When you generate a statistical forecast, it is important that you use the demand history and not the sales history to project into the future. The sales history, you download from your ERP system, is a record of what you sold each month, which is not necessarily a record of what the true demand was.
Let’s look at a few examples: -
01. No Sales in a Period:
You had no sales in a period because you were out of stock, or you couldn’t build products to sell because there was a raw material shortage. You don’t want to forecast a shortage in the same period next year, so you need to modify the figures to reflect, as close as possible, what the true demand was in that period.
02. Enormous Sales in a Period
You had enormous sales in a period due to a one-off, never-to-be-repeated, order. This sales figure needs to be reduced to what the normal demand would have been in that period, otherwise the forecasting algorithms could try and suggest record sales again in the same period next year.
03. Negative Sales in a Period
Often you might find negative sales in a period, due to product being return from a sale in the previous month. Obviously, you don’t want to forecast negative sales in the same period next year so you need to modify the figures again.
04. Changes in a New SKU Number
Sometimes you may have a product that has changed slightly and is given a new SKU number. This will mean that the new product, which is basically the same as the old one, has no demand history for the forecasting system to use to predict the future. In this case the old SKU demand history should be copied to the new SKU number and the old SKU number data removed from the demand history file.
05. Changes in The Demand for a Product
Lastly, demand for a product can be because of events as opposed to seasonality, and these need to be flagged and taken into account when the statistical forecasting takes place. For example, Christmas is seasonal, it always happens on the same day in the same month, whereas Easter is an event, as it moves from one month to another. Other events can be such things as your company promoting products, or conversely your competitor promoting their products and your sales diminishing. Random major sporting events could also affect your sales and should also be flagged as events. In some industries, public holidays falling on different days of the week can have a significant difference in their sales.
So, taking all the above into account, it is important that you develop a demand history file that is used to forecast your products into the future, and that the latest sales history is appended to this file monthly and then ‘massaged’ into, as close as possible, the true demand for that month, before statistical forecasting can take place.
In addition, make sure all dead products are removed from the demand file, you don’t want to waste forecasting effort on lines of old products that will never be sold again. If they are resurrected in the future they can always be added back into you demand history file.
CFPIM, CSCP, SCOR-P, CPF, CS&OP, PLS, CDDP, CSCA, CDDL, CLTD, DDPP, DDLP, AEF, CSSC, CPIA Chief Executive Officer at Kent Outsourcing Services
A successful career in supply chain necessitates a never-ending thirst for knowledge and the most up-to-date supply chain skills and knowledge. Even though continuing education and certification programs can take time and money, they help get your specialized knowledge or skills recognized.