`4 What Is the Value of Free?`

Sitting amid a large crowd of agitated press representatives and industry executives, it is hard to overlook the work that has gone into staging this event. It begins, of course, with the loud playback of a funky Spotify playlist, attracting much excitement on Twitter and Periscope, where the event is being transmitted live. Neon red lights are shooting up the walls. Then the music is turned down and the lights are dimmed. An ad is screened, showing hip millennials enjoying music while pursuing activities such as running or doing laundry, ending with the tagline “For you. For now.” A sweaty, pale Ek takes the stage and begins unpacking statistics, emphasizing Spotify’s global claim on scale. Slides with numbers, arrows, and country icons pass in the background behind him, vividly changing colors. A product called Now, which offers personalized playlists, is launched by Spotify’s vice president of user experience. Video is introduced as being part of Now. Comedians Abbi Jacobson and Ilana Glazer, creators of Comedy Central’s television series Broad City, join the VP on stage. Hilarity ensues. The comedians leave and Ek rushes back on stage, still pale and sweating, announcing video content partnerships. Another executive comes up, announcing music for running. Musician Tiësto appears and acts as if he is having an informal chat with the VP about composing music for running. The event culminates with a wall suddenly being removed, revealing D’Angelo on a previously hidden stage, with Questlove on drums. Fog machine, two songs played live, cheers from the audience. End of show.

During fieldwork in New York City in May 2015, we got access to a media event organized by Spotify to present plans for expanding its services in response to increasing competition from YouTube, Tidal, and (most importantly) Apple Music. Addressing the press, industry executives, and investors, CEO Daniel Ek introduced major additions to Spotify’s core business. These included the integration of short-form videos, podcasts, and radio, representing a shift from manually selected genres to automatically personalized playlists, as well as “more targeted advertising.”¹ Based on a trove of behavioral user data collected over the years—and on content deals with the BBC, Comedy Central, and multichannel networks such as Fullscreen, Maker Studios, and Rightster—Ek offered a top-down curated media buffet as a precondition for programmatic advertising, a mechanism using personal data and algorithms to buy and sell ads. Promising to “soundtrack your entire day, and then your entire life” and to give “new meaning to the phrase, ‘stay tuned,’” Ek consistently evoked scale. Referring to “25 billion hours of listening” since launch, growing numbers of subscribers, and devices such as social tools and discovery features for further scaling the service, his presentation culminated in the claim that “Spotify is the growth in streaming, and that means so much to us!”²

Reading these field notes three years later, we reflect that the 2015 product launch left at least two questions open. First, what was the actual product being launched? While commentators largely took it to be about the introduction of video series programming and related content deals with Vice, Comedy Central, and Maker Studios, this content never made a lasting impression, with view counts as low as “hundreds” per video.³ However, the curated playlists that match what listeners are doing at certain parts of the day (described in chapter 3) remain a prominent feature of Spotify. Second, to whom was this product launch addressed?

In order to answer these and related questions, this chapter begins by briefly contextualizing Spotify within a broader history of “free” access to internet services. The chapter then adopts a conceptual framework provided by economic sociology, a field of empirical research that challenges the tradition of neoclassical and mainstream economics.⁴ Based on such a definition, the chapter focuses on the one element of Spotify’s music market that has been at the very center of internet culture’s free debate for almost two decades: advertising. Providing a detailed empirical and experimental study of Spotify’s programmatic advertising—and ad-tech networks—we will show that targeting is part of a process that considerably exceeds the mere dissemination of digitized sounds. We also argue that Spotify’s platform or interface is not where its main market materializes. While there is, of course, relevant economic input and output managed via Spotify’s back end and interface, this market is in itself embedded in a more significant one that is not directly observable on or via Spotify. The chapter concludes by suggesting that the study of digital “free culture” needs to include a focus on the embedding of markets into other markets and consequently on the negative externalities usually not considered in platform studies. The real “value of free” is that it helped scale Spotify’s marketplaces, the most important of which (finance) extends far beyond music.

`A Brief History of` `Free`

Free is the most common way of imagining the delivery of information via the internet today. In this view, digital culture largely owes its cultural specificity to the absence of a commercial revenue model for—and private ownership of—global communication infrastructure. Theorists such as Larry Downes, Philip Howard, Jeff Jarvis, Charles Leadbeater, Jeremy Rifkin, Clay Shirky, Don Tapscott, and Anthony Williams have vividly described the internet as ending the rule of monopolies and inspiring decentralized media flows.⁵ A decade ago, using examples such as Wikipedia or the free and open-source software movement, Yochai Benkler famously argued for the internet as a free information ecology, where the “outputs” are “usually nonproprietary” and where “free access to a set of the basic instrumentalities of economic opportunity” is provided. “From the perspective of a society’s welfare,” Benkler stated, “the most efficient thing would be for those who possess information to give it away for free—or rather, for the cost of communicating it and no more.”⁶ Subsequently, the ideas of the internet as a common carrier of freedom of speech and of gratis access to all sorts of cultural production have often come to be conflated. In daily parlance, free now equals a price tag of zero for online services or products that are—or appear to be—free to use and free of charge. This price tag is variously attached to email, blogs, software, news, games, social networking, online storage, scholarship, books, film, television, and music.

