Some people make it their job to try and predict the future, but we aren’t talking about psychics. Demand forecasting is an essential piece to the operation of the CPG industry, and the difference between an accurate forecast and an inaccurate forecast can be the difference of millions of dollars. There has been a long time debate on the best method of demand forecasting: What is the best formula?
A Shot at Predicting the Future
Demand forecasting is as much of an art as it is a science. Knowing when and where promotions are going to happen is important because retailers will stock up six to eight weeks prior to the promotion. An apt forecaster needs to work backwards — if there is a promotion three months from now, the factory needs to know to increase its production levels two months from now to keep up with the increased demand of the upcoming promotion.
Knowing what to anticipate from the retailer is the key to an accurate forecast. While a normal, baseline volume is predictable and easy to produce when you know the constant rate of inflow and outflow, it is also highly unlikely to happen. A baseline volume only factors in predictable elements such as sales and production levels from previous years. However, there are far more variables than pure statistical numbers from past years; there are a large amount of outside influences that can’t be predicted with total accuracy.
Why is an Accurate Forecast is Important?
Producing an inaccurate forecast is going to decrease productivity, limit supply and raise costs. One direct benefit of an accurate demand forecast is that it can reduce the downtime of production lines and increase the productivity of a factory. An accurate forecast reduces the number of times they must stop production lines to sanitize and switch products.
For example, an oversupply of product increases the risk of it sitting idle in a warehouse. Even if the product doesn’t have an expiration date, the manufacturer could end up wasting resources by taking away from working capital and important warehouse storage space.
The Human Factor
There has been a long time debate of which forecasting method is best. Do we base forecasting on statistical records, or do we forecast using other outside variables? Perhaps the question isn’t which is better, but rather how can we incorporate the best of both methods? While you can’t build an accurate forecast without statistical models, you also can’t exclude extraneous variables such as competitor sales, weather, human factors, social media and how the news can affect product sales. (A retailer that suddenly dons a bad reputation will cause an unforeseen reduction in sales.) Twitter trends can suddenly shift and start a good or bad conversation surrounding your products or a competitor’s products. Social media trends are unpredictable and can cause a surge or reduction in demand.
A blend of both methods might be the best solution. Statistical methods work on a baseline, and anything promotional – or in a display – has a human element that is not purely based on historical numbers.
Does demand forecasting operate best on a purely statistical level? No. It needs that crucial human factor.
By: Michael Dickenson and Steve Hultman