On dynamic selection of households for direct marketing based on Markov chain models with memory

A simple, dynamic selection procedure is proposed, based on conditional, expected profits using Markov chain models with memory. The method is easy to apply, only frequencies and mean values have to be calculated or estimated. The method is empirically illustrated using a data set from a charitable foundation. The results reveal some interesting features with respect to the time-dependent behavior of certain subsets of households, whereas the profitability increases by about 9% by using the method compared to a benchmark of sending a mailing to all households. Copyright Springer Science+Business Media, LLC 2007