Planning a supply chain is an exercise in futility. The assumptions around capacity, forecasted demand, and production rates will most likely never materialize the same way in the real world. With so many moving parts, all that you know is that your plan is wrong. That goes for your forecasted demand plan and your optimized supply plan. When you accept that the plan is wrong, the next question is, “how wrong is it?” That’s where the ambiguous “quality of solution” measurement comes into play. When a solution is wrong by definition, how do you tell what level of quality it is? When I evaluate quality of a solution in a planning system, whether demand or supply, there are three criteria I use to determine if it is good enough, or if we need to try again.
Is it Repeatable and Predictable?
If you run the solution with the same data in the same matter with the same settings, do you get the same results? This seems like a given, but with the complex solutions that drive most planning systems, it is harder than you might think. However getting to this point is critical if you intend to tune your solution, and adjust the settings in your supply chain accordingly. This also applies to understanding the behavior of your supply chain modeling tool well enough to predict the outcomes. If you can get a consistent solution with the same data, you have a solid baseline. With a solid baseline you can then use your understanding of the levers to pull on your solution to drive predictable results. The combination of the two will give users the comfort in knowing that you have not given them a random number generator as a planning tool.
Is it Understandable?
A wise man once told me when troubleshooting a supply planning solution that “as long as math still works, we still have a chance” In other words as long as 2+2 = 4 we can figure this out. However, with the complex equations and inputs to today’s supply chain planning applications proving that concept is easier said than done. Is this solution digestible to the user community? Does it make any sense logically? This aspect is the closest to a “quality” evaluation as we have in quality of solution. Ensuring that negative numbers stay where they belong, that you can trace an input to an output. In short, can you stand behind this solution and pass the red face test when questioned.
Is it Executable?
This is usually overlooked, but is the most critical when it comes to developing a demand or supply planning solution. Can you actually use the outputs to run a business? You can create the slickest optimizer, or the fanciest demand model, but if the results do not fit the capabilities of the business, then it’s all just an academic exercise. Are the run rates within reasonable expectations? Have we ever sold that much product in that market? Do we even have space to hold that much inventory? If your solution can not fit the constraints of the real world then you don’t really have a solution at all.
Yes, most supply chain planning activities that predict the future are always going to reside in the gray space. There is no lab test to determine good or bad. However, if your solution follows the criteria outlined above you should be able to effectively navigate that fog of uncertainty with positive results.