I have time series for product usage over an year on daily basis. Product usage exhibits seasonality i.e. it usage increase/decreases by more than the normal usage during that time. When i get the current day's data and its more than normal, i want to check if its exhibits seasonality or not and take some actions accordingly. I am thinking of using Double Exponential Smoothing to achieve this. Is this an right approach or is there an better approach to do it.
A useful introductory reference is Forecasting: Principles and Practice by Hyndman and Athanasopoulos. Depending on your data, you might be able to build a simple additive or multiplicative model with seasonal, trend, and cyclic components.
Besides exponential smoothing, you can look at ARIMA models. Exponential smoothing is based on trend and seasonality while ARIMA is based on autocorrelations in your data.