- Coronavirus is having an enormous impact on purchasing.
- Companies need to perform adjustments on their demand history to account for Coronavirus.
Coronavirus is an international pandemic that so far has had the following effects on demand.
See our references for this article and related articles at this link.
- A dramatic spike in many consumer items related to the virus, such as sanitation products, canned goods, bottled water, and similar items in many countries.
- A dramatic decline in in-restaurant purchases.
- An enormous decline in the supply of items, particularly those manufactured in China.
- While less discussed, it seems likely to conclude that many purchases are not being made due to the population being distracted by the Coronavirus (think big-ticket purchases)
Forecasting is not about responding to short term unpredictable changes in demand like the Coronavirus. This article is about forecasting, but about the the future past the Coronavirus. I will cover the implications of the changes in demand on what will eventually become demand history and what this means for historical adjustment when future forecasts are generated.
The Model of Demand History Adjustment from Promotions
In the book Promotions Forecasting: Techniques of Forecast Adjustment in Forecast Systems, I covered the techniques for applying changes or adjustments to demand history to account for previous and for future forecasts. This book could have been called out for one time events like Coronavirus as well — however, at that time, I was not aware Coronavirus would strike in the future, and pandemics weren’t on my mind at the time.
The Problem With One Time Events
The point of the book was that promotions need to be accounted for in-demand history — if not the statistical forecast will end up using the unadjusted demand history and will repeat the one time impact on demand into the forecast. That is, unless the system is told otherwise, the system will assume that the onetime event is authentic demand. One way around this is by applying outlier removal. If the outlier is larger than the outlier removal, the forecasting system will ignore the event. However, as we cover in the article How to Understand What is an Outlier in Forecasting, automatic outlier removal comes with its host of issues.
The Distinction Between Adjusting Demand History for Promotions Versus One Time Events like Coronavirus
There is a major difference. Promotions occur typically around two times during the history of the product in the food and beverage industry, at least. For a major one-time event like the Coronavirus, a far larger number of products will be impacted. This will place a large adjustment load forecasting departments. On the positive side, it should be straightforward to identify the impact of the Coronavirus as the impact has been large, and the timing of the spread of the Coronavirus is well known.
Therefore, compared to promotion demand history adjustment, the identification of the change in demand will be much easier. Marketing tends to not inform Demand Planning of the promotions, so Demand Planning often has to ask Marketing if the change in demand they see in the demand history was connected to a promotion. However, the overall number of product location combinations that must have their demand history adjusted will be far higher.
Promotion and Onetime Event Management
Within the Brightwork Explorer, we have an adjustment functionality that was designed for promotions but would also be effective for a onetime event like Coronavirus.
Companies will need as much help as they can get making all of the adjustments to the demand history sourced to the Coronavirus, and the Brightwork Explorer would be a true time-saver.