Forecasting sales in the presence of promotional events

Sales promotions are widely used in industry as an effective tool for stimulating sales. Companies increasingly collect data on promotional events but this data is rarely used, neither to estimate real effect of promotional events nor to improve sale forecasts in the presence of events. This paper proposes a procedure for estimating the effects of events on sales and for utilizing the information to improve forecasting accuracy whether future events are planned or not. The paper also proposes a mathematical programming model to optimize the allocation of budget to different types of marketing events and decide timing of the selected events. The proposed procedure for forecasting in the presence of events is tested in the paper. The tests clearly indicate the usefulness of the basic functionality of the proposed procedure. Three different scenarios are solved, each representing different level of knowledge on past and future events. The results emphasize the importance of accounting for both past and future events and show how forecast accuracy can be improved by using the proposed approach.