Successful philanthropic development relies on finding a few major donors in the sea of prospects. In healthcare, grateful patient giving has become a key component of non-profit health system funding. Patients or their relatives may donate in response to the quality of care provided. Development officers rely on imperfect, and often inaccurate, estimates of prospect giving capacity to select the most likely candidates. Since the estimates are known to be error prone, development officers must engage in lengthy secondary research on potential prospects using many additional data sources. Although these data present opportunities to identify prospects more efficiently than using only giving capacity estimates and manual research, they present significant challenges to modeling.
To address these challenges Elder Research started with a Data Discovery project to assess whether the data was sufficient to support predictive modeling. We created a model estimating the likelihood a patient will begin a philanthropic relationship with the health foundation. The figure below shows the relationship between a patient’s estimated capacity and predicted likelihood of giving.
Key to our solution was working collaboratively with development officers to understand their workflows, build credibility for analytics, and communicate model results effectively.
Compared to the baseline process of looking first at patients with the highest giving capacity, our model found 20-30% more patients with a high likelihood of beginning a relationship with the foundation. If a development officer looks at the same number of patients each week, our model prioritizes patients for review by identifying 20- 30% more patients with a high likelihood of becoming a grateful patient donor.