The problem

In the retail industry, the distribution centers are the key components of modern-day logistics that bridge the gap between supplies and customers. The supply chain management team of one of the leading retailers in the United States wanted to explore cost-effective solutions to optimize their distribution network to improve product flow to customers, keeping the cost minimized. With the current allocation strategy deemed expensive and suboptimal, the customer wanted to see if there are opportunities within the existing supply chain network without altering the network itself.

With a constant focus on customer service and satisfaction, the retailer asked Cognitive BI to build a cost-effective and efficient solution to optimize their existing customer warehouse alignment and provide better quality of service to their valued customers.

The cognitive solution

Removing the spreadsheet-oriented design currently at play, an optimization engine was proposed to enhance the current allocations and a mechanism to periodically refresh as and when newer nodes/routes are added into their portfolio.

1. Defining the objective function

• Minimize the cost to serve each customer as a function of distance and transportation costs
• Customer profiling to move certain customers like independent restauranteurs who are easier to move vs national/across border chains that are the most difficult
• Adding the constraints of warehousing capacity and various storage types

2. Business constraints and heuristics

• Don’t move newly acquired customers and long-term relationships
• Prioritizing movement of customers within the same area

2. Nature inspired systems

• Ant colony optimization technique used so that both constraints and objective functions can be leveraged easily.
• Genetic algorithms used as validation as part of periodic refreshes

Key insights

The client has successfully deployed the periodically refreshing engine as a central service allowing them to benefit:

• Customer-warehouse mapping with a reduction in overlaps by over 35% and reduced cost by 6% overall
• The change resulted in zero customer churn in the 6 months post deployment.