Critics worry that algorithmic filtering could lead to overly polished, homogeneous web experiences. “Serendipity,” in turn, has been touted as an antidote. Yet, the desirability of serendipity could vary by context, as users may be more or less receptive depending on the services they employ. We propose a nomological model of online serendipity experiences, conceptualizing both cognitive and behavioral antecedents. Based on a survey of 1,173 German Internet users, we conduct structural equation modeling and multigroup analyses to differentiate the antecedents and effects of serendipity across three distinct contexts: online shopping, information services, and social networking sites. Our findings confirm that antecedents and outcomes of digital serendipity vary by context, with serendipity only being associated with user satisfaction in the context of social network sites.