Recent intelligent smartphone assistants can provide proactive activity recommendations for what users might want to do in the upcoming moments. However, many of these recommendations may not be accepted because the ongoing tasks performed by users vary in their "stickiness" (reluctance to switch away). Past studies suggest that user states in their ongoing tasks might reflect task stickiness. Still, it has not been systematically tested among a comprehensive set of tasks for everyday smartphone users. In this study, we sought to examine the impacts of individual and task characteristics on users’ task stickiness in the context of mobile recommendations based on a large set of tasks. In total, 220 participants completed an online evaluation assessing 55 everyday smartphone activities summarized from literature and interviews. They evaluated each task’s perceived task load, involvement, and stickiness. Individual characteristics such as boredom proneness were also collected. Hierarchical linear modeling showed that: (1) Task load and involvement positively predicted task stickiness. Users may not accept new recommendations when their ongoing tasks are demanding and engaging. (2) Boredom proneness reduced the effect of task load. For people who are more likely to feel bored, their intention to stay in the current task was maintained even when the task load dropped. This study suggests that recommendation system designers should consider both task characteristics and individual differences to enhance contextual appropriateness and personalization.