Background: In today's competitive world, with constant changes in the environment and the inefficiency of traditional corporate systems, centralized management strategies are increasingly being used to improve health and deal with service failure. Service recovery can provide a great opportunity to improve service in addition to regaining the satisfaction and loyalty of dissatisfied customers.
Aim: The main objective of this study is to analyze satisfaction with the performance of an online service retrieval system using machine learning.
Methodology: In this study, first, by systematically reviewing the research literature and interviewing experts, a set of performance indicators of the online service retrieval system were identified. Then, in order to select key performance indicators and prioritize them, data were collected using a questionnaire tool and with the participation of 400 customers of online stores in Shiraz. Then, by comparing the efficiency of machine learning algorithms, the Random Forest (RF) algorithm was used to analyze the data.
Findings: The findings of the study show that in this study, using a systematic literature review, 40 indicators related to the performance of the online service recovery system were identified, which were classified into three levels of error detection, error analysis, and error response using the Failure Mode Analysis (FMEA) framework. Integrated analysis of performance indicators shows that the index "customer satisfaction with the organization's commitments to resolve the problem" has the best performance among other indicators, followed by the index "the rate of use of new equipment and technologies in error analysis" in the next ranks.
Conclusions: In this study, using a systematic literature review and machine learning approach, the performance indicators of the online service recovery system were classified into three levels using the Failure Mode Analysis (FMEA) framework. This study also provides a practical