Tuesday, December 15, 2009

The Push Pull Interface: the Optimal Point

A while ago, we talked about the basic characteristics of push or pull based supply chains. We also discussed how push or pull concept should be interpreted within a single supply chain as well as across multiple supply chains supporting the product life cycle from raw material to finished goods to purchase by a customer.

All that raises a question: how can one decide what will be the best point in the supply chain where the order-inventory interface must be located? Just to refresh, the order-inventory interface is the point in the supply chain where the demand fulfillment process changes from “fulfilling from inventory” to “fulfilling from orders”. While the article on considerations for push or pull provided a good view of attributes to be considered, they do not necessarily help in deciding where in supply chain the nature of demand fulfillment must change from inventory to orders. This is the question, I wish to address today in this article.

To answer the question, let us establish some basic facts. Supply chain for this discussion is represented as a chain of activities and nodes connecting supply with demand. Keeping with the convention, we represent the “supply” end upstream of the “demand” end. The Order-Inventory interface can then be located anywhere along the length of the supply chain. It can be closer to the demand end or closer to supply end or mid-way between the two. The question we are trying to answer is how to objectively decide the optimal point where this should be located for optimal supply chain performance.


Next, let us explore through logic what happens when the order-inventory (OI) interface point moves along the supply chain? We will explore the impact of such movement from the point of what happens to demand as this point moves along the chain. For example, if we move the order-inventory (OI) interface closer to demand, it follows logically that the demand fulfillment lead-time will be lowest, while the flexibility to customize products to the order will be lowest as all demand is fulfilled from available stock.

The following table summarizes the observations.


In Managing the Fickle: The True Nature of Supply Chains, we identified that a supply chain’s primary purpose in life was to combat variability in supply, demand, and lead-time. We had also established that supply chains achieve this objective by creating and managing buffers. Buffers can exist for inventory, resources, and time. The observations in the table above enable us to see how the movement of OI point along the supply chain affects these buffers. As the OI point moves closer to the supply end, the supply-chain’s ability to keep inventory buffers reduces, therefore it must compensate by establishing larger resource and time buffers to accommodate demand variance. Conversely, when the OI point moves closer to the demand end, the supply-chain’s ability to keep inventory buffers is enhanced, therefore it can make do with smaller resource and time buffers. Of course, the supply chains must maintain the smallest possible buffers to provide the desired cost performance for the target service levels. This is shown in the picture below.


The point of evaluating the impact of moving OI point on the supply chain’s ability to create and manage buffers is simple. It allows us to see how supply chains will react to such a change and provides us with an objective view of how supply chain performance may get affected as a result. Higher inventories provide shorter fulfillment time, but come at the cost of higher carrying costs and obsolescence costs. On the other hand, higher inventories generally allow mass-production, thus increasing the resource utilization and reducing the cost of resource buffers because they can now be planned and utilized well to their capacity.

As creation and maintenance of all buffers needs capital, eventually the decision to place the OI point at a certain point in the supply chain gets reduced to balancing the cost of these buffers against the cost of stock-outs and lost sales revenue. If the costs of maintaining inventory is relatively higher when compared to the costs of resources and time, then the equation will tilt in favor of the OI point towards the supply side of the chain. On the other hand, when cost of inventory is low compared to the costs of resources and time, then the equation will favor moving the OI point closer to demand. Of course, these considerations are in addition to the basic nature of the products and demand which must determine the suitability of such analysis. But assuming that a supply chain type has been established (for example, commodities versus personalized versus engineered products), this analysis can help in establishing which echelon in the supply chain is the best for placing the OI point.

Here is a summary of the opposing costs to balance when making decisions about the OI point:

  • Costs related to creation and maintenance of Inventory buffers: Balance the costs of maintaining inventory with the costs of obsolescence and clearance of unwanted/excess inventory against the cost of stock-outs and lost sales.
  • Costs related to creation and maintenance of Resource buffers: Balance the costs of maintaining resource flexibility (ramping up/down as required, adding sub-contract capacity at short notice, etc.) against the cost of stock-outs and lost sales if resources cannot be ramped up to meet demand.
  • Costs related to creation and maintenance of Time buffers: Balance the costs of maintaining time buffers (cost of extra capital to maintain additional operations and inventory to hedge against variability in lead-time) against the cost of stock-outs and lost sales.


These costs vary at every echelon of the supply chain network. If they can be modeled and evaluated at various potential echelons of the supply chain, the location for the OI interface can be determined in a more objective fashion. While some of these costs are hard to determine, judicious approximation would be better than making a decision that is completely subjective. Most likely, the optimal echelons will differ by the product’s demand and supply characteristics. Fast moving products will require inventories closer to the end-demand, while the inventories of the slow-moving products can be moved upstream. Similarly, inventories for commodities will be closer downstream, while personalized products must be maintained upstream and typically as sub-assemblies that can be configured to final customer demand.

© Vivek Sehgal, 2009, All Rights Reserved.

Want to know more about supply chain processes? How they work and what they afford? Check out my book on Enterprise Supply Chain Management at Amazon. You will find every supply chain function described in simple language that makes sense, as well as see its relationship to other functions.

1 comment:

  1. Hello! I´ve been going through the literature on this topic and I find that there is not much clarity in it. Especially, I find quite confusing when it is mentioned that we can have the make to stock and make to order approach in either side of the interface. In the book Factory Physics, for instance, they have the taco example in which this happens. In one of the alternatives we have make to order on the left side and make to stock at the end. So, personalization cannot be achieved at the end. How a make to order sistem works at the initial part of the supply chain is what I don´t understand quite well. Also, it seems that inventory levels and resource utilization is mainly related to on what side MTO or MTS is located rather to where the interface is located. I would appreciate your thoughts on this.