The last chapter closed by arguing that the discourse of science, represented in talk and materiality, is a way for marketing organizations to start the process of uncertainty absorption. By projecting an image of themselves and marketing action using scientific references, marketing organizations can increase their clients’ confidence in their work and, we might also assume, overcome some of the ambiguities of marketing discussed by Alvesson (1994; 2001). They imply that they possess a set of technical skills capable of uncovering information about consumers and markets that their clients lack.

Does this mean that Marketing Science has replaced Marketing Art as the way to bedazzle clients? Well, when we look at the way the research team works at Super we actually see that, while they present their activities using general scientism and specific scientism, their work also involves creativity and imagination alongside data science and computation. Specifically, as we will see, the need to embed clients’ assumptions, interests, and expectations in their research requires Super to be more flexible than the scientistic presentation of their work suggests.

This switch from the discourse of Marketing Science to Marketing Art is clearly demonstrated in Super's facial coding product, Emotionality. As I entered the field, Super had recently launched this new product offering. Emotionality is promoted as a way to measure Kahneman's (2011) notion of System One thinking. Dan describes it to clients as a ‘super high level predictive analysis’ that exposes ‘what people don't know and won't tell you’. In short, it is meant to reveal consumers’ inner truths – ones that are perhaps even more true because the consumers do not know them themselves. In this way, Emotionality is framed as a method that removes the mediation imposed on traditional marketing research through research participants (who must tell you what they think, something that is difficult when they don't know what they think). Instead, computers do the work – or at least that is what Super wants its clients to think.

The problem is that computational analysis presents the research team with too much data but not enough information. It creates what I callsurplus research and empty signifiers.