Historically, in the case of music, the advent of free can been traced to unauthorized services such as Napster and peer-to-peer file sharing, as well as to a subsequent cultural shift from ownership of recorded music to authorized access of vast music catalogs. Since 2012, this conceptual shift from ownership to access has been accompanied by Spotify’s attempt to engineer a shift “from access to context,”⁷ in the sense of using music as such a context for developing a business model based on advertising revenue and subscriptions. The history of these conceptual shifts is significant and worth closer scrutiny.

The idea of free access to information is often traced back to the famous 1980s hacker slogan “Information wants to be free.”⁸ By the mid-1990s, however, it had already turned into a cyber-libertarian market proposition that not only envisioned a “toll-free” cyberspace for “netizens”—to recall the then-prominent jargon—but also an explicit trade-off between users and advertisers.⁹ In opting for free access, users were increasingly asked to agree to a model that would exchange their personal data for the “enhanced targeting of online ads.”¹⁰ While targeting capabilities have massively increased since the advent of the social web in 2000, display ad targeting already existed in the mid-1990s.¹¹ Such personalized ad targeting was seen as a major promise of the internet, marking a break with earlier mass advertising principles. Also, initial attempts in 1996 by the Wall Street Journal and Slate to introduce subscriptions after a period of free access had failed dramatically.¹² The consensus was, in short, that if the internet wanted to grow, it had to be “ad-supported.”

The shift from ownership to access, in other words, went along with an economies-of-scale approach to cultural content distribution, based on projected new revenue streams from ad targeting. The term economies of scale refers to an economic system based on diminishing marginal costs, that is, an expected decrease in the cost of production for any additional copy or unit of an item produced. Newspapers, films, and songs were envisioned as markets with unlimited scalability, since copies could be made available at no cost per copy—after the first one had been produced—premised on global access as a precondition for such unlimited growth.¹³

Turned into a market model for the emerging digital economy, free thus came to signify a curious shift from what it had meant previously, that is, public economies and nonprofit organizations.¹⁴ Free also came to mean both gratis and voluntary, in the sense of being free to the consumer and voluntarily given to the advertiser. For instance, in 1996, the webmail service Hotmail was widely touted as an innovation for exchanging a free web-based email account for the free or “viral promotion” of its very own service by automatically adding a tagline to each email. Subsequently, companies such as Google, YouTube, Facebook, and Spotify came to be known as building on a similar “two-sided market” model, whereby a platform establishes an exchange between two different types of users that value each other’s presence, without this relation being one of buyers and sellers.¹⁵

In theory, it is the platform that incentivizes valuable behavior between formerly disconnected parties—like “a space where people meet and not just where trade takes place.”¹⁶ In practice, this has often amounted to a model similar to ad-financed newspapers or television, wherein free programming allows the sale of audiences to advertisers. For YouTube, Facebook, Spotify, and many other services, this is what the shift from access to context has implied. Hence, understood as a market model, free was premised on the assumption that the exchange value of cultural production itself could be neglected. Videos, books, music, and personal data could be given away without charge, in order to enable platform owners to capitalize on the markets they had created. Although copyright infringement claims were brought up in the early histories of YouTube and Spotify,¹⁷ these and other services quickly succeeded in establishing themselves as mediators between free users and advertisers. As we discussed in chapter 1, the hype around Spotify intensified during 2008, when it was rumored that the new service would “make music free” by relying on advertising. Thus, Spotify is often singled out as a company that positively affected supply and demand by bringing more music to more listeners, decreasing marginal and fixed costs, and exploiting the so-called indirect positive externalities between different sides of the market.¹⁸

In this view, Spotify has enabled advertisers to benefit from the presence of many consumers, while consumers now benefit in their discovery of products and services from many advertisers. But what are the benefits for musicians—or cultural producers more generally—whose content is the incentive attracting both sides? Chris Anderson, probably the most ardent proponent of the model that identified two-sided markets with ad-supported media, claimed that technology had made gratis access to culture inevitable. Announcing a major revival and “shift to ad-supported content” in his book, Free: The Future of a Radical Price (2009), Anderson found that free was now simply a given, not least of all for music: “Music is nothing to pay for … it’s something they do for other reasons, from fun to creative expression. Which, of course, has always been true for most musicians anyway.”¹⁹

Over the past decade, such views and the economic models on which they rely have been widely criticized. When it comes to Spotify, a substantial part of this criticism has been directed against low royalty rates caused by the company’s so-called pro rata revenue share approach. Pro rata means that revenues received by the streaming service are divided to the rights holders based on how many approved plays a certain track has in relation to all the other tracks played at the same time. Popularity, catalog size, and catalog ownership matter, and the approach consequently benefits major labels—not independent musicians.²⁰ Listeners have also never valued the presence of advertisers in the way suggested by economics, as documented by continuous complaints within the Spotify community regarding the frequency, repetitiveness, and loudness of audio ads—especially in regard to the lack of proper ad targeting.²¹

Furthermore, advertising never fully supported Spotify’s free tier, since it failed to generate sufficient revenue to pay for gratis access.²² In 2015, Spotify’s declared “best year ever,” ad revenue accounted for only 10.1 percent of total income, with the remaining 89.9 percent coming from people paying for premium accounts.²³ Spotify has long experimented with various “freemium” models that combine free with subscription access. In 2008, for instance, the company offered four distinct tiers of service, two of which involved free access.²⁴ Yet the notion of freemium also largely fell short of expectations, with the result that Spotify still lost money for more than a decade.²⁵

