The large number of available prices gave rise to price comparison sites that yet lack recommendation services to support their customers in scheduling buying decisions. Using a large data set with 1.46 million daily minimum price observations for four product categories of e lectronic consumer goods, we outline reasons for the slow adoption of recommendation services, evaluate 6.56 billion price forecasts and show that forecasted product prices can be used to build recommendation services to advise customers on their purchase time decision. We compare 16 different methods that can act as the core of price prediction services and give detailed insights on performance as well as advice on model selection for short-, medium and long-term forecasting horizons, outline differences b y category and advocate the transition towards prescriptive price analytics services.