We propose a framework to accurately forecast Chinese tourism demand. Search engine query data is collected to forecast tourist volumes to Beijing. A generalized dynamic factor model is used to create a composite search index. Our method improves forecast accuracy better than two benchmark models.