The visual creation tools in the mashup frameworks are supposed to be simple and accessible. Yet at the same time there is a need to extend the capabilities and complexity of mashups. Therefore, in practice, the frameworks become increasingly hard to learn for a casual user. In relation to the emerging mashup creation techniques in Semantic Web area, in our work we propose an idea to split the development of mashups into two stages: data-level and service-level. Each managed by a separate, although well integrated, environments.