This document discusses the use of predictive modeling, specifically uplift modeling, in political campaigns. It provides examples of how the Obama and Clinton campaigns used predictive analytics in 2012 and 2016 to determine which voters to target for persuasion through volunteer interactions. The campaigns conducted special polls coordinated with applying and not applying the marketing treatment to collect data needed for uplift modeling. Uplift modeling identifies segments of voters that are more likely to be positively influenced by campaign contact compared to those not contacted.