81 – Goals, Surprise, and Returns
Time-Dependent Cox Model for Revenue Attribution
Fan Yang
School of Statistics, University of Minnesota
In marketing analytics, revenue attribution is an important aspect. Most retailers have multiple channels for both ordering and marketing, for example, email order and catalog order for order channel; promotion email and paid search for marketing channel and so on. Retailers want to correctly identify the sources of orders that come in back to the various marketing channels that drove them, in order to realize a more efficient way of marketing. This paper proposes a time-dependent Cox model built to analyze the relationship between the number of contacts made through a diversity of marketing channels and the number of purchases via many order channels. The method is applied to data from a retailer company. Weights of each marketing channel which measure how important a marketing channel is, are given finally for different order channels using the model. Numerical results show that the proposed method is very useful for revenue attribution and provides a good way to estimate weights for marketing channels.