Market mix modeling and what-if simulator

Client

Leading US-based fashion retailer with online and brick & mortar presence

OPPORTUNITY

The client’s omni-channel strategy had made their conventional market mix modelling algorithms/strategies obsolete. The data came in from multiple sources and each source had a different format ranging from TRPs, GRPs, Impressions, Likes, Views, App Downloads etc.Moreover, there was no mechanism to attribute the returning revenue to all the touch points in the buying journey.

SOLUTION

SGA constituted a team of data experts and modeling specialists to deliver the following solution:

• By combining its in-depth understanding of various data sources along with the client’s business knowledge, the SGA team defined business rules to standardize the data and brought all data sources at the same level for primary EDA and ease in modeling
• To decide sales attribution, the SGA team had in-depth interactions with the client, and decided on a distributed linear attribution model
• The SGA team then built a model to explain the effect of channels on sales and quality of fit
• As a value-added service, the SGA team built a simulator for “what if analysis” and added a visualization layer over the model