Selecting Supply Chain Connections

Published on January 31, 2013

Filed under: Supply Chain

Selecting Supply Chain Connections

In a previous article on the supply chain, the following different types of supply chain connections were covered:

  1. Sequenced FIFO
  2. Single supermarkets
  3. Dual supermarkets
  4. Rolling kanbans

These connections set up a single method for signaling the supplier to send / don’t send along with a fixed pathway to send the material.  They establish “normal flow” for the supply chain and let us know when that flow starts to become abnormal. 

While the connections are a powerful tool to link the supply chain into the flow of our operation, we still need to select the right connection to use for each part we purchase. In traditional supply chain management, we might have a meeting or set up reviews to determine how to communicate with a supplier.  As we go forward with Operational Excellence, we want to use a standard methodology, or process, that everyone can follow to determine which connection to use, and not leave it up to management. 

In our process of determining which supply chain connection to use, we need to consider some attributes about our supplier and about the parts the supplier is shipping to us. 

First, we need to consider the lead time. Parts can either have a short lead time or a long lead time, with “short” and “long” being relative here.  In some companies, a long lead time could be four weeks, while in others, it could be four years.  In order to determine what is a short lead time or long lead time at your business (and be able to teach it), make a bar graph out of the active part numbers and sort them by lead time, then split the part numbers down the middle, with short on one side and long on the other.  Doing this will provide a rough indication of what constitutes a short lead time for your company and a long lead time. 

After lead time, we need to consider the demand variation of each part and whether it is a runner, repeater, or stranger.  Runners are parts that have consistent, high-volume demand.  Repeaters are parts that have intermediate demand prone to some fluctuation.  Strangers are parts with low or erratic demand. 

To determine the consistency of demand for our products, we can use something called the coefficient of variation, the equation for which is below. 



When we sort the demand for each part by looking at it spread out over a daily basis, as a guideline, think of any CV we get less than 2 as a runner, anything from 2-2.4 as a repeater, and anything greater than 2.4 as a stranger. 

The next variable we need to look at is part cost.  Similar to lead time, part cost varies across industries, and what constitutes a high-cost part in one industry could be a low-cost part in another.  So, similar to what we did with the lead time, we need to divide our active parts into high, medium, and low cost.

The final variable we need to consider is something that can be loosely called supplier reliability.  Many types of criteria can be considered, but the important thing is that at the end of the day, we want to determine whether a supplier has a rating of “good” or “fair.”  Note that we don’t have “bad” or “unreliable” as an option, because suppliers that fall into a categorization like that are probably not good candidates for setting up connections anyway. 

We then take our criteria (lead time, demand variation, part cost, and supplier reliability) and put them on the top of a grid, and down the left side we put our supply chain connections.  In the graphic below, the grid has been populated with Xs. 

To use this grid, we look to see where our Xs fall within each of the four variables.  Where the Xs fall dictates the supply chain connection we use. 

For example, if we have a part that has short lead time, is a runner, has high cost, and good supplier reliability, then we would have either the sequenced FIFO or single supermarket connection available to use. 

Note that following the standard work did not point us directly to one connection over the other.  It did not tell us that we should definitively use sequenced FIFO or definitively use a single supermarket.  Of course, if we are able to, we would want to use sequenced FIFO because it is the connection with the most flow.  But in a situation like this, we could start by implementing a single supermarket with our supplier and, at a later date, implement sequenced FIFO. 

We would then go through the rest of our parts and their suppliers and use this process to determine a supply chain connection for each one. 

All of this work makes up only the first step of creating flow in the supply chain for Operational Excellence.  We then need to complete the rest of the steps in our process and make the flow visual, create standard work for it, make abnormal flow visual, and then create standard work for the abnormal flow.