Improve Forecast Accuracy by the Consensus Process
Business forecasting is hard. As Mark Twain once said, "the art of prophecy is very difficult, especially with respect to the future." The key to developing an accurate forecasting system, as evidenced by a detailed case study in the Harvard Business Review (pay to view, unfortunately), is to incorporate all relevant information, wherever it may be, in whatever form it may take -- opinion, internal data, market data, whatever. The case studied Leitax, a medium-size digital camera producer. Here's a summary of how they improved their forecast:
- The Problem: Leitax didn't have a single, unified forecast to coordinate their operations. Sales developed their own projections, which production distrusted because the sales organization has the incentive to produce low-ball projections so they can "beat the mark." Production, in turn, operated on their own version of the forecast, with their own assumptions. Finance had yet another forecast that always predicted achievement of financial targets so analysts would be happy. The absence of agreement and coordination created a truly discombobulated supply chain with alternating overstock/stockout inventory levels, and Leitax endured all the associated costs.
- The Solution: Management at Leitax developed a system to synthesize all the forecasts. The two underlying principles were the efficient and timely sharing of information and assumptions, and respectful consideration of different forecasting perspectives. This manifested in a system called the Consensus Forecasting Process (CFP), which established a new group to manage the process, called the DMS. The DMS provided a statistical forecast based on internal company data. Then another group created a statistical forecast based on external market data. Every month, the DMS meets with heads of sales, production, and finance to agree on a synthesized version of the five forecasts, which is used to coordinate planning across the entire organization.
Note: Leitax is a psuedonym