Empirical research suggests that quantitatively derived forecasts are very frequently
judgementally adjusted. Nevertheless, little work has been conducted to evaluate the
performance of these judgemental adjustments in a practical demand/sales context. In
addition, the relevant analysis does not distinguish between slow and fast moving
items. Currently, there are neither conceptual developments nor empirical evidence on
the issue of integrating judgements and statistical forecasts for slow/intermittent
demand items. Moreover, no results have ever been reported on the stock control
implications of these human judgements. Our work analyses monthly intermittent
demand forecasts for the UK branch of a major international pharmaceutical company.
The company relies upon a commercially available statistical forecasting system to
produce forecasts that are subsequently judgementally adjusted based on marketing
intelligence gathered by the company forecasters. The benefits of the intervention are
evaluated by comparing the actual sales to system and final forecasts using both
forecast accuracy and inventory control (accuracy implication) metrics. Our study
allows insights to be gained on potential improvements to intermittent demand
forecasting processes and, subsequently, the design effectiveness of forecasting support
systems.