Customers’ loyalty model in the design of e-commerce recommender systems

Abstract
Recommender systems have been adopted in most modern online platforms to guide users in finding more suitable items that match their interests. Previous studies showed that recommender systems impact the buying behavior of e-commerce customers. However, service providers are more concerned about the continuing behavior of their customers, specifically customers’ loyalty, which is an important factor to increase service providers’ share of wallet. Therefore, this study aimed to investigate the customers’ loyalty factors in online shopping towards e-commerce recommender systems. To address the research objectives, a new research model was proposed based on the Cognition-Affect-Behavior model. To validate the research model, a quantitative methodology was utilized to gather the relevant data. Using a survey method, a total of 310 responses were gathered to examine the impacts of the identified factors on customers’ loyalty towards Amazon’s recommender system. Data was analysed using Partial Least Square Structural Equation Modelling. The results of the analysis indicated that Usability (P=0.467, t=5.139, p<0.001), Service Interaction (P=0.304, t=4.42, p<0.001), Website Quality (P=0.625, t=15.304, p<0.001), Accuracy (P=0.397, t=6.144, p<0.001), Novelty (P=0.289, t=4.406, p<0.001), Diversity (P=0.142, t=2.503, p<0.001), Recommendation Quality (P=0.423, t=7.719, p<0.001), Explanation (P=0.629, t=15.408, p<0.001), Transparency (P=0.279, t=5.859, p<0.001), Satisfaction (P=0.152, t=3.045, p<0.001) and Trust (P=0.706, t=14.14, p<0.001) have significant impacts on customers’ loyalty towards the recommender systems in online shopping. Information quality, however, did not affect the quality of the website that hosted the recommender system. The findings demonstrated that accuracy-oriented measures were insufficient in understanding customer behavior, and other quality factors, such as diversity, novelty, and transparency could improve customers’ loyalty towards recommender systems. The outcomes of the study indicated the significant impact of the website quality on customers’ loyalty. The developed model would be practical in helping the service providers in understanding the impacts of the identified factors in the proposed customers’ loyalty model. The outcomes of the study could also be used in the design of recommender systems and the deployed algorithm.
Description
Thesis (PhD. (Computer Science))
Keywords
Recommender systems (Information filtering), Electronic commerce—Management—Data processing, Customer loyalty
Citation