CPAN Chart: A Novel Customer Perception Analysis System Using Natural Language Processing and Attribute Control Charting

Customer reviews of a product or service collected over a period of time may reveal important directions about the changes required in the current functioning of the organization. In this article, a customer perception analysis chart is developed, which will help the business owners to monitor the proportion of negative reviews on a temporal scale, augmented with the information on the most pervasive consumer grievances. The proposed method incorporates sentiment analysis and topic modeling with an attribute control chart to show the proportion of negative reviews received per unit period of time and highlight the most significant issue of the same period. Using theoretical constructs derived from the domain-specific literature and topic modeling, hypotheses related to the influence of the quality attributes on customer satisfaction level are developed, which are then validated using regression-based confirmatory analysis on the organization's review corpus, providing valuable insights on the relationship between the attributes and customer perception. The primary contribution of this work is to combine the power of natural language processing and control charting techniques to analyze customer reviews and derive actionable insights. The implementation of the proposed method is presented through a case study using the review corpus of a luxury hotel.