Dynamic product quality improvement using social media data and competitor-based Kano model

Recently, social media data have emerged as an important information source for product quality improvement. Previous studies utilize social media data and the Kano model to delineate opportunities for product quality enhancement. However, a favorable performance appraisal from customers does not guarantee that the product gains a competitive edge over its rivals. Performance evaluation may be inaccurate and biased without comparing the product to its competitors. Moreover, few studies research dynamic quality attributes and propose strategies for dynamic quality enhancement. In this study, we propose a dynamic approach to improving product quality using social media data. Firstly, we categorize social media data into texts with positive, negative, and neutral sentiments. By considering competitors, we then dynamically estimate product performance and customer satisfaction using these texts, Dynamic Topic Models, and sentiment analysis. Secondly, we cluster product features into diverse quality attributes using our proposed competitor-based Kano model. Finally, we use a case study of laptops for validation. Experimental results not only validate the lifecycle of quality attributes (changing from indifferent to attractive attributes, which turn to one-directional/must-be attributes) proposed by extant research, but also reveal new trajectories of quality attributes (fluctuating among one-directional, must-be, and attractive attributes). Additionally, the comparison with baseline methods manifests that our method provides manufacturers with timely decision support to improve product quality, formulate product development strategies, and enhance product competitiveness.