Predicting mobile trading system discontinuance: The role of attention

• Published in: Electronic commerce research and applications 2020-11, Vol.44, p.101008, Article 101008
• Main Authors: Kim, Dongyeon, Park, Kyuhong, Lee, Dong-Joo, Ahn, Yongkil
• Format: Article
• Language: English
• Subjects: Attention; Discontinuance; Field study; Machine learning; Mobile trading system
• ISSN: 1567-4223; 1873-7846
• DOI: 10.1016/j.elerap.2020.101008

•Mobile devices have become people’s first go-to informational source.•This study explores whether attention predicts mobile system discontinuance.•Attention has significant statistical power over transaction-related metrics.•XGBoost consistently outperforms benchmarks in the empirical literature. As mobile devices have become people’s first go-to informational source, they are becoming critical for e-commerce companies in understanding how mobile trading devices influence their businesses. This study involves a collaboration with a nationwide financial services company in Korea to examine the role of mobile attention in predicting mobile stock trading system discontinuance. Employing XGBoost and an artificial neural network, we analyze the complete transaction history, as well as the usage and login patterns data from 2017 to 2018 for 25,822 mobile trading application users. We find that mobile attention has significant statistical power over traditional trade-related metrics such as recency, frequency, and monetary value (RFM) in predicting subsequent mobile trading system discontinuance. Moreover, the new prediction methodology, augmented by incorporating mobile attention into the RFM framework and utilizing up-to-date machine learning techniques, consistently outperforms benchmarks in the empirical literature. Thus, this study sheds new light on the post-adoption information system usage literature and furnishes practical guidance to those companies whose business hinges on mobile systems.