Analysing inaccurate judgemental sales forecasts

This paper deals with categorising errors that exist in qualitative sales forecasts, so that it can be defined what kind of development is needed to improve forecast accuracy. A framework for pointing out different types of sales forecast errors is presented. The framework includes analysing demand profiles of customers and the continuity of under-/over-forecast errors. The error types are named as random error, positive bidirectional error, negative bidirectional error, systematic under/over estimation error and unforecasted sales. The differences between the approaches for reducing each type of error are explained. The use of the framework is illustrated with sales and forecast data of a large process industry company. The analysis steps are illustrated and actions for reducing different types of sales forecast errors are suggested. [Received 15 November 2006; Accepted 11 May 2007]