00:08 Introduction
00:27 Today we are going to look at scenario planning and how it compares to probabilistic forecasting. How does the two approaches relate?
00:54 What is scenario planning? How does it work?
02:29 With this method, can you combine multiple variations such as demand or lead times?
05:57 Why is it better to take a computational approach?
08:08 Why is scenario planning something which is still so popular with consultants? Why are companies still using it?
10:18 Why is probabilistic forecasting so different?
13:45 Is the probabilistic approach easier for the end-user?
17:18 Is the implementation of a probabilistic forecasting approach more difficult?
20:41 Can you imagine scenario planning dying out at some point?
Description
Scenario planning was first pioneered by Shell in the 1970’s and since then has been promoted by consultancies worldwide as a powerful tool to help companies prepare for every eventuality. Here, we discuss the effectiveness of this approach and whether it can be replaced by alternative methods, such as probabilistic forecasting.
The two methods are similar as they deal with uncertain futures; as nobody can be sure of what the future holds, it’s normal that you want to explore options. However, how the two methods explore these options is very different. Scenario planning is conceptually extremely simple and creates “what-if” scenarios; it’s elegant and easy to replicate.
Due to the extreme complexity of supply chains - demand, lead times, 1000’s of SKUs, multiple locations etc. - technology is required to help deal with the multitude of possible, future scenarios. As scenario planning is so easy to implement and easily creates mulitple scenarios, from a software perspective it’s literally a question of “cutting and pasting”. Despite this apparent simplicity, manual tweaks are required due to the numerically overwhelming amount of combinations. Therefore, it remains a very human-driven affair, as human expertise is required to cherry-pick these scenarios. In terms of productivity, supply chain planning is already highly time consuming, requiring entire teams of planners and forecasters to create primary scenarios - i.e. average classic forecasts. To add more scenarios and fuel the process, far more manpower is required. When it comes to the day-to-day running of supply chains, scenario planning is rarely used because it’s so costly. Yet, it’s still a technique that consultants and vendors continue to push.
The alternative approaches that have now emerged, most notably probabilistic forecasting, have been able to come into being thanks to more readily available computational power that simply wasn’t a possibility before. Naturally, probabilistic forecasting is a far more computational approach than scenario planning and requires less human input. Probabilistic forecasting is a mechanical way to put probabilities on a vast number of possible future events. As soon as you have this you realise that you don’t need “scenarios” anymore. There’s no radical shift as such, but it’s almost comparable to moving from a blurry, grainy picture to a HD photo.
To conclude, we talk in more detail about the impact of these two approaches on the end-user, the dangers of using averaging to forecast when using scenario planning and how this can quickly multiply the amount of staff required. We also go into more detail about software vendor strategies and why scenario planning is still sold as a viable solution.
