We examine when and how task division improves performance for inventory planning. Specifically, we consider a decentralized inventory management context with two interdependent subtasks: preparing a forecast and setting a service level. Using a behavioral experiment, we reveal that the task challenge moderates the relationship between task division and performance. Our findings indicate that task division improves performance when the task becomes more challenging, such as under high demand uncertainty. It facilitates counteracting behavior, where individuals adjust their decisions to counterbalance their partner's errors, leading to more stabilized final decisions. We identify this counteracting behavior as a critical mechanism driving the benefits of task division, mainly when subtasks are interdependent. We demonstrate the robustness of our findings by examining an egalitarian system, where decision-making authority is shared among team members, and a hierarchical system, where decision control resides entirely with one team member.