The BB/NBD + 1 Model Applied to Rotation Policy

An early dismissal policy for unproductive recruiters is proposed. The policy is based on a bivariate stochastic model for productivity; this model considers both incidence of reporting a positive quantity of sales and the quantity per report. All recruiters are observed for a probational period; those who exceed minimum incidence and quantity requirements are allowed to continue to the end of a (fixed) maximum tenure while others are replaced with new recruiters after the probational period. We show that those who report less often must report a larger total quantity over the probational period in order to compensate for a larger chance variation. The univariate negative binomial distribution (NBD) model which considers quantity only is examined with respect to its robustness as an approximation in this context. It is found that the NBD leads to serious errors when the quantity reported is relatively homogeneous and reporting incidence is bimodal. Extensions to more conventional salesforces and to direct mail applications are indicated.