Abstract
In spite of the fact that nearly all advertising campaigns target specific target markets, no models exist for estimating the exposure of a media schedule to a target market, given commonly available data. We propose such a model, based on an approximate log-linear model approach. We tested the model against available models for estimating pairwise duplications within target markets, for television self-pairs and cross-pairs, and for magazine cross-pairs. It performed at least as well as the existing duplication models. However, unlike the other target market duplication models, our target market exposure model may be extended naturally to estimate entire exposure distributions. We obtained good predictive performance on both television and magazine media schedules. A less computationally intensive variant of our model, based on a canonical expansion approach, was found to perform almost as well.