Cloud-based workflow tools suit well for supporting dynamic and cross-organizational business collaborations that can be commonly found in scenarios such as emergency management, supply chain management, healthcare management, etc. However, in these scenarios, a large amount of concurrent workflow instances could involve a great deal of data. When one task needs data located in different places, especially for private data located in client, the data transfer becomes a challenge. To effectively place these data, the workflow manager must intelligently select the place in which these data including private data will reside. In this paper, an approach to data placement for workflow in cloud is proposed, which places the data on cloud-side or client-side according to the privacy requirements, and adjusts to the data placement according to the control flow to minimize data transfers. By integrating the approach into a workflow system, we can expect improvement in data transfer. Experiments show that the proposed approach can effectively reduce data transfers during the workflows execution.