Impact of forecast accuracy on the service level in supply chain is being examined widely by researchers. The sales data of the personal goods contains lot of randomness which leads to higher mean absolute percentage error (MAPE) value for the forecast based on them. This study aims at reducing the noise in the sales data available for forecasting using right type of techniques. We identified important factors influencing forecasting and discussed their significance. We used the weighted average based method for deseasonalising demand to reduce the fluctuations and obtain the pattern. Further we applied appropriate forecasting tool to the smoothed data. A significant reduction in noise was obtained. The reduction in Coefficient of Variation was also obtained using the smoothing technique. An overall reduction of 64% was derived in MAPE from raw sales data to smoothed sales data. Further, same methodology was applied to additional data sets available from a leading FMCG Company to validate the results. Similar findings were obtained. This study address the problem of noise in the field data obtained for personal care products’ forecasting. By combining the deseasonlity technique with regression method, a practical tool is made available to supply chain decision makers in the organisation.