Good forecasts are often considered the corner-stone of successful manufacturing planning and control processes in companies. In Centre for Logistics we work with the challenges of forecasting.
It is our experience that companies, despite their best effort, often fail to give good forecasts. The typical reasons are:
- Data of insufficient quality (e.g. order records are used, but they do not display a true representation of demand, but rather what we ended up delivering because of negotiations and order modifiers)
- Data of insufficient quantity (advanced forecasting methods often require large amounts of data)
- One forecast is made, but it is used for many purposes (input to puchasing, capacity planning, sales management etc.)
- The forecasting methods are not updated (or one method is used for everything)
- Forecasts are not corrected with qualitative market insight (e.g. sales campaigns, vacations etc.)
- The forecasts are made, but not used.
The consequence of this is that forecasts often do not have the right form, precision, level of detail and time horizon for the multiple purposes they are used for. Since forecasts are input to the planning processes, poor forecasting performance leads to poor planning performance. To compensate companies often have to buffer excessively on capacities, materials or delivery times. This again leads to higher operating costs and lost competitiveness.
In our work we therefore strive to achieve data of the right quality and forecasts in the right context with the right time horizons and level of detail. To do this we must look at other aspects of the company besides the forecast in itself. That is why we have an integrated view on diagnostics, forecasting and planning and control. When working to improve forecasting in companies we therefore start by identifying the needs for the forecasts. Then we diagnose the current state of data, forecasts and planning and lastly, we re-design the process in cooperation with the company.