Inquiry Response: Tips for Building Marketing Mixed Models
After using an outside vendor to build our marketing mixed models, we’re going in-house to leverage our particular business expertise to improve the models. What should we be thinking about given that we use a Bayesian hierarchical time series model and we want to understand the impact of marketing spend at our stores nationwide?
THE BIG IDEAS:
- Try vector autoregression
- Avoid impression metrics
- Consider creative quality
TRY VECTOR AUTOREGRESSION
The Bayesian hierarchical approach makes sense in terms of