Simple Exponential Smoothing (SES) is one of the Time Series Models (TSM) most widely used at industrial level for forescating purposes because it is simple, easy to understand and to implement for many items or families, if the time series actually display a stationary behavior. However, to use SES one must decide the proper value for the smoothing constant (alpha) that drives the behaivor of the forecasting model. In this paper we propose a methodology that helps to find a “good” value of the constant (alpha) for the SES model, in order to obtain a “good” forecasting demand. The paper explains the steps of the methodology proposed and it is applied to 58 monthly series of the industry category with 133 observations that are included in the M3-Competition.