Improving the acquisition of donors
Recruiting new donors is the most difficult task in fundraising. Over the years, database analysis has enabled DSC to define donor profiles. Based on these profiles, segments are defined and selected for new direct mail fundraising campaigns. However, DSC wanted to improve the success rate of its campaigns. They called upon SAS partner Python Predictions, a Brussels-based service provider specialized in predictive analytics in marketing, risk, and operations. Their approach consists of using all available data to predict future events and improve decision making.
Python Predictions helped DSC construct more accurate donor profiles. "We used the existing DSC database for advanced predictive analysis. Specifically, we analyzed donor behavior and its relationship to key objective socio-demographic factors. In this way, we could select a combination of parameters that are good predictors for donor behavior and create ideal donor profiles. We then looked for individuals with similar profiles within the Belgian population to identify new donors. The improved donor profiles made it also possible to develop campaigns that are better tailored to specific target groups," reports Geert Verstraeten, partner at Python Predictions.
Achieving higher response rates and donation amounts
DSC and Python Predictions tested the new donor profiles using two fundraising campaigns and achieved positive results in the acquisition of new donors.
During the first test campaign, direct mailings were sent out for Licht en Liefde, an organization that helps blind people in Belgium. The mailings were sent to both the old and new profiles, making a direct comparison possible.
"The new profile had a 37% higher response rate and a 32% increase in the amount of donations compared to the old donor profile. The revenue per letter sent increased by 82%. This was an excellent result and provides a sound basis for further fine-tuning of our primary campaign," says Ludo Longin, representative delegate at DSC. The second test campaign, performed for the Muco vereniging, proved even more successful. The response rate more than doubled and the donations increased by 11% compared to previous results using the old profiles.
Precisely selecting variables and testing methods
Python Predictions chose to compare different techniques in order to maximize the results. "Building a model includes the selection and combination of the best predictors, in such a way that they accurately indicate the donor potential in each geographical area. The model not only quantifies this potential, but also delivers valuable insight into the location and the detailed profile of the most valuable potential donors," explains Verstraeten.
Continuing with ready-to-use tool
DSC now has a ready-to-use SAS-based tool to select good potential donors without the help of Python Predictions. Ludo Longin notes, "The entire analysis and development of this tool took only three months. As a result, we can now swiftly create more effective new donor profiles, select target groups, implement campaigns, and generate significantly improved results. Python Predictions helped us improve our donor profiles, enabling us to select better potential donors and raise more money from more donors."