The Spotify context makes clear the limitations of the two-sided market model itself, which envisioned and then structured market activities in the first place.²⁶ Anderson and his many followers in business and academia strategically downplayed the possible negative externalities that have, in a somewhat ironic twist, now turned into the most widely observed consequences of this development, such as private data brokerage, platform capitalism, and the unregulated global proliferation of intermediaries.²⁷ If it is fair to say that the two-sided market model has failed to account for the dynamics that have come to drive the digital economy, what are its flaws? Within economics, the answer has been to adjust and to expand the two-sided market model by suggesting the need for conceptual extensions for describing various platforms or by extending from a two- or three-sided market approach to a multisided one. While such adjusted models keep enabling entrepreneurs to operationalize network economics and other theories, they fail to account for anything that is not contained in the model. It is the abstraction of the market itself that confines an understanding of digital economies to forms of platform-mediated exchange.

What is needed, in other words, is a proper empirical description of digital markets. How is music put on the market today? A growing body of work in communication and media studies attempts to answer this question about music’s commodification. On the one hand, this includes industry-oriented scholarship based on interviews with music industry professionals that focus on how the music commodity can be successfully monetized.²⁸ On the other hand, media historians such as Jonathan Sterne and Jeremy Wade Morris have provided more critical views on the digitization of music, focusing on particular media formats, interfaces, affordances, and social practices.²⁹ Morris has provided a particularly detailed account of what he describes as the “digital music commodity,” arguing that once digitized, music always remains the same type of commodity. From this perspective, differences between streaming and downloading are neglected and the interfaces of Spotify or iTunes are merely understood as an arena where an already commodified object appears.³⁰ Although the commodification of music is not entirely determined by the techno-economic setting of a given platform, differences in the interaction design, business models, corporate strategies, or actors involved in Spotify, or any other media company’s service, matter considerably.

When Morris and other scholars take it for granted that it is the same commodity that is made available by “iTunes, Spotify, and many other digital retailers,”³¹ the underlying notion is that Spotify is indeed a retailer, or distributor of commodities produced elsewhere. The retail business is based on the buying and reselling of identical goods. Any company that sells commodities that are substantially different from those it has bought is evidently also engaged in some form of production. As this chapter demonstrates, Spotify indeed changes the substance and form of music as a commodity to such a degree that it cannot be seen as a mere distributor. As we have shown previously in this book, the turning point came with the introduction of personalized music recommendations in 2012 and 2013. At the same time, Spotify is also involved in market activities related to the dissemination of music. How is one to empirically grasp Spotify’s twisted market realities? Where is the market, and what is actually being sold?

`Entering the Market`

The best way to begin an empirical encounter with market actors is to define what is meant by the notion of the market. In drawing from various works within economic sociology, Patrik Aspers suggests this simple definition: “A market is a social structure for the exchange of rights in which offers are evaluated and priced, and compete with one another.”³² Key here is the notion of market as a form of order, for “it is only when there is order that we can speak of a market.”³³ As a form of ordering of economic knowledge and activities, markets imply a place (or forum) where goods of any kind can be exchanged in a voluntary, peaceful, and transparent way, based on legitimate property rights, free choice, and public prices. Markets also tend to go by some name, indicating that they are about one kind of thing and not another and that there is stability or extension over time—a market for used books today will not turn into a market for fish tomorrow. Such stability, or institutionalization of a given market, even pertains to recurring social roles where the market identity of each actor is either tied to only one side of the market or where the roles of buyers and sellers may be switched. In this way, most markets are “fixed role markets,” according to Aspers: a car manufacturer has an identity tied to the role of the seller of cars in the consumer market, while in a stock exchange (or “switch-role market”) buyers and sellers change places over the day.³⁴ In short, for a market to exist it must be clear what is traded in the market; there must be rules governing what (and what not) to do and by whom; and offers must obtain economic value in the market through procedures of competition and evaluation.

This conceptual framework thus posits that markets are primarily coordination devices. Markets correlate and arrange economic interest in many ways: by locating market activity at a certain site (virtual or real) and through pricing, the definition of roles, and mutual agreement on what is being traded. Coordination can also take on different forms. Apart from markets themselves, economic behavior can be coordinated through forms of reciprocity, or networks, and through redistribution, or hierarchy. While a market has buyers and sellers, a network implies open and reciprocal interaction without central control within a larger cultural setting, as opposed to a hierarchy (or decision-based asymmetrical order), such as a corporate organization. Market, network, and hierarchy are ideal-type abstractions to describe various degrees of formalization and control; they are regularly mixed in the real world. Markets thus may take on the character of networks or hierarchies, as implied in the notion of the two-sided or networked market.³⁵ Markets also never exist in isolation: one market is always embedded in other markets, as when a car manufacturer sells vehicles in a consumer market while buying labor in a different market.

While providing a useful approach for describing markets in general, the idea of markets as socioeconomic coordination devices is not easily translated into the realities of music streaming and the Spotify marketplace. Aspers’s simple questions—“What is the market about?” and “How are things done?”—do not lead to equally straightforward answers.³⁶ Given the absence of public pricing for any song, for instance, is Spotify’s subscription service indeed a retailer, as opposed to iTunes’ unit-based download model? Obviously not. In Spotify’s early years, even property rights were lacking, as the service’s cofounders did not hold the rights to the music they distributed.

