifting service developed by a company that constantly adjusted, if not entirely changed, its main strategies and goals. Accordingly, the emergence of music streaming as a cultural practice can be historicized in more than one way. While Spotify will appear as a key actor in any historical narrative, the company has played more than one role—and not always successfully.

`The Precarity of Spotify`

The first thing to note is that the company is not owned by its founders. After founding the company in 2006, Daniel Ek and Martin Lorentzon lost their majority share by 2009. Since then—at least until its IPO in April 2018—Spotify has been principally owned by a number of venture capital firms based in different parts of the world. Their primary interest is not to make Spotify profitable but to make it valuable. Venture capital bets on return on investments at the time of an “exit,” that is, when Spotify is either acquired by a larger corporation or introduced at the stock exchange. Meanwhile, more and more venture capital is needed to cover the recurring losses and keep up the growth. So, while the executives must submit to the owners, the current owners must submit to the expectations attributed to an imagined future or the capital may be lost.⁵

However, ownership does not equal control. Spotify’s very existence remains dependent on the willingness of the so-called Big Three—the global record companies Universal, Sony, and Warner—to renew music licensing deals. In essence, these must not demand more royalties than Spotify is able to pay. The Big Three form an oligopoly that can act as a cartel when dealing with any music streaming service. Importantly, Spotify and its competitors all depend on the same product: a distribution license for what consumers accept to be “all music.” This product can be bought from only one source: the recorded music cartel led by the Big Three. With a few exceptions, Spotify and its competitors offer the same catalog of music. In what sense, then, could Spotify possess a “unique selling proposition”? Would it be possible for the Big Three to start their own music streaming service and let Spotify vanish, after having squeezed it on venture capital? The answer to these questions depends on whether one considers Spotify to be simply an intermediary or rather the producer of a new commodity—a personalized music experience.

Yet Spotify is dependent on more than venture capital and music licenses. To deliver its product to consumers, the service also necessitates a larger infrastructure, as we will discuss in greater depth in the next chapter. At the one end, there need to be data centers to send out music files and fetch back user data. At the other end, consumers need to have appropriate playback devices; ideally, Spotify would have its software preinstalled in everything from smartphones to cars. In between these two ends, there must be enough available bandwidth for music to keep streaming without interruption. To guarantee a favorable infrastructure, Spotify has closed numerous deals with data center operators, hardware manufacturers, and telecom carriers.

None of this is unique for a music streaming service. The dependence on venture capital and infrastructure operators is a characteristic feature of most technology startups that intend to become platforms. As we have argued in the introduction, current platform discourse tends to overstate the power that is in a platform, downplaying the variety of positions these companies may take on markets. Spotify’s current dominance in music streaming does not simply mirror the market dominance enjoyed by other platforms, such as Facebook. The latter is designed to facilitate forms of exchange between users, not to distribute copyrighted works. The opposite goes for Spotify, which makes its survival more dependent on licenses. Copyright also works differently in different media. In music, copyright protects the musical work of songwriters and lyricists and allows for these rights to be monopolized by collecting societies. But there is also a separate right to the sound recording, which usually belongs to a record company. This legal construction was historically motivated by the high costs of recording sound. While that circumstance has changed, the record industry cemented a very strong market position thanks to these “related rights.”⁶

Ever since the tape recorder was introduced, the development of consumer electronics has tended to gradually undermine the enforceability of these rights. This process accelerated in the digital realm with the proliferation of file-sharing networks such as The Pirate Bay, also based in Sweden. The record industry was thrown into a crisis—to which Spotify could present itself as a solution. Spotify’s early history is therefore entangled with (Swedish) politics, and this chapter will demonstrate the service’s ambivalent relationship to music piracy.

Spotify, in other words, relies not only on investors, rights holders, and network operators but also on policymakers. Music piracy must not be too rampant, collecting societies too powerful, or privacy protections too strict if Spotify is to succeed in a country. Rather than being an autonomous actor with the power to shape the future of the music business, Spotify exists at the intersection of industries such as music, advertising, technology, and finance.

Accordingly, the history of Spotify is not just a story of success but also a story of precarity. To keep afloat, and to attract new venture capital that can cover its losses before an acquisition or IPO, Spotify has to sustain the hype around its service, framing it as a lifestyle for users and an economic opportunity for artists. Spotify also has to maintain the image of a company that is always expanding, looking forward, and headed on the straight path toward a future monopoly position. In each country, Spotify has had to adapt to different patterns of music use and make deals with different partners. All these interactions have contributed to how Spotify has changed its interface, away from the centrality of the search box and toward the promise of “the right music for every moment.”

The question has often been asked whether Spotify has a sustainable business model or not. We do not aim to answer that question here. The question itself seems inadequate, as no business can sustain itself on a market. Rather, it must be sustained by others, as the following historical narrative of Spotify will demonstrate.

`How to Historicize Spotify?`

What is Spotify? First, it is the name of a company. Second, it is the name of the music service provided by that same company. Third, it is quite common to invoke Spotify as a metaphor to describe or evoke broader transformations of the media industries and the internet, as when trade journals describe other tech startups as a “Spotify for x.” Such metaphorical invocations of the service convey an image of what the internet is or might be, provide clues to design intentions, and carry normative connotations.⁷ In addition, the company, the music service, and the metaphor have all changed over time. A decade is a short measure of time in business history but not when it comes to digital media.⁸ To provide a rationale and context for the methods and research presented throughout this book, it is necessary to historicize the present and past futures of Spotify. This counters a trend of presentist and functionalist accounts that emphasize seemingly emergent or unforeseeable developments.⁹

Periodization is unavoidable in any historiography or social negotiation of which continuities the present will recognize in its past. Anyone who conveys a historical account—be it a historian, a journalist, or a marketer—has to periodize in one way or another and, thus, to engage in a “politics of time.”¹⁰ The story of Spotify has often been told in the press and in academic publications, with different degrees of detail. It has also been conveyed by Spotify itself.

“Spotify—The Story” is the title of a one-minute video uploaded by the company for the public launch of its service in 2008. In this video, music streaming is presented as the culmination of a media history that developed toward the maximization of individual choice. The voiceover relates:

In a world where everyone loves music. First came vinyl, the cassette tape, the compact disc, the MP3 player. Which brings us to the present day. Introducing Spotify, a world of music. You search, you find.… Whatever you want, whenever you want. Instant, simple, and free.¹¹

In this succession of audio technologies, the absence of radio is striking, as we will discuss later. Spotify’s periodization of media history is also entirely technological, which parallels most accounts of the company’s rise to fame. These stories usually highlight the same breaks in time: “from physical to digital; downloads to streaming; ownership to access.”¹² Streaming appears here as the name of a new paradigm in history. But what about the transformations within streaming and with regard to the changing expectations of what a streaming service should provide? In order to answer these and other questions, we will have to introduce an alternative principle of dividing time that is based not on technology, nor on the ups and downs of the music industry, but on financialization as a structuring principle of media history.¹³ The main question asked in this chapter is how Spotify has managed to receive new funding for running its operations at ever larger losses for over a decade.

This chapter brings together media history and business history, integrating the global (technology and finance) with the local (politics and culture). As a media history, it picks up where most research on digital music has ended. In particular, it takes its cue from Jonathan Sterne’s MP3: The Meaning of a Format and Jeremy Wade Morris’s Selling Digital Music, Formatting Culture, both of which situate the affordances of software in a context of commodification.¹⁴ As a business history, this chapter responds to the challenge of studying digital companies that are entangled in both public and private spheres but are unwilling to grant access to their archives. While eager to finance applied research to promote and develop its own service, Spotify has rejected any request to make historical documentation accessible.

