A process-oriented method for evaluating e-service quality using SERVQUAL and interpretable machine learning

Purpose This study develops a process-oriented framework to evaluate e-service quality (E-SQ) from an end-to-end service process perspective. E-SQ is conceptualized as an external manifestation of service process performance. The proposed approach aims to support E-SQ assessment and service process governance for e-service platforms. Design/methodology/approach A two-stage methodology is proposed. First, SERVQUAL dimensions are transformed from the perspective of service provider into observable service process elements representing key stages such as consultation, delivery and assurance. Second, a machine learning-based nonlinear model is developed to predict E-SQ, complemented by interpretability techniques to identify critical process drivers and interaction effects. The framework is empirically validated using operational data from a real e-service platform. Findings The results show that the proposed model outperforms several traditional methods for E-SQ prediction. Interpretability analysis shows that the delivery quality is the most stable and influential process factor, while the consulting responsiveness reflects an efficiency-depth trade-off. Interaction analysis further reveals synergistic effects between delivery quality and other elements, confirming the importance of cross-process coordination for end-to-end service performance. Originality/value This study extends SERVQUAL from the perspective of service process management and provides a data-driven, interpretable approach to E-SQ prediction and process improvement. The findings provide actionable insights for continuous service process governance on e-service platforms.