So what is exchanged via Spotify? A growing number of subscribers buy access to a music catalog, but the majority of listeners still neither buy nor sell. This has prompted platform scholars to assume that, rather than being a market, Spotify is more like a network embedded in digital culture’s gift-giving or “sharing” habits, based on reciprocal interests between parties.³⁷ Here, platform theory seems implausible, given the explicit commercial interests on both the marketers’ side and the intermediary or platform’s side. One could easily argue that Spotify even contains elements of a hierarchical ordering, as it redistributes music in ways that do not adequately reflect the interests of all the actors in the network. And where is all this happening? If advertisers form a key “side” of Spotify Free’s networked market, the interface cannot be assumed to be the marketplace, as no evaluation or competition related to the “audience commodity” is observable.³⁸ So where are listeners—and listening—commodified? Who is trading?

Any answers to these questions will be only temporary, given that the entire setting of the exchange, or platform, is constantly changing, as we previously described in chapter 1. We therefore have to tackle these questions differently. In order to do so, we may come back to economic sociology’s observation that markets are always embedded in other markets. Harrison White (on whose work the observation is based) argued that many markets are “mobilizers of production in networks of continuing flows,” rather than isolated sites for transactions between parties.³⁹ Markets come in families, and what counts are the relations they entertain. In White’s view, markets can be seen as located in a production chain, so that we can talk of markets in “upstream” and “downstream” relations. In fact, what we call an industry is nothing but a set of markets, of which one is the core or leading market to which other markets are auxiliary.⁴⁰

`Scaling Markets`

It is helpful to think of Spotify as a market whose actors’ behavior depends on what actors in other markets do. Turning away from Spotify’s interface and the issue of advertising itself, it thus becomes possible to circumscribe the service marketplace from the outside—that is, by studying its environment, which also includes markets. And here the empirical work begins.

Returning to our field notes and the New York City event in May 2015, it is important to note that most commentators understood the launch of Spotify’s new features as an offer made on consumer markets. But Ek’s staged performance holds additional implications. In mobilizing graphs and statistics alongside music, talent, and the attending audience, Spotify engaged in a form of “data work,” betting on the affective properties of data visualization.⁴¹ Such data work comes with the routine setting of showcases and tech demos in which big data rhetoric is employed to conjure up scale. For consumers, there is no need to evoke global scales, nor is there such a need for distributors. There is no logical reason for any delivery infrastructure to make claims about scale; after all, roads are built without the promise of “25 billion” cars using them. Scale making, however, is regular practice among financial investors. As Anna Tsing notes in a study on internet investments, finance needs to conjure up scale: It is “a regular feature of the search for financial capital.… Profit must be imagined before it can be extracted; the possibility of economic performance must be conjured like a spirit to draw an audience of potential investors.”⁴² Tsing’s observation holds particular valence for internet economics, in which economies of scale are a key ideological framework.

Back in 2015, Spotify had to stage a “performative drama of financial conjuring” because the company attempted to grow into its $8.5 billion valuation before planning an eventual IPO, which the company finally filed for on April 3, 2018.⁴³ Every feature added to the service was thus also about commodifying the service itself, a service that despite its valuation had not generated any revenue. The market on which this commodity was going to be traded was not the market for music, entertainment video, or ad targets. In the decade-long buildup to its IPO and possible acquisition by Facebook, which was long rumored to be Ek’s end game,⁴⁴ Spotify—like Facebook—had literally become capital or “financialized stock.”⁴⁵

As we charted in chapter 1, between October 2008 and June 2015, the company raised $1.6 billion in seven rounds of investment from twenty-six investors, including Coca-Cola, Goldman Sachs, and Palo Alto–based Technology Crossover Ventures. In 2012, Spotify had begun trading securities on the US financial market; in 2015, it raised $500 million in the form of a loan convertible into Spotify shares; and in 2016, it raised another $1 billion in convertible debt with “devilish terms” that subsequently enforced a rapid joint stock exchange listing in Sweden and the United States.⁴⁶ Further strengthening ties to both finance and other media, Technology Crossover Ventures partner and former Netflix board member Barry McCarthy was appointed Spotify’s chief financial officer in 2015, replacing Ek in key leadership functions at subsidiaries such as Spotify Ltd., Spotify Service AB, Spotify Europe AB, and Spotify Sweden AB. Spotify, in short, had turned online music distribution into a media company and then into stock options, a market over which Spotify itself had no control.

While free was central for scaling Spotify in the early years, by 2015, finance had become the leading market from which Spotify’s other markets and their actors were governed. Using White’s analogy, one might say that the markets for ads and music were located in a downstream relation to financial capital, given the money-losing company’s dependence on such capital. If we describe Spotify as a production chain, then songs, videos, audiences, and ads appear as auxiliary markets, despite the company’s promotional claims to the contrary when facing consumers.

As Spotify’s case is not an isolated one, critics usually attribute such market structures to a “financialization” of media. Financialization, according to Max Haivens, is a broad cultural and social phenomenon, referring to a “contagious” expansion of financial “themes, ideas, tropes, measurements, metaphors and influence into spheres seemingly in no way related to the core operations of the financial economy.”⁴⁷ Financial interest seems evident in the way Spotify evokes scale through reference to a potentially boundless horizontal expansion of the music streaming market—more content, more users, more data, more ads—and to the vertical scaling up across different sectors, where losses in one market may create assets in another. For instance, constantly shrinking revenues for independent musicians, caused by the effects of scale on Spotify’s proportional revenue model, are projected to go along with increasing corporate treasury trading elsewhere. In this view, its market resembles a stack—where trading sites are stacked into or on top of one another, in often opaque, unaccountable, and unsustainable ways—rather than a two- or multisided platform.

