Integrating customer actions into aspect-based service quality evaluation: A text mining framework

Although evaluating the service quality is crucial, conventional methods often fail to produce effective strategies. Existing text mining research typically analyzes in isolation critical service aspects (the ‘what’), or associated customer actions (the ‘how’). Consequently, they failed to connect the identified problems to specific prioritized solutions. Therefore, this study developed and validated a novel action-oriented methodology for deriving prioritized improvement strategies. The methodology consists of three stages: (1) using transformer-based topic modeling to extract critical service aspects, (2) estimating the importance of each aspect for customer satisfaction through an interpretable AI model, and (3) calculating a priority score for each aspect-action pair to create a data-driven ranking of improvement targets. An analysis of 231,705 online reviews of Roblox metaverse services demonstrated the effectiveness of the framework. This approach empowers managers to move beyond generic feedback and strategically allocate resources to the most critical aspect-action pairs, ensuring that service enhancements are more tangible and impactful.