What does exist abundantly, however, is hype about Spotify: speculations, rumors, and promises widely circulated by the press, by Spotify’s own marketing affiliates, and in social media. In writing the history of Spotify, a major issue concerns how to approach this abundance of materials. Rather than discarding hype as noise to be filtered out, we understand media industry realities as always constructed, which requires that one pay attention to an industry’s own outward self-representation and dismiss the belief in an authentic reality “inside the box.”¹⁵ This interest in hype as a source of historiography resonates with Gideon Kunda’s observation that a high-tech company’s “engineering culture” is not to be understood as something substantial but as a “mechanism of control” that works normatively both on its employees and its future customers.¹⁶ Yet hype is more than just the outward construction of a company’s culture; it also relates to the ways in which human agency is structured by time. Hype is a phase usually characterized by an upsurge in public attention and high-rising expectations about possible innovation. Hype comes first, modeling the future out of the present. Thus, it is interesting to see how plans and goals related to Spotify were constructed from its existing culture in a process of modeling and what came after each period of hype.¹⁷

In adopting a framework of following the hype, or a history based on widely accessible sources, this chapter mirrors this book’s overall rationale of following files. It works through vast source materials in order both to develop an empirical narrative of change and to look for what Richard Barbrook has called “the beta version of a science fiction dream.”¹⁸

The main part of the material used here stems from online archives. Keeping in mind that Spotify exists at the intersection of industries, we have chosen to consult trade journals from the diverse fields of technology (e.g., Wired, TechCrunch), music (e.g., Billboard, Music Week), and advertising (e.g., Advertising Age, Marketing Week), as well as some general news outlets. In addition, we have found Swedish press archives to be indispensable, particularly for tracing the earlier parts of Spotify’s history.¹⁹ Besides the press, there are a number of relevant web sources that are not accessible by search engine but rather require more complex forms of retrieval, such as scouring old job listings, excavating deleted blog entries, and mining historical Twitter accounts.²⁰

It is much more difficult to study the software itself and how it worked in previous versions. The source code is not made public, and key functions are not executed on the local machine but are dependent on interaction with central servers. To historicize the software interface, then, we have largely had to rely on screenshots and scattered user reports. During the time period that our research project has been running, however, we have done real-time documentation of the software. As we hinted at earlier, none of these sources will give substantial insight into the inner workings of Spotify—information that business historians traditionally consider the most valuable. The external relations of Spotify are more discernable, while key documents are usually guarded by nondisclosure agreements, including the deals made between Spotify and the record industry.

The following narrative is structured chronologically in accordance with the funding rounds that have allowed Spotify to persist in making losses. While media historiography has largely abandoned such linearized narrative accounts, the format of corporate storytelling adopted here mirrors the significance of “narrative infrastructure” in industrial organization and product development processes. Narrative accounts are frequently produced to create coherence in multiactor, multilevel processes and to reduce complexity and uncertainty.²¹ In line with the interventionist design of this book, we turn this practice against itself by showing how the narratives developed around Spotify resonate with its financial history. Spotify would not be valued at several billion dollars—and would not have had its losses repeatedly covered by venture capital—if there were not a story connecting its open-ended past to a certain and positive future.

Spotify’s dependence on financial speculation thus provides the reason for this methodological choice to follow the hype. The first financing round (Series A) of about $20 million coincided with Spotify’s public launch in 2008. Subsequently, larger amounts of capital have been injected, typically at an interval of twelve to eighteen months. For each round, the existing stockholders have seen their share of equity being diluted, so that the balance of ownership has been displaced in a new direction but always away from the founders, who were also the original investors. A limited amount of financial information is easily available on websites such as Crunchbase: the identity of the investors in a round, the total size of their investment, and the resulting valuation of Spotify. But in each deal, there are also secret conditions that entitle Spotify to act in a certain way. These are harder to discern. In recent years, Spotify has taken in billions of dollars in the form of convertible debt, at terms that create a strong incentive to rush toward a stock market launch. By following the hype and mapping it over an investment timeline, we aim to underline how Spotify’s investors have influenced its development.

`How to Count?`

There are many numbers that document Spotify’s growth and are often used to compare the company to its competitors, but it would be a mistake to take these numbers for hard facts. Not only is the financial valuation of the company highly speculative, as it tends to be biased by deals that guarantee some investors a more favorable “exit,” but there are also good reasons to remain skeptical about other numbers invoked in the comparison between music streaming services. While Spotify currently claims to offer access to thirty million “songs” (as of January 2018), it is difficult to differentiate duplicates and nonmusical sounds from actual songs (as our discussion of “zombie music” will assert in chapter 2). User statistics circulated in the press may refer to the number of registered accounts or to those accounts that have been active within a given day, week, month, or year. What counts as activity here is not clearly defined; does the user have to play a song or just open the app? Furthermore, an account is not the same as an individual user. One individual may use several accounts, and one account may be shared within a household of several individuals. Therefore, more than just a grain of salt is needed when reading user statistics. In short, there is no way of knowing if 150 million alleged users (as of 2017) does really represent 150 million individuals. Counting the number of paying subscribers is a bit less problematic. But even here, the figure may well be boosted by giving discounts or by bundling the subscription price together with a mobile service, for instance.

Even more problematic are widespread claims about how much money Spotify generates for artists. First of all, these aggregate numbers show neither the distribution between record companies, songwriters, and performing artists nor between the superstars and the rest. Few industry insiders, however, dispute the claim that the Big Three have used their influence over Spotify to secure better royalty deals than their smaller competitors. They also get free advertising space and special access to user data. In addition, the Big Three are part owners of Spotify, since they have managed to get shares in return for licenses. All of this makes it difficult to discuss the issue of fair remuneration for artists.

Each year, statistics for the “global music market” are published by the International Federation of the Phonographic Industry (IFPI). These numbers are routinely used by journalists and academics as evidence of the increasing significance of streaming. “It makes up the majority of digital revenue, which, in turn, now accounts for 50% of total recorded music revenues,” the most recent IFPI report states.²² Such industry statistics have to be treated with skepticism. Aggregates such as “digital revenue” are defined in an arbitrary way, given the fact that virtually all recorded music today is digital, at least for parts of its journey from artist to listener. Even a vinyl record is almost always produced from a digital master track; that being said, even a digital stream must become analog when it makes its final journey from the loudspeaker to the listener’s ears. In the widely cited IFPI statistics, digital revenue includes royalties from Spotify, but not from “broadcasters,” although the latter category also includes pure internet services. Record industry statistics thus maintain an obsolete distinction between retail and broadcasting, with Spotify categorized as a retailer.²³ Industry statistics are just as arbitrary in their attribution of costs and revenues to national territories; the recurring claim that Sweden would be “the world’s third-largest music exporter,” for example, is presumably a misleading statement.²⁴ We should similarly be skeptical of the report that Spotify is now profitable in certain national markets, including Sweden.²⁵ While revenues from subscriptions and advertisements may come from national subsidiaries, Spotify is a company to which many costs—such as infrastructure and tech development—are of a global nature. It is therefore difficult to judge its profitability within a given country.

Spotify has also been associated with problems of quantification at a macroeconomic level. The company has become a prime example among prominent economists who claim that digital services are not adequately captured by established ways of estimating economic growth, inflation, and productivity.²⁶ These depend on the measurement of prices over time, but how is one to compare the price of music in 2015 to that in 2005? Is it meaningful to calculate a hypothetical “price per track listened to” when the listener is just paying a monthly fee? Should the price of using Spotify be compared to the price of buying records or to the (nonexistent) price of listening to the radio? If the supply of music in Spotify’s catalog increases, should this be counted as an increase in the quality of the service?

It could also be argued that the very abundance of music is a problem, one which first must be solved by some kind of curation. If that is the case, how can we calculate the hypothetical value of a music recommendation system? Economists Erik Brynjolfsson and Andrew McAfee have therefore pointed to Spotify as an example of how national accounts fail.²⁷ But it would be just as easy to make the opposite argument: intensified surveillance and advertising means consumers are paying a price for their data and their attention that should be made visible in national accounts. Even more traditional calculation techniques, which focus on transactions where actual money is changing hands, run into difficulty locating the financial flows around Spotify. Recent government inquiries from Sweden and the UK have singled out Spotify as epitomizing the problem of measuring an economy built on digital services.²⁸

`The Technopolitical Context`

Founded in 2006, Spotify emerged from a particular era in internet history. Mobile connectivity, for instance, was not the norm at the time but still the exception. Digital culture was not yet a streaming culture premised on constant connectivity but rather a storage culture dependent on ever-larger, ever-cheaper hard drives. In 2006, algorithmic culture was largely unheard of; rather, a culture of data storage and everyday archivism prevailed. Apple’s music player, the iPod, may serve as a case in point: launched in 2001 with a storage capacity of 5 GB, the iPod’s storage rapidly increased and peaked in 2007, when Apple produced a 160 GB model (approximately twenty thousand songs).²⁹

Following the burst of the first dot-com bubble, the early 2000s were also a time when venture capital was at low ebb. This is not to say that innovation slowed down. Rather, it changed character, becoming less commercial and more decentralized. Three of the most characteristic innovations of the time, however, took off without the support of investors or any company in control and were based on open standards: the blog, the wiki, and the BitTorrent protocol. After some years of incubation, these forms of networking soon came to represent the future of the internet—a future that was already known in tech circles as Web 2.0.³⁰ By 2005, the term was well established. That same year, a political controversy over file sharing and copyright also erupted around The Pirate Bay, the world’s largest BitTorrent search engine. These different digital contexts resulted in the formation of a new political party in Sweden, the Pirate Party, which made it into the European Parliament during the party’s Swedish heyday (2006–2010) and subsequently spread to several other countries.

