Embracing Uncertainty in SCM (with Stefan de Kok)

00:08 Introduction
00:30 Stefan, perhaps you could start by telling us a little bit more about your background and Wahupa, the company you co-founded?
03:16 Which are the other classes of uncertainty that we can encounter, besides demand?
04:50 The idea of an uncertain future is very much at the core of Lokad’s approach. How do you approach these challenges Joannès?
06:42 What do you think of the more classical approaches that people are taking to account for that level of uncertainty?
09:59 Where did the idea of probabilistic forecasts come from?
12:34 Stefan, what led you to take this approach of embracing uncertainty?
16:06 Do we need to change the metrics?
18:36 What strategy can we employ, from a software perspective, to help customers deal with uncertainty?
20:17 What does it mean for a company to switch to a probabilistic approach?
22:37 To conclude, would you say that people are ready to embrace the idea of uncertainty?


From demand to supply, and almost everything in between, there is a huge amount of uncertainty within supply chains. Traditionally, this can be managed using buffer stock, but there are often events that occur which can trip up even the most experienced of practitioners. For this episode of LokadTV, we’re joined by Stefan de Kok to discuss why this uncertainty is not necessarily a hindrance but something that should be embraced.

Stefan is the co-founder and CEO of Wahupa, a supply chain management solution for small and medium sized manufacturers. Stefan studied applied mathematics before working in various consulting roles. He came up with the idea for Wahupa in 2003, when he began to gather a talented team around him who could make it happen.

In a supply chain, anything that can happen in the future is uncertain to a various degree. Therefore, we believe all possible futures should be taken into account. Yet, even in the past, there can be uncertainties - if there are data errors for example.

Historically, supply chain science would deal mainly with “what if” scenarios, where normally a number of worst and best-case outcomes would be established. This is a tedious and time-consuming process. However, a probabilistic forecasting perspective where all futures are considered is actually easier to implement - with enough raw processing power available - than a complex system that manages a limited number of “what-if” scenarios. In addition, from an operational point of view, a probabilistic approach also requires far less manpower.

We go into more depth about how most companies deal with uncertainty in their supply chains, mainly through the use of “buffers” and other techniques such as expediting, or even just by ignoring the uncertainty completely. We discuss why, in forecasting, we should move away from the classic idea of predicting percentages of error and instead push for calculating dollars of error.

To conclude, we ask the question whether industries are ready to accept this complex idea of embracing uncertainty and what can be done to help encourage that this concept becomes accepted by the mainstream.