Yet while there is reason to see Spotify’s claims about scale as an effect, or symptom, of a far wider-reaching financialization of media, a focus on processes of “infrastructuring” here speaks rather to Mary Poovey’s reminder that financialization “cannot be understood and should not be theorized apart from an analysis of the infrastructure that supports financial transactions.”⁴⁸ Put differently, we need a still more thorough empirical examination of how Spotify’s markets interrelate—or are embedded into one another—while ordering economic knowledge and activities. Here, advertising makes for a relevant case, given how it contributes to produce value from data, metrics, interactions, and patterns of listening behavior, thus contributing to the ways the world of streaming is made accessible. Advertising is key for infrastructuring the digital media ecology, as it creates the “practical ontologies” that shape the ways music looks, sounds, and feels.⁴⁹ Advertising was also key for scaling up Spotify Free. If the observation is correct that Spotify’s ad market is embedded in the financial market, then it should be possible, we argue, to present evidence that financial ideas, tropes, devices, and processes matter for the way advertising is implemented and practiced on Spotify. While such an empirical approach falls short of matching the data standards of economic modeling, it correlates well with the sociological, historical, and ethnographic methods that formed this book’s starting point.

`Embedded in Finance`

What is the advertising market about? Which goods are traded, and by whom? And how are things done? What are the beliefs, norms, tools, rules, and behaviors considered appropriate to the setting by market actors? Taking up the latter issue first, Spotify’s ad market has to be seen in light of a broader corporate culture that has evolved through practices of brokerage and arbitrage. Introducing the financial concepts of brokerage and arbitrage as analytical terms, we suggest that coordination within Spotify’s production chain followed and appropriated models from the financial world.

Internet critics often note that the overabundant catalogs of songs, videos, books, and academic papers that have come to epitomize platform culture would devalue content in order to maximize corporate gains. If “value depends on scarcity,” scaling up seems to imply devaluing audiences, content, and ad rates.⁵⁰ While not owning, producing, or servicing anything, the argument goes, Facebook, Uber, Airbnb, YouTube, Alibaba, and Spotify are “indescribably thin layers that sit on top of vast supply systems (where the costs are) and interface with a huge number of people (where the money is).”⁵¹ Scale, then, seems to depend both on the allegedly infinite “shelf space” of the internet and on a type of intermediary that would be able—or ruthless enough—to clear that shelf: a broker. As a theoretical figure and sociological type, the broker has a long history whose importance reemerges in contexts of state crisis.⁵² Brokerage not only relates to weakening ties between state and market, as well as the increase of transnational flows, but also to entrepreneurs with both a “special expertise in manipulating boundaries: between legal and illegal, commercial and noncommercial, formal and informal” and with the talent to relocate “transactions from one side to the other.”⁵³ The broker is a specific type of middleman: an actor who gains from the mediation of valued resources that he or she does not control. According to a well-known distinction by Bruno Latour, what distinguishes the broker as a mediator from an intermediary in the broader sense is that while the latter transports meaning without transformation, mediators “transform, translate, distort, and modify the meaning of the elements they are supposed to carry.”⁵⁴ Brokerage has always been morally ambiguous, especially where inequality forms the basis for brokerage opportunities and where the broker is seen to exploit an exclusive, central location in a given network and thus becomes identified with a structural function of regulating the circulation of values.

Spotify was founded with the stated aim of mediating the interests of two conflicting parties: the music industry on the one hand and unauthorized file sharers on the other. Spotting an opportunity in the structural hole between these two disconnected groups, the company’s brokerage role developed as that of a market maker.⁵⁵ Although Ek and his engineering colleagues themselves did not have anything of value to offer—neither a cultural good nor valuable contacts in the field of cultural production—they successfully began promoting trade by reducing the physical, social, and temporal distance between an industry (largely US based) and its economic free riders (mostly via Sweden’s Pirate Bay). In order to do so, they transformed the meaning of online music listening by shifting focus “from access to context,” so that a business model based on advertising revenue could take hold.

Bridging the gap between industry and piracy, Spotify’s initially situational role as a broker soon turned into a structural brokerage function in line with the company’s intensifying relations to finance. In this respect, Spotify’s prevalent classification as a tech company is somewhat obscure, since—from its very start—the company was unequivocally in the business of providing content to audiences while selling those audiences to advertisers. These are defining characteristics of the media sector that put Spotify in line with cable television and satellite industries. The firm’s original articles of association list “Internet-based services within digital media such as music, games, and television, and related activities,” and its first subsidiaries—Spotify Sweden AB (2009) and Spotify Service AB (2012)—were established to expand advertising sales and media buying, as well as public relations and communication, respectively.⁵⁶ Spotify still came to be identified with technological innovation rather than as a traditionally operating media firm, however, because “being thought of as a tech company brings with it the potential for much higher valuations from the investment community.”⁵⁷ Venture investing requires an industrial self-categorization such as Spotify’s—“data processing, hosting, web portals,” according to the Swedish Tax Agency’s industrial classification standard (SNI)—since investors see way more potential in the technology sector than the media sector.⁵⁸ And they especially did so throughout Spotify’s two initial series of venture capital funding, granted during the global financial crisis of 2007 to 2009, a period when the availability of such capital decreased—except for internet startups.⁵⁹

