The effects of malfunctioning personalized services on users' trust and behaviors

Abstract

Online merchants adopt web personalization to customize web content to match online users' needs. Prior research has only looked at the "success" side of web personalization. Little research examines the "problematic" side of web personalization. The objective of this research is to explore how "malfunctioning" personalized web services influence an online user's trust in the personalization agent and the behavioral intention of that user. In particular, this research looks at two types of malfunctioning personalization: irrelevant recommendations and biased recommendations. We draw on trust theories to develop seven hypotheses to predict the effects of malfunctioning personalized web services. We conducted a study with a personalized music download website. We found that irrelevant recommendations led to low trust in the personalization agent's competence and integrity, and biased recommendations led to low trust in the integrity of the personalization agent. These findings provide empirical evidence of the possible problems of malfunctioning personalization and help firms understand and quantify the challenges and limitations of incorporating web personalization in their websites.