ANALYZING USER-GENERATED REVIEWS TO IDENTIFY EXPERIENCE DIMENSIONS AND THEIR IMPACT ON SATISFACTION WITH MOBILE BANKING APPLICATIONS

Abstract

Purpose: This study aims to identify key experience dimensions derived from user-generated reviews and examine their influence on satisfaction with mobile banking applications. By analyzing unsolicited user feedback, the study aims to provide a data-driven understanding of what drives satisfaction and dissatisfaction in m-banking services across Southeast Europe.
Methodology: A total of 56,074 reviews of mobile banking apps from Bosnia and Herzegovina, Croatia, and Serbia were analyzed. The study employed BERT-based sentiment analysis and BERTopic modeling to extract thematic experience dimensions from reviews categorized by sentiment polarity. Multiple linear regression was used to assess the predictive power of these dimensions and sentiment polarity on userassigned satisfaction ratings.
Results: Eleven distinct experience dimensions were identified and confirmed as significant predictors of satisfaction. These dimensions, derived through topic modeling and sentiment analysis, were quantitatively modeled and qualitatively interpreted based on review content, offering rich contextual insights into user perceptions. The regression model explained 58% of the variance in rating behavior. Additionally, sentiment polarity emerged as a strong predictor, further enhancing the model’s explanatory power and underscoring the emotional weight embedded in user reviews.
Conclusion: This study shows that user satisfaction in mobile banking is shaped by a combination of functional and emotional factors, which can be effectively identified through text mining. Including sentiment polarity alongside experience dimensions offers a more robust framework for evaluating digital service quality. These insights can support academic model development and practical efforts to optimize m-banking app design and communication