|Media Mix Modeling for a Changing Media Landscape
| Joshua Lewis
Master's Candidate, Department of Statistics, Colorado State University
Monday, April 30, 2007
Marketers are always seeking to decide how best to spend their marketing budget. This budget covers marketing through mass marketing channels (television, radio, newspaper, magazine, billboards, etc) that reach a large audience all at once, as well as channels that are more direct to particular individuals (direct mail, email, outbound telemarketing, etc).
Media mix modeling aids marketers in selecting an optimal allocation of the different media channels. In particular, media mix modeling seeks to consider and evaluate the impact of all marketing exposures from all channels across all individuals, paving the way for an optimal allocation of budget by channel.
Media mix modeling as it is performed today will be reviewed in detail—considering different ways to incorporate time series and cross sectional analysis. Particular attention will be paid to how the statistical techniques relate to the aggregation of data across individuals, geographies, and time periods. This discussion will include empirical results from cross-sectional, time series, and hybrid models that combine the two. All models will be based on the same simulated data set.
Attention will then be shifted to the changes occurring to the media landscape today, such as the explosion of media channels and greater ability to measure and target individual-level media exposure. These media changes create new demands for media mix modeling, which will in turn lead to significant changes in how media mix modeling is performed.