The file-sharing boom and the political dispute over copyright enforcement were both exceptionally intense in Sweden.³¹ They gave extra momentum to anyone proposing that the media industries should disrupt themselves and switch to “new business models,” which tended to rely on ideas put forward by Wired editor and business guru Chris Anderson, author of the The Long Tail (2006) and Free! (2009).³² We would argue that it is far from certain that Spotify would have survived if not for its opportunistic, rather than innovative, strategy and the inconsistent ways in which the company positioned itself in regard to file sharing: sometimes presenting itself as the continuation of the ongoing illicit disruption, while at other times insisting on a binary opposition between illegality and legality. In retrospect, Spotify’s early Swedish history appears in perfect sync with the legal proceedings against The Pirate Bay, which began with a police raid in May 2006—just weeks after Spotify was founded.

`Finding the Files to Distribute`

Neither of Spotify’s two founders, Daniel Ek (born in 1983) and Martin Lorentzon (born in 1969), had experience working with music, neither in the music industry nor even as subcultural enthusiasts. While it is often reported that Ek played the guitar as a child, there is hardly any evidence that music defined his social life as he grew up.

The early history of Spotify—at least as it has hitherto been told—has thus been made more musical than it probably was. In interviews, Ek has explained that he started Spotify for two reasons: to save the music industry from piracy and to help friends with “really bad taste in music” to “discover better music.”³³ Spotify’s actual history is less straightforward. Swedish sources reveal a haphazard search for business opportunities that just happened to involve music. Indeed, given Spotify’s later development, it is telling that both Ek and Lorentzon had a background in the business of advertising technology. Lorentzon had made a fortune with the company TradeDoubler, which he and Felix Hagnö founded in 1999. TradeDoubler survived the first dot-com crash and soon expanded to become one of Europe’s leading affiliate marketing networks, going public in late 2005.³⁴ A few months later, in March 2006, TradeDoubler acquired a small startup called Advertigo, which developed a technology for so-called contextual advertising, from founder Daniel Ek.³⁵ This is how Lorentzon and Ek met: two Swedish multimillionaires—thirty-seven and twenty-three years old, respectively—both experiencing the boredom that may come with financial independence, as they would later explain. The two decided to start a business together.³⁶ It began to take form during the summer of 2006. That July, the new company was named Spotify—a name without any particular meaning.

Spotify was initially structured as a group of several firms that used a plurality of jurisdictions for owning different assets. Apart from Spotify AB, which is registered as a software company, there is Spotify Sweden AB, which has the official purpose of selling advertising connected to “services for digital distribution of music, film, TV programs, audiobooks, games, and similar content.” In addition, for each country where the service has been launched, new subsidiaries have also been launched. They are all, however, owned by a holding company, Spotify Technology SA, which is registered at a post-office box in Luxembourg. This post-office box is the same one that Lorentzon uses for another holding company, which in turn owns another company registered to a post-office box in Cyprus, in which Lorentzon has placed the financial riches that he made with TradeDoubler and through which he now owns his shares in Spotify: a classic structure for tax avoidance.³⁷ Ek also owns a Cypriot-registered investment company, D. G. E. Investments,³⁸ and Cyprus is furthermore the virtual home of two subsidiaries that were already registered in 2006: Spotify Technology Sales Ltd. and Spotify Technology Holding Ltd.,³⁹ the latter being the original registrant of Spotify’s patents and trademarks.

Spotify opened its first office at Riddargatan 20 in central Stockholm, where software development begun in August 2006. The software was initially for distributing data over the internet from a central server to a large number of recipients using a peer-to-peer (P2P) network, so as to unburden the central server. In other words, the core idea was to minimize the cost of digital distribution by using extra bandwidth available among the server’s users. The kind of data to be distributed was, for Spotify, a secondary consideration at this point. “The media may here represent any kind of digital content, such as music, video, digital films or images,” explained the US patent application that Spotify filed in 2007.⁴⁰ Speaking with the business press, the founders presented Spotify as a general “media distribution platform,” indicating that the ultimate aim was to use it for video distribution. However, video demanded much more bandwidth than sound. “We begin with music streaming this spring,” Lorentzon therefore explained.⁴¹ Visitors to Spotify’s first website were consequently met with the following presentation: “Spotify gives you the music you want, when you want it. Your choice is just a search box or a friendly recommendation away. You’ll be amazed by the speed and control you have with Spotify.”⁴² To try out the technology, Spotify loaded its servers with the music files most easily available, namely those already stored on its employees’ private computers. A large portion of these files were downloaded through file-sharing services such as The Pirate Bay.⁴³ Rights holders had not granted the company the licenses required to distribute the files online. Thus, Spotify began as a de facto pirate service.

`The Beta Period (2007–2008)`

On May 1, 2007, Spotify released its initial beta version to a small circle of acquaintances. Among them were some of Sweden’s leading technology bloggers. This immediately resulted in a number of enthusiastic blog posts, and the comment fields were flooded with requests for invitations. Being invited to use Spotify became a sign of exclusivity, and Spotify controlled the growth of the circle by rationing the number of invites that existing users could pass on. If one were to look at the early user demographic, it would probably be rather affluent and dominated by men between twenty-five to forty years of age who lived in Stockholm’s city center and worked in technology or media. Many of these early users saw themselves as passionate fans of new pop music, but their enthusiasm for digital technology was probably even stronger.

One of the first beta testers was Eric Wahlforss, a part-time musician and entrepreneur who would shortly go on to found SoundCloud, another music streaming service (which, years later, Spotify would consider acquiring). Wahlforss immediately recognized Spotify as “a preview of the future.” It may be noted, however, that his enthusiastic blog post about Spotify had little, if anything, to do with music: “The thing that wows me the most is that the app is faster than iTunes on my local machine. Repeat, faster than iTunes. And now we’re talking fancy peer-to-peer architectures, special audio codecs, custom databases, etc., etc.… If this thing scales it will be bigger than Skype. Big, big ups to the Spotify team.”⁴⁴ Spotify’s first interface did indeed look very similar to iTunes: the user could search for music and add tracks to personal playlists. In addition, it also offered a radio-like mode of listening in which the user was asked to select one of eighteen predefined music genres (and one or several decades). Compared to today’s personalized radio stations, this seems primitive indeed, but in fact, this was how Spotify Radio worked until late 2011.

During its beta period, Spotify consolidated a kind of on-demand doctrine as a service centered on the search box, giving access to “whatever you want.” The user was effectively conceived of as a sovereign individual, who already knew exactly what he or she wanted to listen to and did not need help with music recommendations. This doctrine was reinforced by the sample of beta testers.

The hype intensified in early 2008, with a focus on two claims. First, the new service would “make music free” by relying entirely on advertising for revenue. Second, that the move “from ownership to access” was indeed happening and that the personal archiving of MP3s would soon become an outdated practice. If commentators saw anything lacking in the beta version, it was that it did not yet live up to the promises of a Web 2.0 service, given that users still could not easily upload their self-made music. This, however, was a feature that many expected to see implemented soon.⁴⁵

`Period A (2008–2009)`

The first public version of Spotify was launched in October 2008. But what exactly did it mean to “launch” a music service that was already up and running, serving thousands of users each day? Maybe it is better to say that Spotify was legalized. For a year and a half, Spotify’s beta had, in effect, been run as a pirate service, distributing music to invited users without any license to do so. As previously stated, in many cases, the music originated from The Pirate Bay and other file-sharing networks, but this changed when Spotify started to sign deals with record companies and collecting societies, moving itself into legal territory.

For many of the existing users at the time, what happened in October 2008 was that parts of their playlists suddenly became unavailable. Spotify had to remove unlicensed music from its service, and in early 2009, even more music disappeared at the request of record companies that enforced country-specific listening restrictions. In addition, only users with a special invite could access Spotify Free; this did not change with the official launch. The only real addition was a paid version of the service, Spotify Premium. Being legal, the company could now also begin to really sell advertising, which gradually became ever-more present for users of Spotify Free.

Spotify’s launch, in other words, was not the launch of a new service but the launch of new efforts to monetize an existing service. Spotify began expanding into new territories, with different versions of the service being offered in different countries. An October 7, 2008, press release from Spotify announced that the service was being launched in eight European countries: Finland, France, Germany, Italy, Norway, Spain, Sweden, and the United Kingdom.⁴⁶ Around the same time, Spotify also opened offices in London, Berlin, and Madrid.⁴⁷ Soon, however, the company encountered problems in Germany and Italy—partly because of local copyright collecting societies—and initially had to retreat from these markets. The official story now told by Spotify is that the service was not launched in Germany until March 2012 or in Italy until February 2013. The publicity around these later launches did not mention that these were really second attempts.⁴⁸

Spotify’s initial European launch—just like the US launch three years later—marked the conjunction of two interdependent business deals. One deal granted Spotify access to music licenses, while another gave it the money to pay for those licenses. With over $20 million in venture capital, Spotify could now run its operations for a year before a fresh funding round. The leading investor in the first round was the Swedish firm Northzone, which received an 11.9 percent stake in Spotify as well as a seat on the board. Later, information leaked that the biggest record companies had signed a deal that guaranteed not just a certain level of royalties but also equity in Spotify, amounting to a cumulative 17.3 percent share, a proportion which has since diminished with each new funding round. The fact that major record companies own shares in Spotify has complicated every discussion about fair compensation for artists, songwriters, and independent record labels. The dilution of shares in this round did not alter the basic balance of ownership. Throughout Period A, the absolute majority of shares was still held by Spotify’s two cofounders, Ek and Lorentzon.

