CONSTRUCTING SUSTAINABILITY THROUGH DATA: THE SOCIO-TECHNICAL PRODUCTION OF MARKETING INSIGHTS

This thesis explores the socio-technical construction of marketing insights in the age of Artificial Intelligence by focusing on sustainability, a theme increasingly central to consumption practices and discourses. The theoretical background of this work lies at the intersection of sociology, market studies, and infrastructural studies. After reconstructing the development of market research and how it has changed since consumer culture has become increasingly datafied, it proposes analysing it through a performative and infrastructural lens, conceptualising marketing insights as outputs of a market infrastructure that embeds specific values, constraints, and ways of seeing the market. To investigate this idea, this research employs a mixed methodology. It is rooted in ethnographic observation of a market research agency anonymized as “Predict”, which uses AI to analyse consumer data and forecast future trends and is accompanied by an analysis of the data Predict uses to create insights on sustainability, including a computational and visual analysis of a sample dataset of about 120,000 posts collected and processed by Predict. The combination of these methodologies allows the research to open the “laboratory” of market research and explore not only how insights are presented and constructed by market researchers but also the raw data used in this process. The findings are presented across three interconnected empirical chapters. The first chapter explores Predict as an infrastructure, emphasising how its organizational practices, technological systems, and embedded values shape the production of marketing insights. The second empirical chapter examines the dataset offered by Predict, focusing on how sustainability is represented within the data and arguing that the dataset is curated in alignment with Predict’s infrastructural values. This illustrates how market research prioritises certain types of data over others to maintain alignment with corporate objectives. The final chapter introduces the concept of sanitization, describing how market research filters and categorizes data to align with infrastructural biases, resulting in partial and decontextualized portrayals of sustainability. As such, it provides a definition of sanitisation and identifies its key contributing elements: partiality, speed, and ideology. Despite its limitations, this work aims to open the laboratory of market research and provide a new understanding of the socio-technical creation of marketing insights.