By Howard Roddie, Senior Supply Chain Consultant
How does a pilot learn to land an aeroplane? One scenario at a time, or by comparing multiple scenarios simultaneously? I’m at 35,000 feet at the moment, so the answer matters!
In the world of Supply Chain Planning, the ability to create and store scenarios based on different inputs and parameters is always raised and we are often asked if we can manage multiple “What-if” scenarios.
The answer is, of course, “Yes”, we can create planning scenarios; store the results and compare numerical and percentage differences and we can even plan in weeks, months, quarters and years. But what does it mean? What more do we know at the end of this?
When supply chain professionals first see their plans laid out in front of them with the clarity of a good map it seems logical to ask if they can compare multiple scenarios. This is an even better question than it first appears. Of course, the question that is being asked relates to the capacity of the software tool to calculate, store and display the results. But the question should be about our ability to understand the inputs and outputs of more than one situation at a time.
Consider a typical supply plan and just think about some of the inputs. There’s demand from many customers at many demand points as well as planning policies for each product at each stock point. Then there are a variety of shared constraints to consider (Maximum production capacity, Maximum stock levels), not to mention all of the alternative routings.
Once the inputs are set and a plan has been produced, we must look at the outputs; stock at many levels, production and distribution plans. All of these have to be understood together as part of a single model to be truly useful.
This understanding can be achieved with both the use of simple mental modelling techniques and well laid out visual software tools. But the next step adds layers of complexity that can be beyond our comprehension. Already in our single scenario model we need to know, for example, how a change in production capacity will impact on our stock levels in multiple echelons. This we can show and tell.
Now we try to add another layer: to compare the original to an alternative view of the future. How do we display this without either cluttering and clouding the view or losing vital parts of the simultaneous stories we are telling? How do our minds process the increased complexity of the introduction of alternatives? Can we truly separate the two scenarios? Indeed the more time we spend looking at them, the more they will seem to blend together.
This is not to say we should not plan for alternative futures. Instead, it shows that we need to change the way we look at scenarios. We should always have our current plan for operational and tactical decision-making based on the best available current information. But, possible futures should be taken offline and modelled separately, ideally on different days.
We should develop our scenarios as learning models so we can start to predict and check likely outcomes to our actions, without the fear of failure. As Arie De Geus says of managers without scenario planning tools, “We ask them to learn ‘by experimenting with reality’”
Why do you think Formula1 drivers spend so much time on the playstation? Do they ever drive 2 versions of the car at once? Imagine if airline pilots learnt this way? Would you fly with them? I’m now at 27,000 feet and descending…
 De Geus, A.P. (1992) Modelling to predict or to learn? European Journal of Operational Research, 59, 1-5