View from Reviews: Mining Service Quality from UK Retail Banks Customer’s eWOM

Abstract

The recognised SERVQUAL framework assists in uncovering disparities between customer expectations and perceptions and is widely utilised in the field of service quality assessment and management. This study employs text mining methods, mapping SERVQUAL dimensions and assessing the influence of sentiments on customer star ratings through logistic regression. Adopting the tokenisation method of natural language processing, this study utilises 70807 customer-generated Trustpilot reviews of 33 UK banks to gauge social perceptions of banking service quality. The reviews dataset collected for the period of 3 years from 2018 to 2020, related to savings, personal and current accounts users. The study reveals a positive correlation between all SERVQUAL dimensions from reviews and review ratings, with Tangibility, Empathy, and Reliability being the most impactful. The study also examined review length as the moderator and find that Review length has a weaker effect on online bank ratings for Assurance and Responsiveness dimensions. The weak moderating effect of Assurance and Responsiveness is influenced by stringent UK bank regulations and a culture that expects high service standards. The study contributes significantly to interdisciplinary research, reinforcing the expectancy-value framework in marketing and banking. It guides bank service managers to focus on richer qualitative eWOM with customer sentiments in online reviews. Bank managers with this proposed methodology can integrate open-ended questions through text mining and econometric techniques for a comprehensive assessment of customer satisfaction.