By some estimates, Google processes over 40,000 searches every second! Can managers use this torrent of search data, often free and fine-grained, to predict the sales of their products? The premise is that a search for certain terms signals an intent to buy specific products.

My research says the answer is yes! Based on working with multiple retailers, managers can expect anywhere from a 3-8% reduction in prediction errors when search information is included in forecasting models.