Pull Strategy - Historically, the streams of material entering the supply chain are the result of multiple forecasts that attempt to predict the material requirements and push it to each Point Of Use (POU) within the chain. A point of use could be a manufacturing line or a section of a distribution center used to source finished goods to be picked for customer orders.


Forecasts are inherently complex, uncoordinated and inaccurate with a tendency toward overestimation as a hedge against shortages. The cumulative effect is an over abundance of inventory from raw materials to finished goods. Not only does this inflate inventory costs, it also drives up material handling, administrative, and storage costs. In contrast, the pull model is driven primarily by two key variables: demand for items to be consumed and the lead time required to re-supply the point of use before the existing stock runs out.

Simply put, the pull model starts with the end of the chain and works back to the genesis; demand for material at each point of use creates a replenishment request that must be satisfied within its standard lead time creating what is better called a "demand chain" in contrast to a "supply chain". This approach results in rapid inventory turnover, focuses on the immediate "customer" at each point in the demand chain, and is robust for manufacturing, assembly, distribution, and retail contexts. Incorporating cost, volumetric, and other SKU and facility data can refine pull model optimization to achieve high levels of efficiency resulting in increased profitability.
   
  Pull versus Push - As attractive as the Pull model is for increasing throughput, reducing inventory costs, and streamlining material handling, it is vulnerable. Since the model depends on accurately characterizing demand and lead time at the SKU level, overall performance depends on understanding and quantifying these variables. In addition, the procurement process may be only loosely connected to actual demand at any level in the chain. Procurement behavior anticipating demand may be counterproductive to overall profitability as a result of bargain purchases that generate excess inventory. (These may or may not be driven by marketing and sales forecasts.) The key to success is integrating forecasts, actual demand, and manipulating lead time to achieve balance between Push and Pull. This can not be done without information technology.

Achieving Balance - A manufacturing facility or distribution center only exists to satisfy demand. Their output is almost always driven by forecasts of demand. Demand is easiest to understand at the end of the manufacturing or distribution chain: how many units must be shipped today? Working back up the chain, uncertainty creeps in. How much material must be available in a forward picking location to achieve 100% fill rates on orders that are not yet visible to the distribution center? What supplier delivery schedule should be negotiated to stock the warehouse for pick module replenishments for a seasonal item? Does a quantity discount justify purchasing above the demand level?

As a general rule, the push model is applied more at the beginning of the chain while the pull model kicks in toward the end. Moving pull model control toward the beginning of the chain results in a leaner, more predictable, more efficient fulfillment process. Chiron's i-Lean product can help you shift the pull methodology toward the front of the chain.

i-Lean for Design - Implementing a pull oriented facility starts with a good design that focuses on material flow and storage to optimize point of use performance while minimizing inventory balances and handling costs. i-Lean can integrate the characteristics of thousands of SKU's to satisfy dozens of design constraints to generate detailed recommendations for material flow, point of use and reserve inventory balances, slot placement, and storage media utilization consistent with a pull strategy. i-Lean also provides the capability to directly compare the detail of multiple "what if" variations, complete with all of the associated costs. These design outputs can be used in conjunction with virtually any warehouse management system to operationalize pull model material handling.

A Good WMS for Execution - With a good design in hand, the next step is integration with the execution environment. A good WMS matches demand with resources in support of the pull strategy. The WMS assimilates all of the demand in a facility including outbound, inbound, replenishments, and general material handling tasks and distributes them on a priority basis to the best available resource - the right person with the right vehicle. Order picks are only assigned when there is available shipping space for consolidation at the dock. Both of these techniques are examples of the pull model in action - tasks are pulled through the process based on resource availability. Coupled with outputs from i-Lean, pick module inventory balances are designed to satisfy demand so that fill rates are high, replenishment tasks are assigned to maintain these balances within lead time, and material is positioned throughout the warehouse to maximize throughput and minimize material handling costs.


Forecasting - While we're not big fans of a total reliance on forecasts, particularly toward the end of the demand chain within a facility, we do recognize that good execution data provides valuable feedback to planners. A good WMS houses a treasure trove of data that can be used for a variety of purposes - not the least of which is input to the planning process. The WMS should be able to extract, filter, sort, and export virtually any view of its operational data, from dead stock to detailed inventory location history. This information not only improves planning, but also helps move it further up the chain.


Top