Collaborative Research: Optimizing Direct-Marketing Strategies in Non-Profit Fundraising: An Integrated Framework for Segmentation, Estimation and Control

这一合作奖项的目的是通过将一类复杂的优化工具（随机最优控制）与最先进的估计方法（有限混合模型）联系起来，提供直接营销策略及其对非营利组织绩效（创收、成员保留）的影响之间的联系。这种整合使提议的框架可以作为一种实用工具，根据捐款顺序将校友/捐助者群体划分为同类群体，从而可以针对这些群体采取具体的营销行动。该框架处理捐款序列（即纵向数据）的能力，为直接营销策略如何影响捐助者捐款行为提供了基本的新见解，并为旨在模型校准和验证的数据获取策略（通过调查和实地实验）提供了信息。如果成功，这项研究的结果将提供一种严格的机制，根据未观察到的异质性来推断和划分人口，例如，影响个人捐款倾向的人格特征。虽然有充分的经验证据表明这种差异在统计上是显著的，但现有的支持直接营销活动的模型只能将这些差异解释为人口中的随机变化。相比之下，该方法为量身定制直接营销政策以优化具体绩效标准提供了基础。此外，研究将开发方法工具，可以指导决策的广泛的分割，估计和控制问题在其他领域。例如，在运输系统的管理中，问题是将具有相似特征的设施分段，并为每一组找到最佳的资源分配政策。

The objective of this collaborative award is to provide insights into links between direct-marketing strategies and their impact on the performance (revenue generation, member retention) of nonprofit organizations, through the linkage of a general class of sophisticated optimization tools (stochastic optimal control) with state-of-the-art estimation methods (finite mixture models). This integration allows the proposed framework to serve as a practical tool to segment alumni/donor populations into homogeneous groups, based on their contribution sequences, which can be targeted with specific marketing actions. The framework's capability to process contribution sequences, i.e., longitudinal data, provides fundamental new insights into how direct-marketing strategies affect donor contribution behavior, and informs data acquisition strategies (via surveys and field experiments) aimed at model calibration and validation.If successful, the results of this research will provide a rigorous mechanism to infer and segment the population based on unobserved heterogeneities, e.g., personality traits that influence the propensity of an individual to donate in response to solicitations. While there is ample empirical evidence that such differences can be statistically significant, existing models to support direct marketing activities are only capable of explaining these differences as random variations within the population. In contrast, the methodology provides a basis to tailor direct-marketing policies to optimize specific performance criteria. Moreover, the research will develop methodological tools that can guide decision-making for a broad set of segmentation, estimation and control problems in other areas. For example, in the management of transportation systems, the problem is to segment facilities that share similar characteristics, and to find optimal resource allocation policies for each of the groups.