‘How to make an accurate business forecast?’ It’s a question that has been around for some time. The good news is, it's getting easier. The availability of advanced technologies — such as machine learning — continues to grow. On top of that, more and more people have mastered these technologies and can implement forecasting algorithms. But there's another side of the coin. Now that forecasting is considerably easier than it used to be, it is tempting to implement advanced algorithms. Before doing so, you may want to consider whether it's essential to do so. These three questions will help you determine the answer!
Think this is an obvious question? In our experience, that's not always the case. We'll illustrate with an example. Suppose you want to forecast the number of departing passengers at each airport terminal per day. It seems like a logical thing to do, as you'll be able to recognize clear patterns: at the airport, some weeks or days tend to be busier than others. It is wise to schedule crucial airport processes more efficiently based on forecasting outcomes.
In reality, though, you'll spend time forecasting numbers that are already known. After all, flights are scheduled well in advance, and a low percentage of tickets are sold at the last minute. So, an airport should try to obtain this information from airlines rather than implement intricate models that forecast passenger numbers based on historical trends.
Say, you work at a large production company and want to forecast end product sales. The goal is to optimize your production schedule and inventory management. Again, it seems like a logical choice, as you can discern clear seasonal and weekly patterns (among others).
But in some cases, it's better to start by spending more time understanding your biggest customers' production and sales processes. You might be able to conclude agreements on timing as well as the volumes big customers will purchase. To do so, you may even want to negotiate on price. After all, this approach will allow you to optimize your processes without having to deal with any forecasting inaccuracies.
Of course, there are cases in which forecasting is the best option. But are you always the right party to implement the required models? Or are there partners in your value chain that are better equipped to do so? It's important to consider whether you have the knowledge and information you need to forecast. If not, you might want to use a partner's forecast. And if this partner belongs to your value chain, you might be able to transfer part of the forecasting risk to them — for example, by having them carry most of the chain’s inventory.
Sometimes, the answer is closer to home — there are instances in which you should just leave the forecasting to another department in your company!