IT service delivery is challenging to study. It is characterized by interacting systems of technology, people, and organizations. The work is sometimes reactive, sometimes carefully planned, often risky, and always complex and collaborative. In this paper we describe how we’ve learned about IT work, using a variety of methods including naturalistic observations, contextual interviews, surveys, and diary studies. We provide examples of our study results, showing what we’ve learned with the different methods. We argue that to effectively study such systems, a variety of methods may be needed to complement insights and validate findings. We found that naturalistic observations were extremely time and labor intensive, yet offered us the time and space to observe unplanned events and long-lasting tasks, bringing out the full complexity and risks involved in real work. Contextual interviews and diary studies provided fewer details, yet gave a broader context to individual’s work. Surveys provided an even broader picture, going beyond individual differences, yet they were limited by details and issues of sampling.