The accuracy of demand forecasts is considered a critical ingredient to reduct stock-outs and improve service levels. However, when demand is strongly influenced by the weather, it makes sense to consider refining the demand forecast by leveraging a weather forecast.
When General Eisenhower was asked why the D-Day landings were such a success he reportedly replied “because we had better meteorologists”. With technology continuously increasing the overall accuracy of weather forecasting, a common question we are often asked here at Lokad is whether meteorological data can be leveraged to optimize a company’s purchasing decisions.
In this episode of LokadTV, we explore this concept in a little more detail and discuss whether this data can be utilised in any concrete method to improve forecasting accuracy. The weather impacts our lives in many ways: you’re more likely to buy a certain kind of car for example depending on the type of climate you live in.
When you stop and think about it, it’s all too easy to imagine how many products can be affected by the weather, such as ice creams, barbecue meats and winter coats to name but a few. In addition, from a technical perspective, weather data is now available on a daily and very accessible basis. It therefore seems highly logical that weather data could be of some considerable help.
However, we learn why, from a supply chain perspective, weather forecasts are often too short-term to be useful and why it is actually extremely difficult to leverage this data. We discuss how the inherently localised nature of weather can introduce complications and debate how much trust can really be placed in meteorological data.
To bring things to a close, we explore the various implications of climate change and try to understand whether today’s questions of global warming will have any impact upon seasonal forecasts.
00:37 What’s the idea behind the use of weather data to forecast demand?
01:47 How well does it work in practice?
03:01 What did we learn when we tried to use this data at Lokad?
06:30 What are the technical challenges which are introduced?
08:35 Is there something more reactive we could introduce? Is there not something simplistic that can be useful?
11:55 Could it be something we use in the future when we forecast more accurately and have greater control of our supply chains?
14:19 How about climate change? Can we leverage this insight to improve our seasonal forecasts?
17:03 What other data should we be looking at? What is more important than weather data?
19:28 What is our main conclusion today? It is possible to use weather data, but there are other things which are more important?