Over six months, the Data Driven Supply Chain team helped a national restaurant chain understand how characteristics of its distribution network could be used to predict Distribution Center (DC) service levels, and how to navigate inventory tradeoffs balancing cost, service level, and space.
Background
It’s a question faced daily in supply chain management: “How much inventory should I hold in my Distribution Centers – navigating tradeoffs between cost, service level, and space?”
No business wants to unnecessarily tie up cash in inventory. Carrying too much inventory, particularly of slow selling “C” SKUs, drives up warehousing and inventory costs and consumes precious space at capacity-constrained DCs. Additionally, SKUs held in inventory face shrink, obsolescence, and, if perishable, spoilage risk.
At least a little safety stock is necessary to buffer against variability in both supply and demand. Still, adding ever-more safety stock isn’t the answer to improving service level. Besides the cost and space consideration, more safety stock brings diminishing returns: every additional unit of safety stock will improve service level less than the unit before it. Therefore, an inventory tradeoff must be navigated between cost, service level, and space.
DDSC was engaged by a national restaurant chain to help navigate these tradeoffs.
The company maintains DCs across the US and stocks inventory there to service its restaurants. For most SKUs, the restaurant chain maintained a 95% service level goal at a restaurant/SKU basis. Other items, including core ingredients and consumables like paper products, were assigned a service level goal of 99%+ at a restaurant/SKU basis.
DDSC was asked to identify optimal DC/SKU inventory policies that balanced supply chain cost, service level at a restaurant/SKU basis, and DC utilized space: the Goldilocks inventory policy.
Considerations
At first, the restaurant chain was interested in learning how different supply chain characteristics impacted actual service levels at the restaurants, which were consistently hovering below goal.
Supply chain characteristics included:
- Product category
- Product temperature channel (ambient / refrigerated / frozen)
- Restaurant distance from its servicing Distribution Center
- Product flow characteristics (e.g., crossdock at DC vs. stock at DC)
The Data Driven Supply Chain Approach
The Data Driven Supply Chain team worked with SMEs across the company to collect data and build a statistical model, quantifying the relationship between service level and supply chain characteristics like those listed above. This model identified several key drivers that impacted service level. Some of those key drivers could not be modified (e.g., product temperature channel), whereas others could be modified (e.g., distance to the servicing DC, modified by opening more DCs to get the DCs closer to restaurants).
The team’s investigations revealed that the proximity of DCs helped, but only to a certain extent. Once a DC was within a certain distance from its restaurants, any additional improvement in fill rates achieved by locating closer to the stores was minimal. This analysis provided valuable information the client could use when considering its future DC network buildout.
After this analysis, DDSC was asked to recommend optimal inventory policies for each DC/SKU combination: policies that best navigated the tradeoffs between supply chain cost, service level at a DC/SKU basis, and DC utilized space.
Using open-source data science software, the Data Driven Supply Chain team developed a unique simulation-based approach to strategic inventory decisions. This simulation quickly evaluated dozens of inventory policy options for each DC/SKU combination and calculated KPIs associated with each inventory policy: total supply chain cost, DC/SKU service level, and DC space required.
Graphically displaying these KPIs allowed the team to visually understand inventory tradeoffs; have discussions about the relative importance of cost, service level, and space to the organization; and ultimately settle on an optimal inventory policy for each DC/SKU combination.
While not addressed in DDSC’s engagement, this approach could also be extended to help organizations navigate inventory tradeoffs within the four walls of a capacity-constrained DC.
Outcome
The Data Driven Supply Chain team delivered a Proof of Concept of the inventory simulation method using several of the restaurant chain’s distribution centers and a strategic subset of SKUs.
Showing the value of this approach, DDSC provided both technical and non-technical training to internal employees on expanding this capability to all their SKUs and DCs.
When fully implemented, this approach will help revolutionize the restaurant chain’s approach to inventory management.