Which Facebook splits should be used in MMM: Abstract
For social media planners, Facebook offers a multitude of buying options, each influencing Facebook’s auction system, campaigns’ delivery and efficiency. The role of econometricians working on Marketing Mix Modeling (MMM) is to identify the most effective and efficient ones and deliver recommendations helping to increase the ROI of media.
However, the sheer number of splits available in Facebook MMM UI data presents a challenge to econometricians. It is not possible to include all of them in the models. There arises the need to first identify and then use only the most influential of them so that the recommendations based on MMM projects are most useful to the planning teams. The identification of the most important splits (both statistically and impact wise) is the subject of this paper.
In the analysis, five main split criteria for Facebook variables, which can be used in
Marketing Mix Modeling, were investigated:
- Campaign Objective
- Device
- Format
- Placement
- Platform
Campaign Objective and Platform seem to be the most influential split criteria. Of all split criteria, Campaign Objective had the highest impact on the statistical quality measures of the models, while Platform importantly changed the contribution to sales, which is a critical business outcome of MMM. Splitting the variables by Format or Device brought less improvement, and Placement could only be used in very specific cases since it has only slight impact on the analyzed measures.
