Markets are getting more dynamic which requires more advanced forecasting techniques. Traditional forecasting models use historical data to generate predictions and therefore are not picking up on events that change the market. When integrating market drivers in statistical models, for example price development of feed stocks we can model the dynamics of the market better. This project was carried out for Rockwool, the world leading producer of stone wool-based insulation materials and applications.
Starting points
The company had a long-term demand planning process, which was characterized by an inaccurate forecast, difficulties in translating market indicators to specific company market developments and no possibilities to generate fact-based demand scenarios.
Project
The challenge in long-term forecasting is to find the relevant drivers that predict market changes and development. First, we organized a workshop session with the relevant stakeholders to determine potential leading market drivers. We then collected and cleaned the data and organized a second workshop to agree on a short-list of indicators. Next, we set up the statistical forecasting model including the market drivers and we developed a modeling tool specific for the company. We handed over this tool and trained the users.
Results
Our model supports Rockwool to understand the market (five quarters into the future). The company changed from ‘gut feeling’ to fact-based forecasting with the possibility to create and evaluate different demand scenarios. This provides an excellent starting point in the budget discussions and the corresponding marketing approach.