Is your inventory putting you in a constraint jacket?

Let’s look at Inventory Optimization as a mathematical programming problem. You remember linear programming from a Quantitative Methods or Operations Class don’t you? No, we are not going to get deeply into mathematics here.

In mathematical programming, there is an objective function that you are trying to optimize subject to a set of constraints. In our case, we are trying to minimize inventory subject to a few key constraints:

• There should be enough inventory so that Customer Service is greater than or equal to some agreed upon value say 98% unit fill. Most companies set separate service levels for A, B, and C SKUs


• The inventory levels of SKUi at Warehousej must take into account
  
  
  
  • Transit times
  
  
  • Warehouse storage limitations
  
  
  
  


• Manufacturing Lead Times and Capacities for SKUi made at Factory


• Let us not forget that Demand is usually not static but varies. We have to take this into account.

We have scoped out the issues with sizing determining Optimized Inventory levels of just one SKU across a network of factories and warehouses. But, let us not forget an added level of complexity to our set of constraints. The assets that make SKUi mostly likely make several other SKUs in the same family so the utilization of assets must be considered over the breadth of the products produced on that asset.

Wow… this can become very complicated very quickly. There are many constraints at play that influence our Inventory Optimization efforts. The equations are so complicated, especially across a large number of SKUs, that many of us use sophisticated software to help keep track of constraints to get the job done. The software help we solicit may simply be a big time phased spreadsheet, a la MRP systems, to offera variety of complex algorithms for heuristic solutions. Rarely are we able to fully get a true optimization because of all the constraints and variability in demand.

This notion of constraints may have triggered another thought. That being the Theory of Constraints: the concept popularized by Dr. Eliyahu Goldratt in his popular 1984 book - The Goal. Our view of the Theory of Constraints is that there is a hierarchy of constraints that prevent us from optimizing inventory. There is often a dominant constraint that limits how low we can take the inventory and still provide acceptable service to customers.

To Optimize Inventory, we must know what this constraint is. Here are some real world show stopping examples:

1. Demand Variation is historically very high making for poor forecast accuracy. Did anyone say < 50%? This is often caused simply by having too many SKUs due to poor SKU management.


2. Month end or Quarter End peaking that could account for 40-50% of the period’s sales in 3-7 days. This is caused by years of training customers to wait until the end of the sales period to buy because they know we will be wheeling and dealing to make the numbers. This peaking is often beyond our production capability and thus requires the dreaded pre-building of inventory.


3. For real fun, combine constraints 1 and 2.

No matter how sophisticated your algorithm or software, these constraints simply dominate and force high inventory levels. These problems are policy and process. They are only solved by working with Sales, Marketing, and Finance to change policies and implementing the process improvements to change the business model. This is tough work requiring management team coordination and shared objectives to get the job done.

And you thought this was going to be easy.