New products do not have a sales history that can be represented as a time-series. As a result, time-series forecasting models don't work for new products. Forecasting demand for new products requires alternative forecasting models capable of leveraging data, such as product attributes, that do not come as a time-series.