Sometimes called ethnography, observing work and behavior in its real environment can collect essential insights into why, how, when, where and with whom an action or an item is significant.  My work in this area has focused most recently on the workflow of health care providers in the outpatient and surgical areas, to understand the role of technology - specifically software and computers - in enhancing the patient care experience. 

Method

It sounds easy, doesn't it -- just go watch someone then come back and tell us what they do, why and how.  It's a hard method to explain because it requires a deep empathy and listening state of mind.  It's easier to refer to work goals accomplished using this method:

Case #1: I have conducted studies on shoppers in the retail environment, whose use of a shopping kiosk enabled them to enhance their shopping decision-making process, and for us (and my client) to gather a lot of critical functionality to expand the role of that kiosk and that retailer in the customer experience. 

Case #2: For a museum client, I documented the behaviors and intellectual requirements for museum visitors who used a beta PC tablet during their tour to access the diverse and exhaustive archives of the museum in that "moving image" category.  In each case, the users were seen posing questions to technology and needing triggers from the technology to know what to ask. 

Results

User requirements gathered from such studies also benefits the design process in refreshing the vocabulary to reflect the intuitive language of the customer.  Scenarios best capture the dynamic of the user, their fellow users, the context of their usage, and the supporting materials and technologies, in producing the desired outcome.  Vocabulary unique and compelling to that audience is collected for reference as a core taxonomy during the design process.

Timeframe and preparation

Setting up the study (defining, finding and scheduling appropriate users) takes the most time.  The observations can last anywhere from 3-5 hours, sometimes supplemented with journals or pager-triggered "where were you now" techniques that can extend the participation of the user to several weeks.  I have typically not asked users to extend their involvement past the moment of their activity so that there is no constraining actual behaviors and intents.