Predictive models for annual fundraising and major gift fundraising

For most nonprofit organizations, the selection process for determining the best individual prospects for capital campaigns or annual fund drives are ad hoc and intuitive. Our research provides two statistical models developed from the alumni database at Northwestern University for both major gifts and annual fund prospects. One model predicts which individuals will give $100,000 or more over three years, the other, $1,000 or more. Our work provides a means of more efficiently selecting fundraising prospects. In our analysis, we combine geo-demographic data with the internal data typically found in alumni databases. A metric is presented to test the viability of the models when compared to single-criterion models. We show that past giving is the strongest single factor in predicting future giving. However, full models provide superior overall results.