The client had been testing their online grocery channel to offer better online and delivery service. They wanted to expand this service into more stores and newer markets. The client was looking for a solution that could help them identify potential stores in the new market, where they could extend their online grocery operations for pickup.
The analytics experts at Tredence proposed to solve this problem in a two-step process.
Mixed effects modelling technique was implemented to identify key performance drivers. Various factors around store attributes, market characteristics, customer properties, pre and post purchase experience were considered. These drivers were then used as input in to the Generalized Linear Model (GLM) technique to identify potential stores
Machine Learning algorithms were used to overcome the model instability caused due to lower data points.
The algorithms were tested and validated recursively on random samples created from the small pool of data. The final algorithm was used to score the potential stores in different markets. Sensitivity analysis was performed to account for individual factors.
Achieve an incremental revenue of $60MM by launching Online Grocery pickup service in predicted potential stores
Identify drivers affecting online grocery pickup orders such as market share, online grocery awareness, etc.
Improve accuracy over multiple iterations using various Machine Learning techniques
