Terra’s Forecasting Benchmark Study
Key Findings 2013
It was an interesting year for forecasting. As the economy continued to recover in North America, supply chains became considerably more complex while shipments for the group as a whole remained essentially flat. Complexity from innovation reversed a three-year trend of improved forecast value-added (FVA) which dropped slightly in 2012. Average weekly forecast error also dipped slightly from 53% in 2011 to 51% in 2012. However, the spread between figures for top performing companies and the average narrowed for both FVA and error, highlighting a growing challenge to predict demand in fast-moving markets. Bias for the group remained unchanged at 7% with the consumer packaged goods industry consistently over-forecasting for every year encompassed in the study. Other takeaways from this year include:
- Rise in network complexity was fueled by an annual 10% growth in items since 2010.
- Upswing in seasonal products added further complexity with item-locations rising nearly 20% in 2012, compared to only 5% for non-seasonal.
- Increased reliance on promotions and new products drove sales, but made forecasting and execution more challenging with elevated error levels and bias 4-5X higher than regular sales.
- FVA for top performing companies was significant – 2X times greater than the average – suggesting opportunities to improve demand planning across the industry.
- Demand Sensing continues to provide consistent value across all velocities and business activities, including new products, promotions, seasonal and regular turn volume.
Demand Sensing performs consistently across all business activities, underscoring the value of new mathematics that automatically adapts to changing market conditions whether in a downturn or recovery.