Uplift modeling is often associated with political campaigns, advertising and healthcare. In any group of voters, potential customers or patients, for example, some individuals will be "on the fence" about a future course of action but can be influenced by talking to a candidate, stylist or physician's assistant. The goal of uplift modeling is to only expend time, money or effort on individuals that really need to be messaged before they will take a particular course of action.
Uplift modeling requires data scientists to build data models that will accurately identify the characteristics of individuals who require personal contact before making a decision. Building the necessary control and test groups can take time, and refining the predictive models can take a lot of patience. Because the criteria for identifying such individuals can be so nuanced, testing is often done incrementally. For this reason, uplift modeling is often called incremental modeling.