This document discusses using interactive machine learning for sales forecasting and "what-if" simulations at Electrolux, a fast moving consumer goods manufacturer. It highlights the importance of forecasting for marketing, manufacturing, and warehousing. It then describes Electrolux's solution which combines multiple data sources, probabilistic forecasting models at scale, and intuitive analyst interaction to define and optimize promotional campaigns. This has resulted in up to triple the forecast accuracy for promotional sales compared to baseline manual methods.