In this paper, we look at the donation behavior of donors who periodically contribute a preset amount to a particular charity. The charity firm makes extensive use of direct mail to nurture these donors, and in fact sends two types of mails, one that seeks to retain the donors (retention-seeking mails) and the other requesting them to upgrade their donation amount (upgrade-seeking mails).We study the different effects of the two types of direct mail on the donation behavior. To these emails, a donor has to respond by making two decisions: should I continue donating or not? and should I upgrade or not? We propose a multi-response model that accommodates not only the differential effects of the two types of mail on the donors, but also the possible correlation across the two response functions for each donor. We fit our model to a very large data set to explore whether our model can extract the unique impact of demographics and those of the two types of mail on the two aspects of donation behavior. What makes it more interesting is that the charity firm deliberately uses the observed donation behavior of people to design the two types of mail. In our model and estimation, we explicitly account for this endogenous effect to ensure that we measure the unbiased impact of the two types of mail on the two responses of the donors. Our results show that once we account for the endogenous effect and separately model the effect of different types of mail on each type of donor response, we get a much clearer picture of the true effects of these mails, as opposed to a simpler model that does not incorporate these effects. Firms are advised to not only carefully analyze the donation behavior of the donors but also carry out simulation exercises to understand the effects of mails in totality before taking action.