A user-centered approach to developing digital collection websites
Over the past two years, the Digital Library Systems and Services department at SUL has developed a user-centered approach to building websites. Our methodology involves early and iterative feedback from the primary audience of SUL’s web resources – academic researchers. The intended result is web applications that help users achieve their research goals while at the same time increasing the efficiency of the software development process (thus, lowering the time to development and the cost).
By way of example, in the fall and winter of 2012/13, DLSS had the formidable task of developing three sophisticated and specialized online collection websites in the span of six months. In addition to successfully completing these projects on-time and on-budget, a secondary goal was to develop a framework for rapidly developing additional collection websites using proven design elements, common patterns of user interaction, and modular code that could be easily reused.
The process briefly described below was first used as part of the main library website redesign, and has been refined and utilized for the development of three digital collection websites that have not yet been launched: the Revs Digital Library, the French Revolution Digital Archive, the Bassi-Veratti Digital Library (all of which are in the engineering phase as of this writing).
After an initial phase of discovery that includes data collection and functional reviews of similar sites, the User Research Analyst interviews likely users of the collections sites. A semi-structured interview protocol is designed beforehand to provide consistency across interviews. The interviews are conducted in person when possible, but phone interviews are also used with researchers at other institutions. Interviews typically last between thirty and sixty minutes, and are recorded, transcribed and coded. The interviews typically contain a contextual element in which interviewees access currently used collection websites, carry out self-selected, typical actions and discuss the strengths and problems of these sites. The goal is to document what users do, as well as what they say they do.
The interview process attempts to learn about the scholar’s research interests, their general research process, their goals in working with library collections both physical and digital, how they currently work with digital collections, desired features and functions, and how they see digital collections supporting their research in the future. The interview protocol gives the interviewer and subject flexibility to take the conversation in many directions.
As part of our process to build three upcoming digital collection sites, the user research team interviewed twenty-eight researchers between September 13, 2012 and January 25, 2013. Among the twenty-eight subjects, we interviewed thirteen faculty, nine graduate students, one undergraduate, three library staff, and two external stakeholders. Nineteen of the twenty-eight (67%) interview subjects are affiliated with Stanford. Others are affiliated with partner institutions with faculty that have a research interest in the collections. Interview subjects were nominated by project stakeholders, and by the interview subjects themselves (for example, faculty who were interviewed nominated colleagues or graduate students).
Once the interviews are transcribed, the user research team will analyze the transcripts, identify relevant quotes and produce gists or summaries of the interview. Along with other data collected during the discovery phase, the interviews are used to create detailed personas.
"A persona is a character sketch that represents a particular segment of the target audience…With personas website designers can focus on how the website will be used instead of how the technology should work. Instead of asking how a feature should work, the designer can ask, 'What would Francis do?'" (Mulder, 2006) Personas are in-depth character sketches, which not only describe research goals and behaviors, but also some personal characteristics. They are composite sketches derived from patterns seen in multiple interviews. They are not, and should not, be characterizations of a single scholar.
Our goal in DLSS is to create generic personas that can be reused across projects. This is because the research process used by a scholar may include many SUL web resources, including the library homepage, SearchWorks, collection-specific sites and SDR. At the same time, it is important to acknowledge specialization and the fact that some online collections include unique target audiences. For this reason, the set of personas is evolving to include some that are reusable across collection websites, and some that are site specific.
Personas are a reference that are consulted throughout the design and development process to validate that the site is being built with the target audience in mind.
Information Architecture & Requirements
The Personas have become the foundational artifact of our design process. However, depending on the project, the design team will develop a variety of other artifacts to support both interaction and technical design.
Information Architecture documents describe the organization of content and functions on a website. These can be represented many ways, but most often by block diagrams visualizing the relationship between site pages, content types and functional activities.
Requirements documents are useful summary lists of features and functions that typically serve as a higher-level overview of what the site will do. Our design process often produces a variation on the MoSCoW list, defined must-have, should-have, and nice-to-have lists of features and functions.
Wireframes are sketches of virtually every critical page or page type defined by the information architecture. These can be image-only mockups or interactive HTML mockups that demonstrate some of the user interaction. Wireframes are often the first opportunity both the engineering team and stakeholders can get a real sense of what a site is going to look like and how it is going to behave.
Once wireframes are produced and refined they are shared with stakeholders and often many of the original interview subjects to validate the design and expectations for the site. This becomes an iterative feedback process until the design is far enough along to start engineering.
In our process we make a strong distinction between interaction design and visual design. Visual design is limited to color palette, font, decoration, branding, imagery and iconography. Whereas information architecture, layout, navigation, and content organization are part of interaction design.
In the SUL web environment we are attempting to achieve continuity in interaction design so that both design and technology can be reused from collection site to collection site. This creates efficiencies in the production process, and helps users gain familiarity with the environment as they work in different collections within the SUL web space.
At the same time it is important that specialized collections, especially those that are collaborative in nature, have a distinctive look and feel that resonates with the content and goals of the site, and are appropriately branded.