These initial conjunctions of various deals—later known as “the launch”—indeed confirm Spotify’s place at the intersection of different industries: music and technology, advertising and finance. The timing is also notable: Spotify’s launch happened to coincide with a global financial meltdown. It occurred only weeks after the bankruptcy of Lehman Brothers, precisely at the time when business magazines were busy producing front pages about the possible end of capitalism. The financial crisis was followed by the so-called Great Recession, with immediate consequences for media industries. However, it was not until the second half of 2009—Period B, in this context—that this economic downturn had a visible impact on Spotify.

In strategic terms, Period A was characterized by a focus on mobility. Spotify still had no mobile application, but in early 2009, it recruited many new developers to begin creating an app for the iPhone. Whether this app would then be accepted by Apple was still not certain.

During the same period, Spotify’s center of gravity was shifting away from Sweden. Not only did the number of British users soon surpass the Swedish, a newly opened London office became Spotify’s new company headquarters, even though tech development was still based in Stockholm. Accordingly, during 2009 and 2010, the international press consistently presented Spotify as a British company. Patent registries confirm this. From these, one can learn that Spotify’s first patent—registered by the Cypriot subsidiary in July 2007—was transferred in October 2008 to Spotify Ltd. (UK), only to be transferred again in June 2011, this time to Spotify AB (Sweden).⁴⁹ Hence, for almost three years—Periods A, B, and C, in our periodization—Spotify was a British and not a Swedish company, despite often being portrayed as a Swedish success story.

`Spotify as a Metaphor`

During 2008, the year of its launch, Spotify was widely hyped in the Swedish press. The company was frequently described as a revolutionary service offering immediate access to every conceivable kind of music. In exchange for just a few advertising jingles, all this music remained available for free. From early 2009 onward, many commentators also began to tout Spotify as a model for transforming the digital distribution of cultural goods in general. This idea was first promoted by Bredbandsbolaget, a major Swedish internet service provider (ISP), which had already struck a deal to offer each customer an invite to Spotify Free. Interviewed in January 2009, the CEO of Bredbandsbolaget proposed that there should also be “a Spotify for movies,” which would arguably be “the most effective way to combat illegal file sharing.”⁵⁰ At the time, there was no company plan to offer such a service in Sweden, but the vision caught on in the public conversation, mainly because it was politically appealing at that time.

The first half of 2009 came to mark the climax of the disputes over copyright enforcement in Sweden. The criminal case against the founders of the Pirate Bay search engine, which had been in preparation for several years, finally went before the district court of Stockholm in February 2009. With these court hearings still ongoing, the Swedish parliament voted in favor of implementing an EU directive that would open new possibilities for copyright holders to sue individuals suspected of illicit file sharing.⁵¹ Later that spring, the court announced that the four defendants connected to The Pirate Bay had been found guilty of abetting copyright infringement, and each one was sentenced to one year in prison and ordered to pay a total of €3 million in damages.⁵²

Press coverage of these events was largely “unfavorable to the positions taken by rights-holders,” the US embassy noted in a confidential cable later made public by WikiLeaks. For the Swedish government, the “political sensitivities” pervading issues of copyright were at this time “very delicate,” according to the embassy cable.⁵³ Leading politicians were clearly uncomfortable in taking sides, and when pressed, many tried to avoid the question by simply referring to Spotify as the solution. The US embassy explicitly mentioned the risk that strong media attention to the file-sharing issue might give “the Pirate Party a boost in the EU Parliamentary elections in June 2009.” This is exactly what happened: after receiving 7 percent of the Swedish vote, the Pirate Party entered the European Parliament. While political developments in a country such as Sweden might come across as marginal, they form an important part of Spotify’s early history. Were it not for the political conflict over file sharing, Spotify’s trademark would clearly not have received so much attention in the Swedish mass media.

In July 2009, news broke that Sweden would suddenly get its “Spotify for movies.”⁵⁴ The discussion that followed illustrates the metaphorical meaning of Spotify at that time. In collaboration with Bredbandsbolaget, Voddler launched a beta version of its film service, with the backing of 100 million kronor of venture capital investment (approximately $10 million). The timing coincided with the annual Almedalen Week, during which virtually all of Sweden’s politicians, lobbyists, and political reporters convene on the island of Gotland for seminars and free drinks. Only a few weeks had passed since the electoral success of the Pirate Party, making file sharing one of the hottest issues debated during the week. Consequently, the major Swedish tabloid Expressen published an editorial hailing the Spotify/Voddler model not only as “the way out from the file-sharing quagmire” but as potentially even pointing toward “the way out of the financial crisis.”⁵⁵

By its premiere date, Voddler had hoped to offer free streaming of Hollywood blockbusters. It was an aspiration that would turn out to be unrealistic given the movie industry’s territorial licensing and distribution practices. The allegedly universal model exemplified by Spotify and Voddler was characterized by the combination of three features: streaming on demand, P2P networking, and offering content at no cost to the consumer. Of these three aspects, the latter two pointed to a continuity from file-sharing networks such as The Pirate Bay to new streaming services such as Spotify and Voddler. The solution to mass pirating was supposedly found in the file-sharing experience: adding advertisements and using the resulting revenue to pay copyright holders.

`Period B (late 2009)`

In October 2009, Daniel Ek appeared uncharacteristically humble. Offering “a few thoughts on the past year,” Ek’s post on Spotify’s blog read as an exercise in self-criticism:

I care more than most about figuring out a revenue model that doesn’t devalue music … Someone asked me a while back, during a fireside chat, what was the biggest mistake I’ve made so far with Spotify? I can’t recall my answer, but I’ve since thought more about the question. I would say that the biggest mistake that I’ve made is that Spotify, unlike any of the other businesses I’ve been a part of, depends on our partners (artists, composers, labels etc.) and I haven’t always acted with this fact at the forefront of my mind.⁵⁶

Ek’s statement should certainly not be read as a personal blog post directed at a general audience. Rather, its publication appears to be part of an ongoing negotiation over music licenses, indicating that at least parts of the record industry were still suspicious of Spotify. Ek’s public demonstration of remorse over past arrogance was hence a tactical move, probably well grounded in discussions with investors who had just agreed to provide Spotify with a second round of capital to stay afloat.

Until late 2009, Spotify had been regarded primarily as a free music service, based on the founders’ vision that advertising can make music “free but legal.” This vision died over the course of 2009. The reason for this—and for Ek’s newfound humility—can be found in the global financial crisis. The Great Recession of 2009 effectuated a steep downturn in the total market for advertising. In this context, it became difficult for Spotify to convince investors and license holders of advertising support as a main source of income. As is well known, newspapers were hardest hit, which may well explain why journalists at the time quickly adopted a more skeptical stance toward the concept of free—in contrast to the uncritical reception that Chris Anderson’s ideas had enjoyed before the recession.

The financial crisis also affected the availability of venture capital. During 2009, it was reported that digital music services had a harder time attracting investors. But the downturn was only temporary. Central banks responded to the financial crisis by lowering interest rates and launching programs of quantitative easing, effectively forcing investors to chase greater risk. In brief, risk was sought in two main areas: among internet services and in emerging economies outside the West, especially Brazil, Russia, India, and China—the so-called BRIC countries. In mid-2009, the Russian investment fund Digital Sky Technologies (DST) made a $200 million investment in Facebook, marking the beginning of a new boom in the valuation of tech startups associated with social media.

