3 September 2017
The term “Sales Forecasting” is such a commonly used term, that I had completely forgotten how inappropriate it actually is: almost everyone who refers to “Sales Forecasting”, usually means they want a Demand Forecast. Sometimes it takes a short moment of reflection to appreciate these not-so-subtle differences. As is often the case, this thought popped up as I was trying to prepare a data set for analysis.
Just because data are easily available, that doesn’t mean they are suitable for your purpose. Nor that you can draw valid inferences from them J Accurate sales data are much more readily available because the primary business process “naturally” captures accurate and reliable data on sales. If the product did not get delivered, your client will let you know. The bill doesn’t get paid. Etc. Primary business processes have a natural tendency to gravitate towards rather accurate reporting. If something goes wrong in the primary business process, this quickly trigger strong feedback loops to correct the deviation.
Do you see what the problem is when you use sales data for Forecasting? When you naively assume that the sales accurately reflect demand, then what about stock-outs? Returns, etc.? For marketing and business development purposes you (almost) always want information about demand, not your historic ability to meet that demand as reflected in your sales. If that distinction is too subtle for you, then maybe Data Science isn’t for you. Or you may wish and pray that only clean and reliable datasets cross your path. My journey has typically been rather different – and I love it!