As well as greetings and sales, January also brings forecasts for the year ahead and a series of themes that have been playing over and over again in our marketing ecosystem: blockchain, connected objects and robots, big data and artificial intelligence.
- Big data is not a new subject: in fact, 2016 saw the recruitment of many data scientists in companies - sometimes complementing, sometimes conflicting with research departments, mainly in the field of quantitative research.
- Artificial intelligence has more recently entered the debate, bringing a more qualitative and explanatory dimension to results that have hitherto been considered from a purely quantitative angle.
Let's take the example of verbatim comments collected en masse on social media: automating their semantic processing quickly led to dead ends, with some of the meaning emerging from figures of speech such as the litote or the antiphrase: without an in-depth analysis of the context, it was impossible to know whether "that's great" really meant... "that's great", or on the contrary "what stupidity"!
The late Umberto Eco, in Lector in fabuladistinguishes four levels of analysis for any narrative: "discursive" structures (very close to the linearity of a text), "narrative", "actancial" and finally "ideological... invested with value judgements"; in the Fashion SystemRoland Barthes refers to the "phraseology of the newspaper (which) constitutes a connotative message, intended to convey a certain vision of the world".
A simple semantic analysis will certainly only address the most elementary level of messages: simple denotation in Barthes, discursive structure in Eco, leaving aside the majority - and the depth - of meaning: Artificial intelligence will take an interest in the complexity of the discourse as a whole, in order to distinguish between a simple statement and an antiphrase: a left-wing voter's "that's great" response to a measure taken by a right-wing politician will have a completely different meaning to a right-wing voter's "that's great" response to the same measure.
In short, it represents undeniable progress in the field of qualitative and quantitative research... although it is not the ultimate answer to every problem: at best, it is a new analytical tool - certainly a powerful one, and one that we will have to learn to master.
The fact remains that, even if it is highly instructive, spontaneous expression on social media - even if all the risks of manipulation are ruled out - will only ever reflect part of what consumers think; there will always be a significant amount of unspoken information that needs to be gathered through observation, interviews or group meetings.
Over the next few years, the range of qualitative research is likely to expand significantly, which is good news if we are to get a better grasp of an increasingly complex consumer.