How did Spotify fit into this new business climate? A growing number of commentators began to criticize Spotify’s idea of free music. Meanwhile, a new consensus regarding the advantages of subscription was forming within the media industries. In the United States, subscription-based television services, for example, fared remarkably well through the recession. For a giant such as Time Warner, growth in subscription revenues more than compensated for shrinking ad revenues during this period. Subscription services appeared to be recession-proof.⁵⁷ The record industry took notice and began to promote the subscription model as the way forward. “We’re bullish on subscriptions,” a Warner Music executive explained in late 2009.⁵⁸

As we have previously discussed, storytelling about Spotify is an integral part of Spotify itself. The narrative includes claims about continuity, as well as changes, in business models. “Spotify’s business model is, and has been since launch, more about the mix of subscription and ad supported,” Daniel Ek blogged in October 2009.⁵⁹ This may be true if one understands launch as referring to the business deals made a year before. But as we have argued, it is not an accurate description of Spotify’s original business model. The idea of selling subscriptions was rather forced upon Spotify by the holders of music licenses, and the economic crisis exerted an even stronger pressure in that direction. It was expedient at this time, however, for Spotify to modify its own history—and to distance itself from the pirate heritage. In the same blog post, Ek also promised that Spotify would not only offer music streaming but would also sell individual downloads on a pay-per-song basis.⁶⁰ This marked a break with the earlier claim that streaming was the future of music and affirmed the continuity between iTunes and Spotify as purveyors of “legal music.” Download sales were integrated into the Spotify client, at least temporarily, which was certainly not Spotify’s own idea but rather a demand from the record industry.⁶¹

In hindsight, it is striking how reluctant Daniel Ek was to affirm the idea of building his company on subscriptions. Instead, he repeated the claim that Spotify would become profitable by combining a multitude of revenue sources: “The new business model in music is a mix between ad-supported music, downloads, subscriptions, merchandising and ticketing.”⁶² There were indeed far-reaching plans to make Spotify a marketplace for concert tickets, band T-shirts, and other merchandise.⁶³

A clever way to introduce a new subscription business is to bundle the monthly bill together with another one that the customer is already paying. This was the reasoning behind Spotify’s strategy of seeking deals with telecom operators. The aim was to package the music service together with mobile services. The consumer would not really notice the payment, and the service would hence “feel like free.” Consequently, Spotify made two such major deals: with Telia in Sweden and with Three in the United Kingdom.⁶⁴ At the same time, in October 2009, Spotify also began to buy advertising. Previously, marketing had relied on word of mouth, fueled by the sharing of invites to the free service. Now a television ad appeared on Swedish channel TV5. To one of Spotify’s earliest employees, lead designer Rasmus Andersson, this signaled a change in the spirit of the company: “Yeah, it’s true that we’ve become a Big Company™. Scary, a bit sad but mostly exciting.”⁶⁵ If nothing else this indicated a tension within company culture, at precisely the time when the original founders were about to become minority owners in their own company.

`Period C (2010–2011)`

The year 2010 and the first half of 2011 were a frustrating time for Spotify, as it waited for an opportunity to launch its service in the United States. The company was not exactly at a standstill, as it continued to expand its European user base. Revenue—now mostly from subscriptions—did multiply from 2009 to 2010, but all earnings were swallowed by additional costs, which grew at an even faster pace.⁶⁶ Industry analysts agreed that Spotify needed to scale. Everything was set for its US launch, but the Big Three record companies were reluctant to allow free on-demand streaming in the American market. Warner Music, in particular, refused to grant Spotify the necessary permit. “Free streaming services are clearly not net positive for the industry,” Warner’s CEO declared in early 2010.⁶⁷ Conversely, the possibility of launching in the United States without the free tier was not an option for Spotify. The stalemate was not resolved until after Warner Music was acquired by a Russian businessman in 2011.

Calling 2010 a lost year in the history of Spotify would be an exaggeration, but this period does stand out for its slow expansion and relatively low level of press coverage. Indeed, the amount of money invested in Spotify’s Series C funding round was also small: just $16 million,⁶⁸ compared to the two preceding rounds of $22 million and $50 million. The new money was in fact less significant than the identity of the investor: Sean Parker, the cofounder of Napster. As Parker took a place on Spotify’s board of directors, this seemed to emphasize a certain continuity in which Spotify represented the fulfillment of the very same disruption that Napster had started.⁶⁹ More importantly, however, Parker had also been the founding president of Facebook and still played an informal but important role in its leadership. He was in a good position to forge a partnership between Spotify and Facebook, which would result in far-reaching attempts to integrate the two services. By early 2010, just after Parker joined, it was already possible to observe a Facebookian influence on Spotify’s strategies, well documented in patents registered by Facebook engineers at that time.⁷⁰

The marketing jargon at the time focused on social, sharing, and platform. In the language of Period C, Spotify would no longer be just an app that gives you access to music but a platform where you manage your music. The user interface also developed, probably more than in the two preceding periods. First, it became social, in Facebook’s particular sense of the word. Users could now build a personal profile within the app, add friends, and then drag-and-drop to share music. The profiles were connected to Facebook accounts, to help users find more friends, as well as to facilitate company extraction of data among users.⁷¹ If the push toward subscriptions largely relied on deals with other subscription-based businesses (telecoms), the efforts to become a platform was likewise premised on integration with an existing platform (Facebook).

Chapter 2 will chart these different data integration strategies in more detail. Suffice it to say that, at this time, Spotify also announced its first partnership with an audio hardware manufacturer.⁷² Another new feature was that the client could now index and play music files stored on local computers. Thus, users could listen to music that was not present in Spotify’s catalog if they had downloaded the files from other sources (legally or illegally). Indexing of the local hard drive was not just an option, it happened automatically. After installing Spotify on a computer, a program would go through the entire hard drive looking for stored audio files, which were then reported back to Spotify’s servers. This did not spark a major debate over privacy, and we simply cannot know what happened to all the data that Spotify collected regarding personal MP3 archives. There remains the possibility that it was shared with record companies as part of a licensing deal.

Less effort during this period was put toward developing ways to discover new music. A “related artists” function was added to the client but still no personalized recommendations.⁷³ Industry observers did not consider this a flaw but praised Spotify for being fast, clean, and simple.⁷⁴ “Competitors tend to boast about the size of their catalog and their tools that help users make sense of their massive amount of music,” Billboard stated in late 2010. “In contrast, Spotify assumes what people want most is a fast and easy-to-use product.… Without a doubt, Spotify is the best subscription service on the market today.”⁷⁵ Comments like this demonstrate how, at this time, personalized recommendations were still not regarded as an essential feature of music streaming services.

In May 2011, new restrictions were imposed on Spotify’s free service. Users who did not pay for a subscription could now only listen for ten hours per month and only five times for each track. This resulted in some negative word of mouth and led commentators to finally declare the idea of ad-supported music to be dead. This was not an initiative by Spotify Sweden, however, but one dictated by the major record companies, who would not let Spotify launch in the United States before seeing hard evidence of a rise in paying users. Yet Spotify soon made a move in the opposite direction, opening up its mobile version to users of Spotify Free.⁷⁶

The experiment seems to have satisfied two of the Big Three record companies, and only Warner Music blocked the road to America. As it happened, Warner Music was also up for sale and was acquired by the Russian businessman Len Blavatnik. Another bidder was Sean Parker from Spotify’s board of directors. According to Parker, he only left the bidding process after getting a guarantee from Blavatnik that, under his ownership, Warner Music would let Spotify launch in the United States without giving up its free version.⁷⁷ According to a leaked contract, Spotify also had to accept paying hundreds of millions of dollars in royalty advances to the Big Three and provide free advertising space to Sony Music before the roadblock was finally lifted.⁷⁸

`Period D (2011–2012)`

In the middle of the summer of 2011—on July 14, to be precise—Spotify launched its service in the United States. The launch had been made public the previous evening via press release, only a couple of hours after Spotify had signed the licensing deal with Warner Music. The launch, however, was not widely advertised. As usual, Spotify relied on word of mouth, and the key to get the hype going was the offer of free access to music streaming (i.e., the same offer that somewhat paradoxically had delayed the US launch for years). To try out Spotify, Americans had to either pay for a Spotify Premium subscription or get an invite for a Spotify Free account. Each user received a limited number of invites to pass on to their friends. This way, Spotify created a powerful marketing machine based on the free efforts of its own users. The limitation on invites not only ensured that network capacity would not be overloaded but also contributed to hype that associated Spotify with the lifestyle of those circles that were first to receive invites—primarily young, hip New Yorkers. This was largely a repetition of the marketing pattern that had proven successful during the Swedish beta period of 2007 and 2008 (except for the political element in the Swedish hype). Two months later, registration for Spotify Free opened to anyone with a Facebook account.⁷⁹

The reception of Spotify in the US press was enthusiastic. In describing the service, reporters emphasized free access as Spotify’s “unique selling point.” Many accounts also stressed that the service ran with extraordinary speed; a song started playing almost instantly after it was clicked, partly thanks to the P2P network used by Spotify to distribute data. Social features, such as playlist sharing, were also mentioned but not the radio-like features, and Spotify was hardly ever described as a service helping users to discover new music.⁸⁰

When Spotify launched in the United States, the company received $100 million in fresh venture capital, more money than in the three preceding funding rounds combined. Taken together, the three investor firms gained approximately 10 percent of Spotify’s shares, and the company was now valued at $1 billion. The investment deal had already been signed in early 2011, on the condition that it would not be triggered until Spotify reached a licensing deal with the Big Three to allow a US launch.⁸¹

With this influx of money, Spotify could run for a year and a half before the next funding round. Apparently, a main condition of the investment deal was that Spotify would expand its operations in Western Europe. Later in 2011, Spotify launched in Denmark, Belgium, Austria, and Switzerland, followed in early 2012 by a relaunch in Germany and then in Australia and New Zealand. This pattern of expansion confirmed the Western geography of Spotify; not until 2013 would the service be available in parts of Eastern Europe, Latin America, and Southeast Asia.

