Crowdsourcing‐based new product development (NPD) involves consumers in contributing ideas as ideators in the ideation phase, and providing inputs as cocreators in self‐selected subtasks in the product development phase. The novelty of this NPD approach furnishes little understanding about the effect of crowds' self‐select participation into ideation and development subtasks, and this study aims to bridge this gap. This article draws on the attention allocation perspective of online knowledge sharing to identify the domains of ideator expertise based on the types of development subtasks they self‐selected to perform as cocreators in prior projects. It separately examines the impact of different domains of ideator expertise (marketing and engineering) and the interaction between ideators' expertise and cocreators' inputs on crowdsourcing‐based NPD outcomes. Large‐scale, longitudinal data from a crowdsourcing‐based NPD platform reveal that ideators' engineering expertise helps convert ideas into final products more than ideators' marketing expertise. In contrast, ideators' marketing expertise helps these products achieve more sales than ideators' engineering expertise. Moreover, final products achieve more sales if ideators' marketing expertise embedded in initial product ideas is later augmented with either marketing or engineering‐related development inputs by crowd cocreators. However, ideas generated by ideators with engineering expertise achieve fewer product sales when those ideas are complemented by more marketing‐related development inputs by cocreators. These findings extend crowdsourcing‐based NPD theory and furnish insight on managing crowdsourcing‐based NPD platforms.