Prediction Limits

Morning everyone. First thread, so take it easy on me.

I am using a holt-winters times series to predict future call and transactional volume. In addition to the forecast, I would like to provide a prediction interval or range. Looking at previous forecast deviations from actual results, the distribution is normal. My first thought was to calculate the standard deviation of my previous forecast deviations and incorporate that value along with the properties of the normal curve to build out the range. For example, if my forecast for September is 69,000 calls and the standard deviation of the differences from previous forecasts is 12,687, I would expect that 68% of my call volume would fall between 57116 and 82490.

Is this accurate? If not, any recommendations on how to build out the prediction interval or range. Thanks for any assistance.