Abstract
To understand what makes customers loyal, we analyzed thousands of reviews in both Arabic and English. We framed our research using the DeLone and McLean IS Success model to see how specific factors like system quality and service quality lead to satisfaction and loyalty. We used a mixed text-analytics approach to turn these reviews into data. First, we gathered reviews and used topic modeling to find the most common themes. Next, a team of coders created a guide to label reviews based on system reliability, responsiveness, customer support, and security issues. Finally, we used a machine-learning model to apply these labels to all reviews. To find out which factors are the strongest predictors of loyalty, we used tools called LASSO and Ridge regression, which help pick the signals that matter most from the text. Our findings show that four main themes drive user satisfaction. System reliability is the foundation, as users feel frustrated when they cannot sign in or when the app stops working. Responsiveness is also vital, as users value fast loading times and immediate confirmation messages. The third theme, customer support, shows that service quality depends on how well the bank’s agents resolve problems. Lastly, security assurance builds trust when features like two-factor authentication are easy to use.