Examining Solver Performance in Crowdsourcing Contests: Does Ambidexterity Matter?

The performance of solvers is crucial to the success of crowdsourcing contest platforms. Sustained solver performance entails a combination of exploration and exploitation activities, i.e., solver ambidexterity. However, it can be arduous for solvers to engage in ambidexterity with limited knowledge of what its optimal levels are and little research informing this topic. Thus, this study examines the relationship between solver ambidexterity and performance, which is stated to be positive for workers in organizational research. We challenge this assumption and propose that the costs associated with ambidexterity will limit its efficacy beyond a certain level—i.e., we hypothesize an inverted U-shaped relationship between ambidexterity and solver performance. Moreover, how contest conditions shape this relationship is unclear. Drawing on the bounded rationality model, we hypothesize three moderators of the relationship, i.e., task reward, task diversity, and in-process feedback. We tested our model using a panel dataset of solvers from a major crowdsourcing contest platform. Our results support the inverted U-shaped relationship between solvers’ ambidexterity and performance. We find that the highest-performing solver cluster showed a ratio of 5.33 exploitation activities to 1 exploration activity, contradicting the prior premise that both activities are required to a similar extent. Additionally, task diversity and task reward are found to steepen the inverted U curve, while in-process feedback flattens the curve. Our study contributes to theoretical knowledge of the relationship between solvers’ ambidexterity and performance and its contingent conditions. The results also offer novel insights for solvers and platforms to manage ambidexterity and the trade-offs between exploration and exploitation.