Application of natural language processing to enhance qualitative research used for marketing

Poj NetsiriMarketa Lhotáková

Abstract

Understanding consumer behavior can help improve marketing of the products. Market research normally applies conventional qualitative analysis to discover the reasons why consumers act and purchase products in a certain way. However, qualitative analysis with small samples is insufficient to make population-level summaries. On the other hand, qualitative analysis with large samples is time consuming. In addition, poor quality of qualitative research due to human error and bias from the researcher can lead to misleading findings. Therefore, to overcome these problems, Natural Language Processing is applied to extract consumer behaviors from large-scale samples of product reviews. The result from 809 product reviews (source: CarMax in US) of preowned luxury cars (Mercedes-Benz) indicates top 10 relevant keywords “ride”, “smooth”, “luxury”, “nice”, “feature”, “excellent”, “beautiful”, “comfort”, “style”, and “expect”. These terms correlate with consumer perceived emotional, social, and quality values that could positively influence customer purchase intention toward preowned luxury cars.

Keywords

Consumer behaviour, consumer perceived value, qualitative analysis, natural language processing, topic modelling.

DOI

https://doi.org/10.18267/pr.2023.kre.2490.13

Cite this paper

Netsiri, P., & Lhotáková, M.  (2023). Application of natural language processing to enhance qualitative research used for marketing. In: 23rd International Joint Conference Central and Eastern Europe in the Changing Business Environment: Proceedings. Vydavateľstvo EKONÓM, Bratislava. pp. 171-180. https://doi.org/10.18267/pr.2023.kre.2490.13

 

Download the PDF

 

Back to Proceedings

The websites of the Prague University of Economics and Business use cookies to optimize their content and services provided. Please give us your consent to use of them.

Select the services for which you want to allow the use of cookies: