The prevailing paradigm of supply chain planning solutions reduced to its bare essentials is this:

Build an “optimal” plan based on a snapshot of the current state of the supply chain and what is believed to be the most likely estimate of the future. Execute plan. Repeat as often as necessary.

There is a crucial flaw in this approach: the approach overlooks the fact that reality could, and almost always does, differ from the most likely estimate of the future. Demand is never as forecasted; execution is never as planned. As a result, reality and plan diverge almost as soon as the plan is published.

Therefore, state-of-the-art systems that embody this paradigm take several measures to compensate:

  • Re-plan at a high frequency:
    • Revise the snapshot of the current state to reestablish a new starting point
    • Revise forecasts to reflect the most recent (and hence best) estimate of the future
  • Build buffers into the system to ride over the inevitable bumps in the road.

The incorporation of buffers is usually done using inventory planning modules that compute safety stock levels by looking at demand and supply uncertainty.

This approach overlooks a few important factors:

  • The sources of uncertainty in supply chains are far more diverse than generally acknowledged
  • The nature of that uncertainty is poorly understood
  • Most importantly, when reality and plan diverge, the ability of the supply chain to react is limited by the lead times involved
  • Fortunately, the strategies that are available to plan in the face of these uncertainties go far beyond safety-stock planning. In most of the projects described below, we took a careful look at the sources of uncertainty, characterized them succinctly, and devised policies that would make the plan robust in the face of that uncertainty.