This paper reports on the cross-disciplinary research that resulted in a decision-support tool, Team Machine (TM), which was designed to create maximally diverse student teams. TM was used at a large United States university between 2004 and 2012, and resulted in significant improvement in the performance of student teams, superior overall balance of the teams as well as overwhelmingly favorable reactions from stakeholders. An empirical study is conducted comparing the performance of teams created by TM compared to teams manually allocated by a subject matter expert. The findings show that optimally balanced teams perform better than those created manually. TM serves as a broad-based example of how to integrate business analytics into interactive search by conflating human judgment with algorithmic efficiency in the context of team formation. The contribution of this research to the academic community is a model and solution method which can be easily implemented to solve a very important and recurring issue faced by many MBA programs and empirical evidence of its value. The outcomes of this study include empirical evidence of increased team performance, significant administrative time savings, improvement to team diversity, and increased satisfaction from students and administrators with the team formation process.