Leveraging Comparables for New Product Sales Forecasting

. Sales forecasting is a central activity of operations managers in most retailers. Dependent on the industry, forecasting sales can be a hard problem. Working together with two large industry partners, we find that predicting sales for new products is especially difficult. Whereas standard forecasting models use past sales data to predict on the short term, predictions for new products need to be made far in advance without any sales history. As a result, many companies resort to qualitative methods. [1] suggests that surveys, expert opinions, and looking at average sales for similar products are the most widespread techniques for predicting demand of new products. Unfortunately, according to [1] only 10% of companies make use of some form of analytics. At the same time only 20% of companies were satisfied with their new product forecasting. Altogether, the current state of affairs provides opportunities to improve the new product sales forecasting process significantly. Forecasts of new product sales are important, because they guide many decisions that operations managers have to make before and during a new product launch. Decisions guided by these forecasts span the whole range of operations: capacity planning, procurement, production scheduling, inventory control, distribution planning, marketing promotions and pricing. In the end, the goal is to forecast more accurately. This leads to better operations decisions that will increase the chances of a successful product launch.