Today, Spotify is neither particularly Swedish nor particularly about music. While invocations of the company’s “Swedishness” are needed to sustain the venture capital vision of “European unicorns,”⁶⁰ and to position Spotify at the sexy, cool end of digital innovation, the company now acts as a digital broker whose history of equity rounds, market and debt capitalization, and board of directors firmly ties its brokerage strategies to US-based financial interests. As illustrated by the 2015 addition of television content, Spotify now operates increasingly like a conventional American media company while retaining the benefits of financial and regulatory loopholes granted to European tech firms.⁶¹ The bet on advertising—and on what was introduced as programmatic advertising (or behavioral targeting) at the 2015 New York event—has to be seen in light of these transnational flows. As a digital broker Spotify has, in short, contributed to reconfiguring the relationship between state, market, and music.

But brokerage is not the only financial practice to be found in the way the service coordinates economic activity on and off its interfaces. Of equal importance, we argue, is arbitrage. Arbitrage, in market practice, is a form of trading “that aims to make low-risk profits by exploiting discrepancies in the price of the same asset or in the relative prices of similar assets.”⁶² Not only is arbitrage a widespread form of trading, it also builds markets and determines their scope and the extent to which they become global. When Daniel Ek, Martin Lorentzon, and others first developed Spotify, they famously did so by building on a collection of music they did not hold any rights to themselves. The history of music streaming thus begins with an act of free riding—or arbitrage. As arbitrageurs of music, Ek and his colleagues would obtain scarce goods at no cost, with the aim of making revenue through advertising and later subscriptions, while others involved in the transaction—the composers, musicians, music publishers, artists, and repertoire owners—experienced an implicit loss, at least initially (the majors), if not still today (the independents). While arbitrage loopholes may gradually be closing, what remains is an “art of association” that consists in constructing the equivalence of properties across different assets.⁶³ In other words, the art of making one thing the measure of something else.

Arbitrage thus creates a sense of comparability—that one may measure the value of music against the value of audience segments or subscriptions. But what counts when it comes to cultural goods, and how can they be counted? Although companies such as MyPlay.com experimented with the conversion of music into targets starting in the late 1990s,⁶⁴ they did not develop a business model for this conversion to scale. In order to properly establish a scaled exchange, Spotify had to define a unit of what is measured—a measurement standard—and a forum where exchange could take place. The evolution of Spotify’s graphical user interface documents how this definition came to materialize.

Think, for instance, of the relation between playlists and the ad banners that have been a key feature of the user interface since the earliest versions of Spotify Free. If arbitrage is not about identifying the essential or relational characteristics of two assets so much as testing ideas about their correspondence based on isolated qualities of each, then playlists and call-to-action banner ads have come to mark this correspondence between music and advertising over the years. As we described in chapter 3, associating mood-oriented playlists with psychographically targeted display ads means focusing on a quality of music described as “neo-Muzak,” or as a turn from sounds toward “personal care products for affect management and mood elevation.”⁶⁵ Spotify is “buying” such products increasingly under value, because prices for most of its (non–major label) music continue to fall due to the effects of scale on its proportional revenue model. And it is “selling” at an increasingly higher price—or is at least attempting to drive up costs for advertisers in order to do so. While such arbitrage trading is usually deemed unfair and is expected to end as soon as the buying and selling sides match up, Spotify’s pro rata share model and resulting royalty payments are unlikely to change.⁶⁶

Brokerage and arbitrage thus provide us with an analytical framework to describe Spotify’s corporate practices as tied into the logics of financial markets. Broader tendencies within company culture suggest that key decision-making incidents relate to a governing market and actors that are not primarily or solely about music. Finance is thus not simply another side of the same market, as implied by the notion of the two-sided or multisided market. Rather, it is a form of internally ordering the production chain that ranks finance above other considerations. We now need to examine what the advertising market is about, who is involved in that market, and what relations it holds to the financial logic assumed to govern the entire chain.

`Automation—or Programmatic Advertising`

Programmatic—the industry shorthand for programmatic advertising—is an obvious case to be considered in a context of financialization, given that its practices were directly and literally modeled on those of the stock exchange. Introduced in 2008, programmatic is a new transactional procedure of media buying that is projected to replace manual insertion-order (IO) inventory by 2018.⁶⁷ Put simply, it is a mechanism for using personal data and algorithms to buy and sell ads. This includes the automation of online ad buys via interconnected online “trading desks” that allow the auctioning off of inventory within milliseconds. Such trading desks are not accessible via Spotify’s desktop or mobile interfaces but come as separate interfaces connecting buyers and sellers directly via real-time bidding procedures. This means that the entire ad buy market remains entirely invisible to consumers and musicians—and, importantly, that pricing is never made public. Programmatic is a matter of coordination between marketers, ad-tech firms, and Spotify, with anonymity on some sides of the market, as marketers often do not know on which platform they bet.

