We are currently implementing this kind of systematic approach in Intel IT. Our pilot studies this year have determined some best practices and recommendations around how to most effectively utilize user research data for internal IT improvements. We have found that user research must be conducted in a broad, strategic space where it can be fed into the overall direction for a company's IT organization (Figure 3). This is especially effective in IT organizations that are heavily utilizing OTS applications. It is the best place for a company to ensure that user concerns are considered in long-term planning processes, and it ensures that the research will not be constrained by strict project timelines or be limited by project scope. It also allows more directed research to be conducted on actual projects and to be used by program managers, business process engineers, human factors engineers, transition change management, or other usage and process experts to guide their strategy and design of specific tools or applications.

As new applications are purchased, configured, and deployed, user data can then be used to optimize OTS deployment at the capability level. For example, examining business and usage considerations together can help project teams determine when modifications to OTS solutions are needed. As the previously mentioned data on modification usage highlights, employees may not have the broad or longer-term perspective to accurately determine which system functionality is critical and which functionality can be adapted through procedure or process changes. Broad and systematic research that looks at usage patterns across time and the entire job context of multiple roles can be used to succinctly identify the most value-added modifications. Additional directed research that is specific to a given capability can then fill in the more specific usage details needed to move forward in development.

Many different corporations are increasingly using ethnography and other types of user research during their product design cycles but they have not been used as frequently to look internally at a company's own employees. While our efforts are somewhat innovative, they borrow heavily from the work that other groups have done at Intel to bring field research into product design. Numerous new Intel products such as the Community PC designed to meet specific user needs in rural India and the China Home Learning PC designed to address specific computer needs in Chinese households are two of the more recent examples of this new focus on people-centric product development.

As such, there are not a lot of case studies yet that detail some of the possible missteps or difficulties one may encounter during efforts to implement a user research effort. We provide some of our key learnings from the past year as a starting point to this topic.

The first thing we learned is that it takes at least two researchers. The best research analysis comes from having at least two researchers involved in any given effort. This reduces bias and improves the quality of the analysis. In addition, if the research is being undertaken at a project level, the researchers need to lead and own the research results and process so that they can structure things to ensure the highest quality of data and research outputs.

Another key lesson we learned from this past year is that it is important to get management buy-in and agreement before starting. The research process and outputs to both participants and management should be clearly articulated before beginning any research effort to establish a shared understanding and vision of the research goals and outputs. This helps to gain access to the users that need to be talked to and ensures the data will reach the right people to be fully utilized to improve decision making. Conducting a few key pilot projects showing the kinds of research results they can expect will be beneficial in highlighting the unique benefits of understanding the contextual nature of how people work.

A third lesson we learned is to talk to a wide variety of people. The users targeted as participants should include both the actual employees as well as managers of groups of employees in the study. This allows a full picture of the organization and its culture and processes. It is also crucial to communicate the results back to both groups for two main reasons: to validate any findings and to enhance understanding of the value and process of user research.

Another lesson we learned is that quality field research and data analysis take time. We have found that it takes approximately 12-16 weeks to run an ethnographic study, longer if including multiple geographic locations or large number of participants. The scheduling and actual field research generally take at least 4-6 weeks and the data analysis and final outputs take another 4-6 weeks or longer depending on the outputs to be created.

A major lesson we learned was to utilize a variety of techniques to make users come alive for decision makers. Providing research results with detailed quantitative numbers as well as a summary analysis will create a clear path from the field research to creating visualizations. Making users come alive for decision makers is important for gaining support and understanding of research findings. The use of user profiles, "Day in the Life," and other narrative techniques help communicate work in an easily digestible format that can be used across an organization.

One final lesson we learned is that the researcher can often provide an overall "research coordinator" perspective to decision makers. By working with market researchers and other people who have conducted research on a particular area, an overall analysis can be provided that builds on previous research and provides one research summary to decision makers. This type of strategy helps them avoid dealing with fragmented pieces of user data and enables them to make better, faster decisions.

Figure 3: User research approach
click image for larger view