Determining the optimal amount of mailings being sent to customers is crucial and requires the specification of relevant mailing variables. Therefore we distinguish different mailing types and sizes as contribution over the extant literature. from a modeling viewpoint we use a mixture of Dirichlet processes to deal with unobserved heterogeneity and policy functions to treat endogeneity. For a data set from a mail order company we demonstrate the occurrence of (super-) saturation effects. Finally, we investigate whether and how consideration of endogeneity leads to different managerial implications. To this end we apply dynamic optimization to determine optimal mailing strategies.