Apart from international expansion, this period in Spotify’s history was also characterized by a specific direction of service development. Data-driven personalization was still not widespread, and the developing arm at Spotify still treated neither discovery nor algorithms as defining concepts. Rather, their priority in 2011 was to transform Spotify from a music service into a platform that was integrated with other platforms, most notably Facebook. If the “platformization” of Spotify and the alliance between Spotify and Facebook had been established by Sean Parker in the preceding period, it was now confirmed by the fact that Series D investment was led by the Russian firm DST, which had made a large investment in Facebook two years earlier, in the middle of the economic recession.⁸² This move by DST in 2009 had marked the beginning of a new gold rush in tech startups, this time with social media as the main marketing term and driven by trillions of dollars released by the Federal Reserve to stimulate the economy.⁸³

Arguably, to become social was never more important for Spotify than in 2011 and 2012. But the attempt was not without complications. At the F8 conference in September 2011, Facebook CEO Mark Zuckerberg presented a number of new features, many of which related to the new concept of “frictionless sharing.” This was another step in Facebook’s ongoing redefinition of what it means to be social online. First, social had been associated with the active dissemination of words and images. Then the thumbs-up Like button had been introduced in 2009, reducing sociality to one click. Now Facebook was set to abolish even that and automatize the sharing of activities between friends without the need for any active choice. Instead of choosing to share a particular song, users who opted to add a music app to Facebook’s platform would now share all songs they listened to. The prime example presented by Zuckerberg was Spotify, and he even invited Daniel Ek to the stage for a brief guest appearance.⁸⁴

The integration of music services was clearly visible for users of Facebook. At the left-hand side of the interface now appeared a Music Dashboard, including statistics for which songs were trending among friends. On the right-hand side, the news ticker showed in real time what friends were listening to and what app they used for listening—giving Spotify’s trademark valuable exposure. And in the central home feed, Facebook would occasionally provide updates.⁸⁵ “Music is one of the most social things there is,” Ek explained on Spotify’s blog. It is “inherently social.”⁸⁶ In other words, music is not intimate and has nothing to do with privacy. Not every Spotify user agreed—which we will discuss in more detail in chapter 2. The integration had gone too far and was partially rolled back.

In late 2011, Spotify also launched its “app platform,” inviting external developers to create new features to be offered within the client. Some of the apps provided news or lyrics, others specialized in playlist recommendations or ticket sales. With this move, Spotify also created a symbiotic relation with a number of “media partners,” including Rolling Stone, Pitchfork, and the Guardian. They would feature links to Spotify from their music-related articles, and it would not be far-fetched to imagine that these media partnerships also affected the news reporting about Spotify.⁸⁷ Finally, in this period, Spotify also developed a new radio function to replace one based on music genres dating back to the beta period. But there was still no personalization of Spotify Radio; the listener first had to select an artist to be given the option of listening to a “Related Artists” station.

`The Reinvention of Radio`

Spotify’s launch in the United States meant that it began competing with a number of music streaming services that were not available in Europe. The year after its US launch also happened to be a year when these services began to compete in a new way. One reason for this was the increased hype around Pandora Internet Radio, as it began trading on the New York Stock Exchange in June 2011, exactly one month before Spotify launched its service in America. This raised the question of whether Pandora and Spotify were direct competitors or wholly different services.

While both were indeed music streaming services, Pandora differed from Spotify by not being an on-demand service.⁸⁸ The user could not select individual tracks or build playlists but was instead supposed to lean back and enjoy a personalized radio station. Pandora also was far ahead of Spotify when it came to data analysis and was widely recognized as the best music recommendation service. Until this point, Spotify had not prioritized the “lean back” experience. This may be partly related to the fact that Europeans do not spend as much time in the car as Americans do. The main explanation, however, relates to copyright policy. In the United States, a streaming service without on-demand functionality may be classified as radio and therefore does not need to strike licensing deals with each record company. It would simply pay a certain royalty to a collecting society, SoundExchange. The result was that Pandora had to pay much less than Spotify for every minute of music streaming and hence could stick to an advertising-based business model. It also meant that Pandora could not expand beyond those few nations with a similar copyright system.

Pandora asserted that Spotify was not a competitor but “largely complementary”—and circulated a statement attributed to Daniel Ek that supported this view: “Daniel Ek says he thinks Spotify is the future of the record store, and that Pandora is the future of radio.”⁸⁹ Interestingly, this was an attempt to enlist media history in a commercial effort to fence off a future market for media services. Yet this historical divide was already in the process of collapsing, resulting in intensified competition. It also mirrored a shift in the relationship between the tech giants that took place around 2011. If corporations like Google, Apple, Facebook, and Amazon had previously focused on wholly different products—sometimes even partnering with one another—they now expanded their field of operations so that they were pitted against one another as direct competitors.⁹⁰

In the field of music streaming, services such as Slacker and Rara were already trying to combine the two models—radio and on-demand—within one interface. During the spring of 2012, a new service named Songza created a lot of hype for “falling right into that sweet spot between Spotify and Pandora.” Songza had no viable business model, and a few years later, the service would be discontinued after Google acquired the company. But its approach to music recommendation would have an enormous influence on Spotify’s subsequent trajectory. Songza did not ask the user to search for a favorite song or artist, or even to choose between genres. Instead, the guiding concepts were activity and mood (a tactic we will discuss in detail in chapter 3). Depending on the personal profile, the time of day, and the day of the week, Songza would give the user a choice between six different activities, each represented by an icon. For example, on a Wednesday morning, Songza could offer a choice between work, exercise, or staying in bed and then provide a soundtrack for this activity. Possible choices for a late Friday night could include “Making Out,” “Getting High,” or “Bedtime.” Activities could also be filtered by the selection of a predefined mood, such as “Aggressive” or “Introspective.” The resulting playlists would be put together by Songza’s own music experts. This focus on human curation was, at the time, seen as a unique selling point for Songza in its competition with Pandora and Spotify, which had both put efforts into developing algorithmic systems for music recommendation.⁹¹

During 2012, a new consensus emerged among music industry commentators: streaming services would henceforth compete with the best music recommendation features. Not long before, Spotify had been praised for being clean and simple. Now it was criticized by some for being “just a huge database of songs” and not assisting its users in choosing the right music.⁹² Billboard wrote about “the resurgence of radio” and pointed to the multitude of startups involved in reinventing this old format by trying to make it either personalized, mood-based, or local—and sometimes even by mixing music with news and weather reports. According to Billboard, this was all because of “the fact that people love to simply lean back and listen.” Hence, Spotify’s “lean forward” approach suddenly began to appear as a weakness rather than a strength.⁹³

`Period E (2013)`

At the end of 2012, Daniel Ek made several statements indicating that Spotify would now adjust its ambitions for what a music service should deliver to its users. In 2012, the focus had been on growth in numbers, but 2013 would be the year for Spotify to address a very different issue: “The abundance of choice. How do you make sense out of 20 million songs?”⁹⁴ Until this point, Spotify had followed the on-demand doctrine, which tended to ignore the question of meaning making. The user was treated as a sovereign individual who already knew precisely what music he or she preferred to hear. Spotify’s task was simply to respond to a request from the user, providing “whatever you want, whenever you want it.” The individualism of this approach had been somewhat moderated by the “social turn” of Spotify, yet in effect, social was a term defined by a few dominant services and platforms, in particular Facebook.

For most of 2012, when Ek was asked about how Spotify would address the abundance of choice, he would hint at a social solution. According to the doctrine of social music discovery, friends would successfully help one another navigate the abundance of music, provided that the service allowed for frictionless sharing. But what if your friends had bad taste? Toward the end of the year, Ek shifted tactics and recognized the need for authority: “Here’s what our users are telling us: Spotify is great when you know what music you want to listen to, but not so great when you don’t.… The biggest unsolved question for most users is, how can you help me figure out what I’m going to listen to? And for artists, it’s, how am I going to be heard?”⁹⁵

Spotify’s first solution was presented in December 2012, when it introduced a new, personalized recommendation function that would provide each user with “the most relevant content,” as well as the ability to “follow” musical suggestions by “artists, trendsetters, editors and experts.”⁹⁶ This was a step away from the symmetrical sociality of Facebook (where friendship is a two-way relation) toward the asymmetrical following system that characterizes Twitter (where a small number of users tend to become hugely influential).⁹⁷ “Now you can get music recommendations from only your most trusted musical influences,” Spotify wrote in a press release.⁹⁸ In other words: friends cannot always be trusted to recommend the best music for you. Curating millions of songs was a task too large to be left to users.

