Data warehousing is an integral infrastructure for data-driven decision support systems. The main measure of a successful data warehouse is the degree to which it meets its purpose. Unfortunately, traditional requirements analysis methodologies have been inspired by programming concepts, leading to a semantic gap between the intended system and its objectives. To reduce this gap, we propose a software engineering methodology called Tropos, which is founded on i* organizational modeling framework. This approach not only allows us for a deeper understanding of the operational environment, but also the interactions between software and human agents. The methodology is illustrated with the help of a real case study.