Task Characteristics and Incentives in Collaborative Problem Solving: Evidence from Three Field Experiments

We study, using three sequential field experiments, collaborative problem solving in knowledge work enabled by information technology within the context of the customer support function in a leading high-technology firm. Experiment one examines the performance change after introducing a new collaborative problem-solving process, specifically whether the use of a team of experts across departments to solve problems can help reduce problem-solving costs. In addition to the extant process of supporting customers using problem solvers within a specific department, the experiment allowed two forms of engaging problem solvers outside the department: (1) formal handover (transferring the task to experts in an external department) and (2) using a new, collaborative process in which experts across two departments jointly work on the task. Interestingly, we find that the cost reduction occurs not because the collaborative process is always superior to formal handover, but because there is a shift of intradepartmental customer support work toward the new collaborative process. Building upon the findings of experiment one, experiment two aims to identify the conditions under which the new collaborative process works or fails. We discover that task features, such as novelty and time constraints, play a significant role in determining the appropriate mode of engaging an external department for problem solving. These findings are then utilized to develop an information system that provides recommendations on how to seek help through either formal handover or collaboration. In experiment three, we examine how users react to the recommendation. We find that local (department level) incentives can cause problem solvers to deviate from machine recommendations. We analyze the underlying reasons for this deviation and demonstrate how global (firm level) incentives can be aligned with local incentives to increase compliance with machine recommendations. The findings of this study offer practical implications for firms that aim to develop and implement information systems to support knowledge-intensive problem-solving tasks.