The introduction of new discovery functions, however, was far from an innovative move. On the contrary, Spotify was rather late to follow this trend in the music streaming market. But when it finally did recognize that there were limits to the on-demand doctrine, Spotify changed its business strategy. The company gradually reoriented itself toward the aim of providing not only access to music but also recommendations for music that users would not have requested themselves. This meant that Spotify began to transform itself from being a simple distributor of music to the producer of a unique service. At the time, the nature of this service was still not clearly defined, but it began to take shape through a series of acquisitions that were made possible by the hundreds of millions of dollars that investors were willing to throw at Spotify. This time, the funding round was led by Goldman Sachs and also included the Coca-Cola Company, which had entered a marketing partnership with Spotify.

The first company acquired by Spotify, in May 2013, was Tunigo. For two years, this service had provided Spotify users with playlists based not only on genres but also on specific activities or moods, curated by a small team of music experts—clearly inspired by Songza. Tunigo had already been the object of much hype on Spotify’s official blog and was one of the original apps when Spotify launched its app platform. After the acquisition, Tunigo’s features were integrated into the Spotify client. In a larger perspective, this marked an incipient retreat from the platform strategy; significantly, the app platform that Spotify had launched in late 2011 would be dismantled in 2014. Tunigo had about twenty employees who would form the core of a new unit within Spotify. They would be joined by a growing number of “music editors” recruited to create local playlists for each country where Spotify was available.⁹⁹ This curatorial turn took some time to implement but can clearly be traced through Spotify’s job listings, which were preserved by the Internet Archive. A snapshot from January 2013, for example, lists seventy-seven jobs, of which the vast majority were in tech development. None of these jobs involved music curation.¹⁰⁰ In fact, it was not until 2014 that Spotify began recruiting people in different countries for positions such as “Music Editor/Playlist Curator”:

At Spotify, our vision is to provide the perfect music for every moment.… We’re growing our world-class music curation and editorial team and are looking for a multi-talented self starter, go-getter with limitless ambition and an undeniable hunger for progress and passion for music. This role calls for someone who can identify and execute the best music playlist listening experiences for a multitude of moods, moments, and genres, has a passion for performance-oriented analytics, and has her/his ears to the ground in the music community.¹⁰¹

Part of this job would also be to try out “playlist hypotheses, i.e. the perfect running playlist or music for a date night or a dinner with friends.”¹⁰² Running, dating, dining—these were indeed typical “moments” for Spotify, and by deploying such terms, it followed Instagram, Twitter, Tinder, and other popular apps.¹⁰³ Spotify even appropriated Tunigo’s slogan: “Music for every moment.” Indeed, it came to suggest something quite different from its previous slogan: “Music whenever you want it, wherever you are.”

The early Spotify saw its task as providing access to music, filling the life of consumers with more music, as if they felt a need to remedy a general lack of music. But around 2013, Spotify’s curatorial turn reconceived its product based on a different image of the consumer. Now the selling point was to provide not more music but better music. This meant, however, that Spotify needed to somehow define musical quality, something that it had so far avoided doing. The service had to steer a way between pure relativism (i.e., each individual has her own standard for what is good music) and pure absolutism (i.e., certain pieces of music are inherently better than others). This polarity was transcended by the idea that musical quality is dependent on context. This is a utilitarian approach in which music is understood as functional for certain activities, as we will demonstrate in chapter 3.

`Period F (2013–2015)`

The amount of venture capital invested in Spotify in each funding round continued to increase, not linearly but exponentially. In November 2013, a venture capital firm that had been active in Silicon Valley since 1995, Technology Crossover Ventures (TCV), made its biggest investment ever: $250 million. Proportional to the equity, the deal was read as a valuation of Spotify at over $4 billion.¹⁰⁴ The high numbers could be interpreted as a bet on world domination for a company that, paradoxically, was not even close to being profitable. This tremendous elevation also made it improbable that Spotify could be acquired by a firm other than Facebook, Apple, or Google. Conversely, from then on Spotify would often be rumored as a possible acquirer of other music services, such as Pandora or SoundCloud.

As the sole investor in this massive funding round, TCV likely had some influence over Spotify’s strategic priorities for 2014, which followed a distinct pattern. Spotify did expand to a few large countries: the Philippines, Brazil, and Canada. It also started a subsidiary in Russia and probably invested a lot of money in launching its service there—only to withdraw again because of Russia’s economic crisis and its new, repressive internet legislation.¹⁰⁵ Above all, however, Period F was characterized by consolidation and centralization of the service. A truly significant change, enacted in Spring 2014, was Spotify’s dismantling of its P2P network, which until then had guaranteed its supply of bandwidth.¹⁰⁶ From the very beginning, Spotify had provided users with a remarkably fast service and avoided paying market price for the bandwidth by harnessing the resources of those same users. It was a technology that Spotify had patented and that made the service different from other music streaming platforms. Now, with hundreds of millions of dollars, Spotify could finally maintain its own servers for the distribution of music.

The shutdown of one of the internet’s largest P2P networks, which was then replaced by central servers, must be considered a major infrastructural event. Yet it was hardly recognized by ordinary Spotify users and was not even communicated on the company’s blog probably because the move toward centralization did not fit the company’s preferred image.¹⁰⁷ Not only was the P2P network silently dismantled but so was the app platform. This was a platform Spotify had once described as “the next big step in musical enjoyment,”¹⁰⁸ yet it took less than two years for Spotify to stop accepting new apps.¹⁰⁹ Some of the most popular app features were copied or acquired, as in the case of Tunigo.

During the spring of 2014, Spotify made its second acquisition: the music analysis firm Echo Nest. The Echo Nest was known for powering the algorithmic recommendations in Spotify, as well as in several of Spotify’s competitors. The acquisition may be read as yet another move toward centralizing the infrastructure of streamed music. While it was said that the Echo Nest would continue to serve all of its existing customers, Spotify’s acquisition was obviously considered a move to fend off competition. In fact, after just one week, Spotify’s competitor Rdio abandoned the Echo Nest and had to find another supplier of music intelligence.¹¹⁰

Spotify seemingly preferred to challenge other streaming services in the arena of personalized recommendations. But later in 2014, another kind of competition arose. Taylor Swift, the superstar singer-songwriter, made a public statement and withdrew all her music from Spotify. In a later op-ed for the Wall Street Journal, she complained that Spotify’s free service was undermining the value of music and expressed the opinion that an album should have a price.¹¹¹ Her move was explicitly anti Spotify, since rival services such as Rdio were still allowed to stream her albums (other than the most recent one). “Swiftgate,” as this incident soon became known, marked the beginning of a new kind of competition over the exclusivity of content that would intensify in the coming years.

`Period G (2015–2016)`

After much anticipation, Apple finally launched its own music streaming service, Apple Music, in June 2015. Late to join the streaming market, Apple’s obvious advantage was the control of a line of hardware, as well as its App Store. Furthermore, Apple could afford to accept even larger losses than Spotify—and for a longer time. Spotify’s response was to try setting different terms for competition. The tone was set a month before Apple Music’s launch, at an event in New York City in May 2015—an event that we will describe in detail in chapter 4. If there was one common factor in the new features presented there, it was Spotify’s attempt to become much more than a mere distributor of music.

A typical example was the premiere of Spotify Running, a new feature that used the smartphone’s sensors to detect the pace of a runner and play music at the same speed. Spotify also started cooperating with electronic music producers to create special compositions that could change or be rearranged according to the pace of running. This amounted to the establishment of a new musical format, beyond recorded music in the usual sense. Without a doubt, this feature was an attempt by Spotify to move beyond the realm of distribution toward becoming a producer of unique musical experiences. “Music is moving away from genres,” Ek said at the May 2015 event. Spotify sought to compete with Apple Music by providing more sophisticated curation of playlists, tailored for “the moment.” Beyond mere personalization, the algorithms would also take into account spatial data (where you are), temporal data (the day of the week, the season of the year), and maybe even the weather. If the assumption had previously been that musical taste is a property of the individual, now Spotify seemed to implicitly accept the view that it is rather “an aggregate of the supra- and the subpersonal.”¹¹² The new recipe for delivering the best recommendations followed from two key acquisitions, first Tunigo (expert curation) and then the Echo Nest (algorithmic curation).

At the event in May, Spotify also declared its intention to go beyond music and offer other types of media, such as videos and podcasts. Videos were to be presented by a number of selected media partners, which included a cadre of major American television networks. To deliver all those features, and to compete with Apple in general, would not be cheap, so early in the summer of 2015, Spotify took in half a billion dollars in a Series G funding round led by Swedish telecommunication giant TeliaSonera. This was more money than in all previous rounds combined, and the investment resulted in Spotify’s valuation more than doubling, from $4 billion to $8.5 billion.¹¹³ As soon as the money was secured, Spotify acquired yet another music intelligence company, Seed Scientific—with Apple Music as one of its customers. It was clear that Spotify hoped to succeed by building a service on the most advanced algorithmic recommendation systems and to fend off its rival from music intelligence suppliers.