In principle, such automated trading of ad buys yields several advantages. It allows the sale of overabundant inventory that human sales teams have failed to sell. Thus, ad auctioning is a clear indicator that the economic value of the offer (the ads on Spotify) has not been settled and cannot be settled without an auction. While Spotify had ads long before programmatic, it is only through auctioning that a functioning market could be established. Programmatic also creates larger scales more efficiently by allowing advertisers to aggregate interactions with audiences across the web without first having to select audiences manually across thousands of publisher sites. For ad agencies and media planners, establishing automated trading desks was a defensive measure against Google and other players that ensured their own relevancy in a field where brands were able to buy directly from ad networks.⁶⁸ Low barriers of entry and high profitability created additional incentives that fueled ad-tech market growth.

Spotify promotes its platform to advertisers as an innovative setting for programmatic targeting techniques. The company’s 2015 New York media event gives an idea about the value it saw, and sees, in “more targeted advertising.” Ad targeting techniques include both demographic targeting and targeting content at users with particular habits, mindsets, and tastes that align with a predefined target persona. Playlists tailored to specific urban activities (such as “Morning Commute”) and moods (such as “Life Sucks”) are combined with data on genre preferences, age, gender, geography, language, and streaming habits, alongside information from third-party data providers about broader interests and lifestyle and shopping behaviors. It is, in short, a business model based on technologically aided information exchanges—not music. Music is rather promoted as merely functional for defining and microtargeting audience segments. The main promise of programmatic advertising is its competitive edge when it comes to demonstrating advertising’s new “relevance to the consumer.”⁶⁹

While Spotify upholds the success of its efforts to match ad content to listening behaviors, members of the Spotify community regularly criticize the service for its lack of accurate targeting. They note, for example, that some ads do not match basic data sets, including the user’s age, gender, location, language, genre preferences, and listening context. Such criticism is partly misdirected, however, as it wrongly puts the blame on publishers for what are, in effect, decisions made by advertisers, most of whom do not microtarget their ads but instead opt for broad media reach, always depending on overall ad campaign goals and disposable budget. Spotify itself has pointed out that ads serve the dual purpose of generating a revenue stream and encouraging advertising-adverse users to pay for Spotify Premium, a goal demonstrated by the fact that, as of August 2016, the most repetitive and frequent audio and display ads indeed promoted Spotify’s own ad-free service.⁷⁰

In order to develop a more tenable critical position vis-à-vis the market coordination enacted through programmatic advertising procedures, better empirical data are needed. Occasionally, Spotify employees leak such data in semipublic contexts, but industry leakages are seldom helpful for developing industry-critical perspectives.⁷¹ Between August and October 2016, we therefore conducted a number of simple experiments in collaboration with programmer Roger Mähler at Humlab. Our objective was to gain insight into the ad-tech market, an issue not reliably covered via trade journals or interviews.⁷² The experiments used the Ghostery browser plugin for Spotify’s web player, which allowed us to map various stakeholders involved in placing ads, and Fiddler, a tool for capturing network data.⁷³ By establishing a Facebook user ID connected to Spotify, we were able to identify some of the advertising supply-chain vendors involved in the placing of ads related to that specific user profile. While these findings are indeed limited, they allowed us to confirm a few assertions about Spotify’s production chain.

Ghostery and Fiddler provided us with a list of various companies that take part in networked interaction when browsing through a user’s playlist stats. Some of these companies provide trackers, widgets, and analytics, while others measure performance. However, most are related to advertising, including (but not limited to) the following groups of actors to be found in programmatic ad sales:

Supply-Side Platforms (SSPs): AdScale, PubMatic, Rubicon, and YieldLab

Demand-Side Platforms (DSPs): AdRiver and Sociomantic

Ad Exchanges: Admeta, AppNexus, Facebook Exchange (FBX), OpenX, and Yahoo Ad Exchange

Ad Networks: Adkontekst

Ad Servers: Adtech

Data Suppliers: Seed Scientific and the Echo Nest

Data Management Platforms (DMP): BlueKai and Krux

Measurement: Moat and Google

Verification: ComScore and Nielsen

As in an actual stock exchange, the auction of Spotify audience segments is constant and continuous. Real-time bidding is managed via ad exchanges and digital marketplaces. In this brokerage setting, a publisher such as Spotify partners with one or several yield-optimization or supply-side platforms (SSPs). This partnership is designed to maximize the sale price for Spotify’s impressions and to aggregate and manage Spotify’s relationships with ad buyers as well as with ad exchanges and ad networks. Through the SSP, the publisher may determine whom it wants its inventory bought by, the type of ads it wants displayed, and the range of pricing: preferences that are set before trading starts. Once a publisher’s sales partner or SSP (such as Rubicon) begins throwing available ad impressions into automated ad exchanges (e.g., Admeta), demand-side platforms (DSPs), such as AdRiver, analyze and purchase them on behalf of marketers and brands based on attributes such as location or specific targets, with the objective of buying ad impressions as cheaply and efficiently as possible. In addition, publishers may also presell part of their inventory at a fixed price to one or several ad networks (e.g., Adkontekst) for a particular targeted audience. From a marketer or brand perspective, the exchange may start differently: by simply buying a “whitelisted” inventory, which would also include Spotify; by accessing Spotify’s inventory through a DSP for open auction bidding; or by directly contacting Spotify in case something more than just inventory is wanted, such as when first-party data are required in order to target a brand in relation to content or demographic data, as in so-called private (prenegotiated) marketplaces.⁷⁴

Trading is fueled by consumer data that are extracted by various data suppliers and aggregated by one or several data management platforms (e.g., BlueKai). As soon as the inventory has been sold, several media- and format-specializing ad servers (e.g., Adtech) will deliver the ad units to the website, while the viewability and effectiveness of the ads are monitored (by Moat or Comscore). Although the basic currency of the exchange is cost-per-mille, buyers and sellers measure differently: a marketer may optimize toward viewability or click-through settings in the DSP platform, and a publisher toward number of buyers or revenue per thousand impressions. The entire exchange process from the client’s initial linking up to the publisher content server to the final placing of the ads takes less than thirty milliseconds.

