The Construction of Audiences through Big Data Analytics

The first ECREA 2014 panel session is one that we have a paper in as well, but we start with Göran Bolin, whose interest is in the construction of audiences through big data-driven research. There is now a large discussion of what constitutes big data, of course, but Göran is bypassing this, focussing instead on the uses of such data to envisage media users.

Much such data are drawn from social media, but it seems that the social is getting further estranged through this; big data means an intensification of statistical and quantitative approaches, and to some extent constitutes a shift in the nature of scholarly enquiry, focussing on mapping networked accumulations. Such accumulations stem from the metrics produced by social media platforms, for example (such as like and friend counts); and interesting experiments are being conducted by taking away such metrics and thinking through how this changes the experience of such spaces.

But statistics have long been part of the project of modernity; mass media audiences have long been quantified, measured, and even anticipated. Yet the move to big data is also a move towards a panspectric rather than panoptic mode of tracking; it is about detecting rather than describing activities, and generates non-representational correlations.

Historically, media audiences have been understood as readers, listeners, and viewers; advertising turned them into a commodity. Later, audiences were understood as active agents and workers; they were then positioned as raw materials, and finally used to extract productive behavioural patterns.

Today, big data has become a buzzword, but the reality of big data analytics and correlation is a great deal more messy; online systems are rarely exhaustive and reliable, data remain patchy and unstructured, and there is a great deal of further work to be done to develop reliable and stable tools and frameworks for working with big data. Analysis, in a media context, is utilised by producers to anticipate user demographics and behaviours, and by users to orient themselves in relation to the media spaces they operate in.

On the one hand, this positions users as part of the mass (as not special); on the other, it positions them as unique individuals who are being profiled (as special). User data are being translated into familiar categories (gender, age, income, etc.) to determine distribution curves and anticipate commercial efficacy; into sociograms that enable surveillance agencies to find individuals ("black swans") to anticipate risks; and into algorithmic mindsets (creating networked accumulations) to anticipate viral mechanics.