Apple responded by presenting its music service as a warmer and more “human” alternative, compared to the allegedly all-too-algorithmic Spotify. While Apple Music emphasized its radio station Beats 1, where human DJs play mostly mainstream pop, this positioning of man against machine should be taken with a grain of salt. Within industrial capitalism, all commodity production infers the combination of labor power and machinery; this holds equally for the production of curation as a commodity.¹¹⁴ The question is mostly one of the proportion between human and algorithmic curation. Spotify has continued to hire music editors with expert knowledge in local culture, even as it invests in technology for “music intelligence.”

In the summer of 2015, these efforts resulted in the introduction of “Discover Weekly,” a playlist of personalized music recommendations delivered every Monday to each user of Spotify. This was a highly successful addition to the service and an important lever for success in the music industry. A year later, it was reported that a significant proportion of Spotify’s users—no longer “trapped behind a search box”—were regularly listening to “Discover Weekly” and also adding its tracks to personal playlists. “Discover Weekly” would be followed by more weekly playlist offerings: “Fresh Finds” and “Release Radar.”¹¹⁵

For users of Spotify, the discovery of new music thus became scheduled for specific weekdays, which is reminiscent of traditional radio. It was first remediated by Songza, then copied by Tunigo, and subsequently integrated into the heart of the “new” Spotify. However, not every bid was successful. Spotify’s attempt to integrate video was a failure, especially considering the great amount of money it had invested in efforts to enlist major media partners, who only provided content for a short while. Moreover, competing with Apple obviously requires enormous resources. In the beginning of 2016, Spotify therefore began to accept funding in the form of convertible debt. This meant that instead of receiving shares in Spotify, investors were given the promise that they could buy shares at a certain discount in the future, at a time when Spotify was expected to finally make its stock market launch. If the launch was to be delayed—in this case, beyond 2017—the interest rate would rise incrementally.

`Toward an IPO`

As a particular form of credit, convertible debt—which many consider quite risky—has created a very strong incentive for Spotify to transform itself into a publicly traded company. And no small amount of debt was taken in 2016, as $1.5 billion is roughly equal to the total amount invested in the company before that year. As of early 2018, Spotify had spent much of its money on a virtual shopping spree, acquiring ever-more technology companies with very different focuses. All these acquisitions give indications of Spotify’s future trajectory, but it is still too early to integrate these in the broader company history of Spotify. For now, we can say only that Spotify was finally introduced at the New York Stock Exchange on April 3, 2018, as this book was going through its last round of edits.

This chapter has charted the business and media history of Spotify in some detail. While it is probably the first academic account of this sort, it is not the only possible one. Other scholars might choose to write Spotify’s history from the perspective of the music industry or with a focus on individual executives within the company, for example. Here, we have instead structured the narrative in a way that highlights Spotify’s dependence on venture capital and rights holders, which makes it surprisingly precarious. Rather than being an autonomous innovator with a divine power to shape the future of music, we have shown how Spotify has often been rather late to follow trends set by others. Several times, the service has proudly introduced new features, only to dismantle them shortly thereafter. The core business model has also changed several times: first it was not specifically about music, then it was about free and ad-supported music, and then the free service was reconceived as a way to market the subscription service. For many years, Spotify’s interface was based on the idea of a user always possessing perfect knowledge of her own musical preferences, but then the curatorial turn occurred and personalized music recommendations gradually became the central feature.

Rather than positing the analog against the digital, or seeking to characterize some innate characteristics of streaming, we have highlighted a turning point in the development of Spotify and other streaming services, where music discovery becomes the main theme of competition. As noted early in this chapter, at the time of its launch in 2008, Spotify inscribed itself in a particular media history, making itself appear as the successor to the record store. A few years later, however, music streaming services began to rediscover the heritage of radio, and Spotify followed this trend. Yet we find that this shift is not reflected in much scholarship, where “music streaming” appears as a stable technological condition. The business history of Spotify, to which this chapter contributes, shows how the service first conceived of itself as a technology company working with digital distribution while not being particularly about music. More recently, Spotify has attempted to move beyond a distribution-oriented company profile by transforming into a producer of unique music-related experiences. This shift has coincided with a curatorial turn, in which the best music is the music that is best suited for a given context. This curatorial turn includes a localization of music recommendations, which are produced by music curators employed at Spotify’s national offices. Finally, we must note that Spotify, after years of international expansion, is still a rather Western service—significantly more so than competitors such as Deezer or Apple Music. Spotify is still not available anywhere in the Middle East or Africa (except in Israel and South Africa, where it launched in 2018), nor in Russia, India, Korea, or mainland China.

`Notes`

20. For example, we wanted to find reports from those Swedish bloggers who were the very first to try out the beta version of Spotify in 2007. Most of those blogs are no longer online. In some cases, the content is available via the Internet Archive (https://archive.org/) but is not directly searchable, so it is necessary to first identify links to the specific blog entries. There was once a popular site named Delicious (https://del.icio.us) that was used for social “bookmarking.” There we found a public account used by the Swedish tech entrepreneur Henrik Torstensson, who was one of Spotify’s very first beta testers. As early as 2007, he collected some links tagged with “Spotify,” producing a link list (https://del.icio.us/torstensson/spotify) that we found very useful for our research. Many of the links pointed to dead websites, but the content could sometimes be found through the Internet Archive, giving examples of the early reception of Spotify. The short life of many websites is a good reason why the history of digital culture must be written sooner rather than later. In addition, Spotify’s official blog should not to be overlooked as a source of information on the many changes to the service, as well as how these were communicated to users. The Twitter accounts of the company and its executives are also useful. Daniel Ek, for instance, has been tweeting as @eldsjal since late 2007. We found no indication that Spotify has engaged in retroactive deletion of tweets or blog posts but cannot be sure; this is an issue of source criticism that any kind of business historian will confront.

43. We know this from personal experience, as one of the authors is also a musician who made recordings that were available in the Spotify beta—only to disappear at the public launch. This has also been reaffirmed by other persons who made similar observations at the time: “Most importantly, I can confirm the pirating part as I was using the Beta, and thought at the time it was a cool thing to do.” Magnus Boman, reply to “<nettime> Spotify Threatened Researchers Who Revealed ‘Pirate’ History,” Nettime (mailing list), October 9, 2017, https://nettime.org/Lists-Archives/nettime-l-1710/msg00049.html. Niklas Ivarsson who was then Head of Licensing at Spotify, also stated in an e-mail to one of us authors in November 2008: “All content for the Beta tests were taken from our own or friends music collections.” See also Andersson Schwarz, Online File Sharing, 32.

45. Eric Wahlforss, “Please Let Me *Not* Own My Music: On Spotify, SoundCloud,” personal blog, March 3, 2008, http://eric.wahlforss.com/2008/03/03/please-let-me-not-own-my-music-on-spotify-soundcloud/ (site discontinued); Henric Karlsson, “The swedish model vill erbjuda musik gratis,” [The Swedish Model wants to offer music for free], Göteborgs-Posten, March 5, 2008, http://www.gp.se/kultur/the-swedish-model-vill-erbjuda-musik-gratis-1.1153611; Adam Svanell, “Reklam ska göra musiken gratis,” [Advertising will make the music free], Svenska Dagbladet, April 6, 2008, https://www.svd.se/reklam-ska-gora-musiken-gratis; “Årets entreprenörer 2008,” [Entrepreneur of the year 2008], Internetworld, April 25, 2008, https://internetworld.idg.se/2.1006/1.157629/arets-entreprenorer-2008; and Andres Lokko, “Internet är allas vår skivsamling,” [The internet is our common record collection], Svenska Dagbladet, August 15, 2008, https://www.svd.se/internet-ar-allas-var-skivsamling.

107. The technical reasoning behind Spotify’s peer-to-peer shutdown was soon explained by Gunnar Kreitz, one of Spotify’s most senior network engineers and one of five names on its first patent. In addition, he is a postdoctoral researcher in theoretical computer science at Stockholm’s KTH Royal Institute of Technology, where he held a presentation on Spotify’s peer-to-peer network. He still saw many benefits of using this network, but indicated that it would be better for video distribution, which demands less bandwidth and tends to be fragmented in smaller pieces. And indeed, when Spotify began developing this technology back in 2006, it was with video distribution in mind. See Gunnar Kreitz, “Spotify—Behind the Scenes: A Eulogy to P2P (?)” (PowerPoint presentation, KTH Royal Institute of Technology, Stockholm, Sweden, May 7, 2014), https://www.kth.se/social/upload/536a05d8f2765472d425ac0a/kreitzspotify_kth_kista14.pdf.