Most businesses within the FMCG over forecast their demand.
A forecast must be correct on average, because both under and over forecasting demand leads to problems:
|
| Over-forecasting
demand results in over-stocking of a product when the forecast is used to convert stock targets into a number of cases to make. This in turn incurs needless stock holding costs.
|
| Under-forecasting
demand leads to a failure to serve demand, and therefore lost customer sales.
|
Positive Bias Example
The following graph shows forecast bias for six SKUs sampled from a previous client’s data. All the SKUs have a positive bias in their forecast – and two of these have a bias in excess of 90%.
|
| |
Bias should be measured and monitored
Rather than use the common textbook definition of bias, we use a normalised bias calculation. This gives a much clearer indication of what effects any bias within the forecast will have on stock levels.
For example:
- A positive 10% forecast bias will result in stock levels that are 10% higher than necessary;
- A negative 10% forecast bias will result in stock levels that are 10% lower than necessary.
We recommend conducting Bias analysis over a year and using the following guidelines
|
Forecast performance feedback
It is important that those responsible for creating the forecast receive regular feedback as to how the forecast is performing and what impact it is having on stock levels, customer service and production.
|
|