This paper considers the question of how a sales manager should design the optimal
compensation scheme for his salesforce when it consists of salespersons of varying
selling skills, i.e., when the salesforce is heterogeneous. The manager's problem
is to reward the salespersons based on observable, uncertain sales achieved by the
salespersons. Under the assumption that both the manager and the salespersons are
risk neutral, the optimal compensation scheme is derived. It consists of the manager
offering a menu of plans, consisting of a quota, a payment for meeting quota, and
a constant commission rate for sales above or below quota. Such schemes using constant
commission rates are also called menus of linear plans. Salespersons choose the quota
which best suits them, achieve sales, and are then rewarded based on their actual
performance. This scheme, variants of which are often observed in practice, is shown
to be optimal for sales environments characterized by commonly encountered sales
response functions, and a large class of frequency distributions of selling skills
in the salesforce. The problem is solved using the methods of principal-agent models.
The key differences in managing homogeneous and heterogeneous salesforces are highlighted.
Finally, the paper discusses the issues involved in practically implementing the
optimal compensation scheme.