Inventory Optimization

Coupa provides a robust method to perform Demand Analysis based on a set of statistics in multi-echelon models. Each echelon is a group of sites that perform the same or similar functions. During this process, it aggregates demand, performs analysis of any outlying demand and generates a set of demand statistics.

Using these statistics and user-definable threshold values, Demand Analysis is able to segment the demand into a set of classifications such as smooth, erratic, slow and lumpy. Demand classification distinguishes between the different types of demand for each product at each location in a multi-echelon model. Demand classification information is provided in demand profile tables.

Once the demand analysis is complete, Safety Stock Optimization calculates the required safety stock taking into account demand and lead-time variability:

• Demand variability – This is variability in the amount of product that a customer wants. It is the difference in the amounts demanded between demand buckets.
• Lead-time variability – This is any kind of variability in lead-time from order processing to handling to transportation. It is defined in the Source Lead Time in the Site and Customer Sourcing Policies tables, the Fixed Order Time column in Production Policies, and in the Transport Time column in Transportation Policies.

Safety Stock Optimization uses a target service level when calculating safety stock. This is a measure that represents how capable a site is of providing product to others downstream. Because companies have different methods of calculating these service definitions, Safety Stock Optimization also supports three specific service definitions:

• Type 1 (Probability of not stocking out)
• Type 2 (Quantity fill rate)
• Type 3 (Ready Rate)

In addition, Safety Stock Optimization supports an alternate Type 2 service definition: Type 2 (Undershoot).

You select the service definition to which Safety Stock Optimization must adhere and you provide a Service Requirement, which acts as the minimum constraint on the service. Safety Stock Optimization then determines the safety stock levels, as well as the recommended inventory policy, and associated parameters (reorder point, reorder quantity, etc.).

Safety stock is determined based on the safety stock rule (multi-echelon vs. single-echelon), demand class, and lead-time demand statistics. Safety Stock Optimization uses the mean and standard deviation of the lead time when calculating safety stock. You can enter a number, or enter a normal distribution in the format N(mean,std dev). Additionally, Safety Stock Optimization takes into account probability-based sourcing policies, bills of material and product conversions. For additional information, refer to Lead time inputs.

Safety Stock Optimization makes use of a fixed network structure and transportation modes. If you have run your model with Network Optimization, you can use the NO-IO Conversion transformation to “fix” the network structure with the sites, policies and other components selected during Network Optimization.

Rolling Horizon Modeling enables you to model future forecasts. It provides time-phased safety stock and inventory policies.