Somewhat complicating the picture, several of the aforementioned firms perform multiple roles in the supply chain and may be directed toward either publishers or marketers, or both (e.g., AppNexus). None of them is exclusively working with Spotify. Spotify claims to partner exclusively with Rubicon as SSP, but in our experiment, Ghostery listed a number of specialized SSPs that may serve specific ad format- or region-related purposes.⁷⁵ Dozens of firms, clustered in digital space but operating from often remote countries, offer highly specialized, competing or complementary services that are partly tailored to regional ad markets, subject to a constant transformation of technology and industry, and always actualized in their relations depending on a given brand and target. Automation thus entails a super-intermediation of media buying. In the words of the Internet Advertising Bureau, it leads to an “exploding number of ad technology products and services, new cost models, and an evolving patchwork of bundled buy-side offerings and solutions.”⁷⁶ As Jana Jakovljevic, Spotify’s head of programmatic advertising, admits in an interview, it is a “lot of middle-men …, definitely confusing, even more so for the buy side.”⁷⁷

Our exploratory and experimental mapping of Spotify’s ad-tech infrastructure allowed us to draw a few tentative conclusions about the company’s ad market. For consumers and musicians, this is indeed an opaque market. Even for actors representing the advertising side, the ad market has a peculiar form of order. Its technical environment and network of involved actors are highly unstable, not only because they involve a constant switching of roles but also because of the constant expansion and accretion of the network itself. Ad tech coordinates the market through auction procedures, yet the idea of a stock exchange is misleading when it comes to the distribution of costs and benefits between actors. Ad tech increases the costs for publishers and marketers, instead of reducing them. Programmatic media buying is more expensive (especially regarding costs per exposure) than insertion-order advertising, adding an “ad tech tax” to the budget, often without transparent pricing.⁷⁸ It also disrupts long-established relations between ad agencies, media planners, and publishers. Here, automation may serve to dispose of human intermediaries and to restructure and renegotiate the input of any actor. It is, in short, not so much a form of trade modeled on the stock exchange as a form of incrementally stacked exchanges.

This chapter has shown that Spotify’s ad targeting is one element in a process that goes beyond the mere dissemination of digitized sound. Spotify is a mediator, rather than an intermediary, that actively reproduces the meaning of the songs it is supposed to distribute. This reproduction is engineered through a music classification and recommender system whose output is data for advertising. The ad market constitutes but one in a family of markets connected by the Spotify brand. Given the opacity of some of these markets, the internal ordering among them remains speculative. It is likely, however, that the financial market governs all others, given the valuation reached for Spotify and the way corporate practice seems tied up into the logics of finance. None of its markets build primarily on network effects, but there are implications of hierarchy in the way songs, ads, and targets are subsumed under such logics. Somewhat ironically, these logics even put wealthy publishers such as Spotify in a precarious position, because it turns them into brokers depending on other brokers for realizing globalized claims on scale—middlemen on top of middlemen.

As a more general result of this empirical market study, future research is needed to highlight the way digital markets are always embedded in other markets.⁷⁹ We have argued that such mutual embedding, or relatedness of markets, is exactly what defines an industry. So far, Spotify, Facebook, and Google have often been studied primarily as platforms in the sense of techno-economic configurations somehow isolated from or elevated above the manifold activities that constitute markets on- and offline. Platform studies emphasize that digital markets are ontologically distinctive and multisided, imbuing internet businesses with ideas of newness and networked egalitarianism. As the ad market case study has shown, there certainly are positive effects in the interaction between Spotify’s various market actors, such as increasing ad revenue. But when considered as a media industry, Spotify and other services constitute production chains that also cause a number of negative externalities. While programmatic advertising has only spread since 2008, the experience is of a higher volume of ads everywhere. Its spread has disrupted established relations between ad agencies, media planners, and publishers. With its millisecond processing of personal data and predictive behavioral targeting capacities, programmatic also engineers a new, uncanny intimacy between publishers, cultural content, and audiences.

“Finance can only spread as far as its own magic,”⁸⁰ Anna Tsing has observed. Our research has shown that Spotify’s magic indeed exists and cannot be discounted as fake. This includes a new belief in the magic of music as commodity, despite music as a business having been declared dead on numerous occasions. It includes the wonders of free and Spotify’s subsequent conversion of free listening into scale and value. Spotify’s magic also invokes acts of corporate staging, cultural border crossing, and the manipulation of boundaries in ways that connect the world of finance to our own imagination. The value of free, in this context, may rest on us imagining other forms of freedom.

4 What Is the Value of Free?

A Brief History of Free

Entering the Market

Scaling Markets

Embedded in Finance

Automation—or Programmatic Advertising

Notes