Budapest BI 2015

To maximize the return-of-investment (ROI) for a direct marketing campaign it is critical to contact the right customers for a campaign. One common approach is using a response model which aims to maximize the response rate but not necessarily the ROI of such campaigns. However, the recently developed technique of “uplift modeling” promises to provide much higher detail. Uplift models are designed to maximize the ROI of a campaign instead of just measuring the response rate. By comparing the treated group with the control group, uplift models predict the influence of the marketing campaign and not the customer behavior like in response models. The ROI can be increased by contacting only customers which respond thanks to the marketing campaign and leaving customers untouched who would have responded without contacting them. Churn analysis also strongly benefits from uplift models, as they identify the customer segment which might respond negatively when contacted. The presentation shows different types of uplift models within the R programming environment. Furthermore, it compares response models with uplift models in a case study.