Social media analytics for mining customer complaints to explore product opportunities

Abstract

This research develops a social media analytics (SMA) methodology to analyze product complaints from social media for exploring product opportunities. The methodology contains the following steps. Sentiment analysis is used to select negative posts that contain customer complaints related to unsatisfactory products. The latent Dirichlet allocation (LDA) topic modeling technique is applied to extract latent topics from the negative posts. Then, a new opportunity evaluation framework for handling consumers' complaints is developed to discover important and emerging product topics. Topic engagement analysis based on user engagement indicators is used to measure how well the user attention to each negative topic, while topic emergence analysis based on the generalized linear mixed model (GLMM) is to measure the topic emergence level. Further, this research formulates a new product strategy map that prioritizes negative topics for developing product improvement strategies. Finally, KeyGraph based on the chance discovery theory is applied to further explore product opportunities based on less frequent but important terms in the collected textual data for enhancing consumer experience. A smart home application is used as a case study. Ten topics are extracted and several critical topics are identified, such as “after-sales services”, “home security systems“, and ”media streaming devices“. In addition, through KeyGraph analysis, several product strategies are recommended, including offering quality after-sales services, increasing product and service reliability, and providing seamless integration of various home devices. The developed methodology can help firms discover critical product topics and provide insightful information for developing product improvement strategies.