4 days Qualitative Research and Data Science. The common denominator: NLP (Natural Language Processing)

Over the past years, the Esomar Fusion Event has merged the two conferences “Qualitative” and “Digital”. However, the 4-day conference still continues to be divided into 2 days with the focus on Data and 2 days with more Qualitative Research ; but overlaps and hybrid methods are desirable in both. It is interesting to see how one topic is worked on from both “directions”: Natural Language Processing (NLP).

From Social Listening to E-Seeing

One of the most interesting presentations was the “Using A.I. to spot a client’s irritating experience” presented by Ipsos (Charlotte Zaepfel, Mathilde Guinaudeau).

A data scientist and a market researcher presented in a very amusing way and according to their roles (good cop – bad cop = Algorithms-Focused Data Scientist vs. “So-What?”-focused market researcher) how they link CRM data and social listening data via the open entries in CRM surveys. It is well known that one of the limitations of social listening for market research is that it is impossible to understand who said what in which context.

Here, Ipsos has combined verbatims collected via weblistening with verbatims from the open questions in the CRM satisfaction survey and extrapolated them to further CRM data via a kind of Look-Alike-Modelling.

Under the title “Using social media to understand the beauty evolution through influencers”, L’Oréal (Alberto Rodríguez Romo, Estefanía Yaguez) presented how they increasingly work with images instead of texts (which is not so new anymore), and t the processing of images is, of course more scalable, since images, unlike texts, do not have to be translated.

Focus Vision, with former chef, now Data Scientist Mike Kuehne, has introduced its new video-to-text transcription solution (“Using knowledge models for video highlight extraction”), which allows Qualitative Market Researchers to process verbatims collected via videos much faster using NLP.

NLP

As mentioned at the start, in my opinion Natural Language Processing (NLP) was the connecting leitmotif of the two sub-conferences. Data Science uses NLP to process vast amounts of existing unstructured data, combining the depth of qual with measurements in quant. Qualitative market research solves the problem of sample size and numbers by analyzing much larger amounts of qualitative data much faster.

My special moment : Of course OUR presentation 😉

Benjamin, our Chief Consulting Officer, and Charles Mezerette from leboncoin presented our study Gen Z : some like it old. In this study, we explored the shopping behaviour of 16-25 years olds in France and especially their usages, attitudes and expectations towards 2^nd-hand-products. To do so, we combined passive metering of web behavior with our quantified qual online community approach.

Networking

Each day there were about 80-90 people at the conference, with some exchanges from Tuesday to Wednesday (Data to Qual). At such a “small” conference the exchange between colleagues is much easier. In addition, Esomar organized a small party with tapas, wine, ham and manchego every evening ,where one could continue the conversations in pleasant and easy-going atmosphere.