An Ethnography of Advertising Categories in the Age of Big Data

• In Event: Data and Markets: Devices, Practices and Representations - II

Abstract

What do digital technologies and big data do to the categories and classifications that shape and inform our economic and social activities? Do they contribute to a weakening of pre-existing conventional categories that maintain coordination by fostering the production of new, ephemeral, categories (Boullier, 2017)? Do they exacerbate the effect of existing classifications by enclosing individuals in filter bubbles (Pariser, 2011; Turow, 2012) or by delegating inclusion/exclusion operations to machines (Eubanks, 2018)?
We address these questions by studying the advertising industry. Traditionally, all the actors in the advertising food chain articulate their activities around "socio-demographic" classifications (Napoli, 2003). Recent innovations in online advertising open the box of the algorithmization of categorization operations, by mobilizing a wide variety of data sources to build micro-segments of consumers, and by delegating the manufacture of audience segments to machine learning algorithms. Today these tools equip the ordinary work of media planning professionals.
The purpose of this research is to study empirically how these audience planning devices are used by professionals. How are hyper-targeting tools mobilized in practice? How do they combine with traditional advertising categories? This research relies mainly on interviews with media-planning professionals using these tools on a daily basis. We show that, contrary to the “end of theory” narrative (Anderson, 2008), the algorithmic creation of targets is seldom used, and that micro-targeting tools are mobilized for costly hand-made operations. Instead, the "custom audience", first-party information on consumers owned by advertisers, underlines the growing importance of local knowledge and ontologies.

Authors

• Kevin Mellet, Orange Labs

• Thomas Beauvisage, Orange Labs

• Jean-Samuel Beuscart, Orange Labs

• Samuel Coavoux, Orange Labs