Measuring Service Quality has bas been an area of interest for researchers since 1980s. As the retail banking institutions become more customer centric, the focus on customer service quality is increasing across the world. Pre-existing service quality frameworks such as SERVPERF and SERVQUAL have been applied to evaluate the quality levels in banking. However, these frameworks are expensive, as these instruments need to be replicated across the bank branches. With this in consideration, through this study, we have explored a cost and time effective approach to approximate SERVPERF model based on sentiment analysis of online reviews on various social media sites. This paper proposes an innovative approach to measure service quality in a cost effective way. In this paper, our main objective is to analyse customer reviews to better understand banks' service quality and performance. We have collected large number of online reviews from a website for three private Indian banks. Our data set is distributed into three banks that have a similar proportion of reviews (33% each). For each bank, we have a similar mix of products- Credit Card(15-31%), Loan(60-71%) and CASA(9-11%). Further, we also note that the average ratings across banks and products reveals that customers feel differently about different products and banks. The reviews have been mapped to RATER dimensions of SERVPERF model, followed by calculating sentiments for each of these dimensions by adapting an upcoming field in informatics called as text analytics. Finally, a logistic regression model has been developed to understand importance of RATER dimensions in the mind of consumers. Our results show that improved sentiment on RATER dimensions especially on Tangibles and Responsiveness can lead to enhanced customer satisfaction.
Somnath Chakrabarti, Indian Institute of Management (IIM) Kashipur, India
Deepak Trehan, IIM Kashipur, India
Mayank Makhija, Global Consulting Firm, India
Stream: Business Administration and Business Economics; Marketing; Accounting
This paper is part of the EBMC2016 Conference Proceedings (View)
View / Download the full paper in a new tab/window