In 2004 Fujitsu asked PARC to carry out an ethnographic investigation of their software business, focusing on their development processes, and while doing so to build an ethnographic capability in their own organization. One of our biggest challenges was to convince Fujitsu's system engineers — and the development organization more generally — of the value of ethnography for their business. They are used to translating what they hear from customers about the workflow into a standard framework of system requirements and specifications; it was difficult for them to see the relevance of putting any significant focus on understanding what is going on in the workplace at the level of everyday work practices. Moreover, in their work with customers, system engineers commonly proceed in a carefully planned and highly structured manner, where every activity is expected to yield predictable outcomes. For them, the open-ended nature of ethnographic fieldwork seemed dangerously chaotic and unpredictable. The lessons learned from our experience with both our initial teaching of the engineers and the organizational fieldwork we later did together helped us design a new ethnography-training course that incorporates the task of conveying the full value of organizational and business ethnography.