A Success Story
End-to-End Data Science Solution for Optimizing Product Mix for a Major Multinational Food Manufacturer
Bardess developed and implemented a data science initiative aimed at using machine learning to optimize product mix at stores, resulting in significantly increased sales revenue and reduced costs.
Knowing that they could be extracting more value from their data, the client came to Bardess, already a trusted partner, for ideas. Working collaboratively to review their data and business pain points, Bardess discovered that a continuing challenge was how, given limited shelf space, product mix could be optimized across thousands of stores and product SKUs, so as to maximize revenue and minimize returns.
THE BARDESS SOLUTION
Bardess designed and implemented an end-to-end ML solution for optimizing product distribution mix across all of the client’s stores and products. Firstly, a Python ML model was built to predict revenue for thousands of product SKUs and thousands of stores, leveraging Snowflake for data processing. Secondly, an optimization algorithm was built which took these predictions as input and calculated the highest-revenue product mix for each store. Finally, a BI application was developed for quickly visualizing and understanding the results.
VALUE & BENEFITS
This data-driven analytics solution is able to deliver product mixes that are far more optimized than those that were provided by the previous manual approach. This helps the client to maximize revenue from each store and remain competitive, while minimizing costs incurred by returns.