I have been working with my company's Marketing Mix Modeling solution for years now. We are looking to update our modeling methodology and are looking at different options. What are the current 'up-to-date' methods being used for MMM? For instance, when we built our current MMM solution, Bayesian methods were not really being used, while now it seems that they are very important in the space.

What are the various 'state of the art' model options for MMM in 2022?

  • 1
    $\begingroup$ I have never seen an MMM that has survived an experiment, in the sense of being able to recover the experimental estimate using non-experimental data (i.e., data from before the test, or data only from the treatment group during the test). Even a single channel seems impossible to do well: Gordon, Brett R. and Zettelmeyer, Florian and Bhargava, Neha and Chapsky, Dan, A Comparison of Approaches to Advertising Measurement: Evidence from Big Field Experiments at Facebook (September 23, 2018). Forthcoming at Marketing Science, dx.doi.org/10.2139/ssrn.3033144. $\endgroup$
    – dimitriy
    Jan 31 at 19:54
  • $\begingroup$ But I am curious if that is a widely shared experience, so keeping an eye on the responses. $\endgroup$
    – dimitriy
    Jan 31 at 19:57


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Browse other questions tagged or ask your own question.