Alice asked: "Would you tell me, please, which way I ought to go from here?" "That depends a good deal on where you want to get to," said the Cat. And this same conversation (or at least a variant of it) happens all of the time in today's commercial enterprises as they decide where to point the business and how to get there.
The past twenty-five years has seen great progress in supply chain planning capabilities, and companies continue to make progress in developing and deploying operational plans that are more and more detailed, and which have improved congruence with the financial goals and plans of the enterprise. The continual march of improvement in computer hardware and software have provided today's planners advanced supply chain models and tools that can handle large amounts of data, honoring constraints and optimizing on costs or profits. If it weren't for the real world, we might have achieved supply chain Nirvana by now. Unfortunately we live in a world where it is impossible to know the future and stuff happens. We don't really know what our customers will want, what our competitors might do, or what might happen at our suppliers. In other words, we are trying to optimize our plans under uncertainty.
Most supply chain models are deterministic. We build a model based on assumptions about demand, supply, costs and income and if our assumptions are all correct, and we execute to plan, the expected profits will appear. We know this isn't going to happen, so we build in extra inventory and we continually look for planning tools that can re-plan in an instant as things change. This will only take us so far because the inflexibility of deterministic planning and the anarchy of the real world will always be in conflict. What if we could develop supply chain plans that took uncertainly into account? After all, the financial sector has been building such models for years. No one on Wall Street pretends that they absolutely know what the market will do, instead the build stochastic models that include the probabilities of the inputs and outcomes.
In the future, supply chain models will increasingly use probabilistic modeling, and the resulting models will be much more flexible and better at managing risk than today's models. But why wait? Why not get started now? There is a branch of supply chain planning that has already adopted stochastic principles: inventory optimization.
Inventory optimization balances desired customer service levels against inventory investment by modeling the variability of both demand and supply, and then placing inventory optimally in your supply network, coming up with the best postponement strategy. Such a plan minimizes risks at the lowest investment possible. This is exactly what we would want in a holistic supply chain plan. Unfortunately, this technology is just beginning to achieve wider application and acceptance.
Why are we not there yet? We could be cynical and say that the talent has been attracted to the hedge funds instead of the shop floor. But that will change as people see the potential benefits at the global enterprise level. A slightly less cynical explanation is that managers are loathe to adopt plans they don't fully understand. This is an obstacle that will pass with time. Twenty years ago, the COO that would site a plant based on the output of a linear programming model was a very rare bird. But that technology is now today an integral part of supply chain planning. Increased adoption of stochastic principles is just a matter of time.