Supply chains have historically been built with forecasting at their centre, with the actual decision making process left to the intuition and experience of senior staff. We explore the alternative approach of putting ‘decisions first’, and how this philosophy vastly improves the supply chain performance.
So what do we mean by a ‘decision first’ approach? Through the definition of economic drivers, practitioners can focus on the decisions which actually matter. Many people believe this to be an easy step, but it’s not. It requires a lot of refinement to get somewhere sane, as you can’t actually know your optimization target until you start optimizing. It will take many feedback loops before improvement starts to show. It must be said, the idea of starting from the base of a crude decision can feel counter-intuitive, with the obvious first step of every data project being to ensure that the data is clean to allow a solid foundation.
But in fact the chances are that the measurement will already be extremely flawed. As far as KPIs go, if consultants are involved, they’re more likely to go for the easy options in order to avoid any danger. Relying on low hanging fruit means that the bigger picture isn’t properly taken into account.
We’ve always maintained that you can’t optimize what you don’t measure - and we continue to do so -, therefore data is always needed to back any claim. We go more into detail about the problems of bad data and why it’s not always so simple to imagine that eliminating bad data will have a magic wand effect.
To wrap this episode up, we discuss what supply chain scientists should spend their time focusing on, and how supply chains as a whole can become more reactive, ultimately by taking leaner approaches and removing unneeded decisions.
00:25 Joannes, perhaps you could start by clarifying what you mean by taking a “decision first” approach?
02:27 What made you take the “decision first” approach at Lokad? How did you come about this realisation?
04:29 This idea of starting with crude decisions is certainly counter-intuitive. Surely the first step of every data project is to clean data to ensure you are working with solid foundations?
06:16 How do you get to the final decisions that drives the metrics? It seems that there is a lot of trial and error, right?
08:53 You can’t optimize what you don’t measure. What is the end goal then?
12:05 The industry is very focused on this idea of forecasting accuracy. Why do enterprise vendors or consultants focus so much on forecasts?
14:43 What are the alternative approaches that people can take?
18:30 What would you do to convince those organizations that are used to take more classical approaches?
21:56 To conclude, do you think the industry is ready to embrace a whole new approach?