2011 Forecasting Performance Benchmark Study

Overview

Member companies in the Forecasting Performance Benchmark Consortium include Campbell Soup, ConAgra Foods, General Mills, Kimberly-Clark, Kraft Foods, Procter & Gamble, Unilever and two other consumer packaged goods companies. This report includes virtually all items and locations for the North American warehouse-delivered businesses of these nine multi-national firms. The scope of the 2011 report encompasses more than 7 billion physical cases, represents more than $200 billion in sales over two years, and accounts for more than one third of the North American consumer packaged goods (CPG) market (excluding alcoholic beverages). It includes 300,000 item/warehouse combinations with 55,000 base codes and 72,000 items across 350 locations.

The dataset spans a comprehensive cross section of the CPG market with 70% food, 30% non-food and 25% seasonal, 75% non-seasonal volumes.

The value of a trusted benchmark is that you know it is directly comparable to your performance. Reaching or exceeding it enhances supply chain efficiency and improves your competitive position. The problem with traditional forecast performance benchmarks is that they are not reliable. Pitfalls include self-reporting (which no one can really trust) and differences in metric definitions and methodologies. The net result is non-comparability and many believe that traditional forecast benchmarks are irrelevant. 

This study is unique in that it provides the opportunity for a truly comparable benchmark of forecast error and bias across the industry. What sets this report apart is that all companies participating in the benchmark study have deployed Terra Technology’s Demand Sensing software. Having a common platform of production data (not self-reported), with common data mappings and a consistent unit of measure for all participants overcomes these problems. 

The analysis provides a composite of Demand Planning performance for the North American CPG industry during the calendar years of 2009-2010. Forecast performance is reported using the widely accepted Mean Absolute Percentage Error (MAPE) and bias metrics. Planning performance is contrasted to both Demand Sensing and to a simplistic naïve forecast (seasonally-adjusted moving average). The addition of the naïve forecast provides a method to compare the value of investments in sophisticated forecasting tools such as Demand Planning and Demand Sensing.

Next - Key Findings