A novel problem for a vintage technique: Using mixed-integer programming to match wineries and distributors

Wineries must find distributors to represent their wines, and distributors seek interesting wines to sell to clients. Wineries and distributors often form partnerships at trade shows. I created a system to qualify potential partnerships for one such trade show, the World Wine Market, by formulating a mixed-integer program that is an embellished transportation model. I based its parameters on answers to a questionnaire I obtained from a subset of trade-show attendees who were seeking partners. Of 675 possible matches between wineries and distributors, the program recommended 31 as the optimal set of matches. This resulted in an allocation of just over 50,000 cases of wine. Participants provided positive feedback. In addition, the program introduced at least one winery and a corresponding distributor who have formed a profitable, long-term partnership.