Studying Ad Targeting with Digital Methods: The Case of Spotify

Online advertising is a matter of public interest. Ten years ago, many of us would not have cared much about how Facebook, Google or Spotify place ads on their sites, and how they target particular constituencies of buyers or voters. This has changed since 2008, when programmatic advertising was introduced, an automated procedure of ad buying. Programmatic advertising largely lacks human oversight, making advertising an algorithm-driven business. The procedure enabled $100,000 worth of ads being placed during the 2016 U.S. presidential election by inauthentic accounts that appeared to be affiliated with Russia. It allows Facebook ad buyers to define target groups such as “Jew Hater,” “Second Amendment,” “Hitler did nothing wrong,” or “Nazi party,” which in turn makes it possible to feed such groups with divisive messages. Platforms have taken an active role in spreading misinformation through advertising. They also monitor user behavior on a large scale. Facebook, for instance, obtains detailed dossiers from commercial data brokers about users’ offline lives, and users have limited means to opt out of their data being used (Angwin et al 2016; Madrigal 2017; Meyer 2017). Spotify, the Swedish music streaming service, has a less controversial reputation. Introducing programmatic ad buying in 2015, however, the company has made no secret of its abilities to collect data on user behavior. In November 2016, Spotify launched a global outdoor ad campaign with ads jokingly showcasing massive aggregate data sets: “Dear 3,749 people who streamed ‘It’s the End of the World as We Know It’ the day of the Brexit vote, hang in there” (Nudd 2016). Spotify does not just collect “an enormous amount of data on what people are listening to, where, and in what context,” as one of its executives stated in public (Terdiman 2015). The company also acts as a private data broker, providing this collection of contextual data to marketers for ad targeting purposes. Spotify offers