The discovery environment is often frustrating for all stakeholders. Tools are fragmented. Metadata is insufficient. Authentication and interface design issues block access to resources even when one has rights to them. Even Google and Google Scholar, often cited as favorites of researchers because of ease of use and relative success in identifying full-text, are limited by questions of coverage and limited transparency in algorithms for search and results ranking.
Publishers have the particular challenge of seeking to provide a positive user experience while having to admit that no publisher platform is a discovery service and thus need to invest in mechanisms to ensure their content is appropriately represented and ranked in both open web and library discovery platforms. And, although there are substantial reasons for a library to take an interest in discovery, not the least of which is ensuring that users find the resources that the library invests in, like publishers, librarians know that competition for user loyalty at the network level is fierce. Library discovery can only succeed in reaching a high market share of researchers from a campus if it is intensely user-centric and provides value different than that which is found on the open web. To address these issues at the University of Illinois at Urbana-Champaign, we developed a set of evidence-based principles for making library discovery user-centric.
Centering on the User Community
Developing user-centric systems is challenging. Users do not always know what they want and they do not always do what they think they do. Users may also change their practices over time. And, particularly vexing, knowing about users does not reveal the needs and wants of non-users. Users are also not a global monolith, uniform regardless of local priorities, pressures, and practices. One community of users may be very different than another, which is also different than a third, etc. Any given library serves a particular community of users. Being user-centric means serving that community of users and prioritizing its needs, regardless of general trends or fashion.
It is often tempting to begin by asking users what they do and do not like about existing systems or possible replacements; however, this is problematic. Asking about current and possible systems centers our thinking on product selection and presupposes a need for a particular kind of discovery environment. Instead, we need to design assessment and evaluation processes that focus on what users actually do and the choices that they make as they do their work in order to reveal the actual behaviors and desires in the user community. We need to bring the user to the foreground of our library discovery decision-making. Only then can we design a discovery environment that meets the needs of the user community and earns their loyalty.
This challenge was particularly acute for us at the University of Illinois at Urbana-Champaign in the early 2000s as the Y2K crisis had forced a migration to a new integrated library system that drew higher than typical levels of complaints about usability. In addition, the burgeoning growth in subscriptions to e-journals as well as emerging backfile digitization projects made clear that radical changes in discovery and access were on the horizon. A working group was charged to develop the University Library’s access strategy and that group has existed since then with myself as a leader or member more or less continuously, though with a variety of names and configurations.
In the remainder of this post, I will fast forward in time and focus on the evidence-based development of the user-centric principles that are guiding our library discovery environment. Though I will share our principles, I want to caution against any temptation to apply our principles to other user communities without a process of local testing and validation.
Characterizing the Illinois Context
At a most basic level, we know our discovery environment must be robust enough to accommodate our extensive collections as well as the heavy use that they receive. We have an annual collection budget of over $16 million, more than 125,000 online journals, over 800 databases, and just celebrated the acquisition of our 14 millionth volume. The collection includes materials in essentially all known formats and languages and altogether the collections are comprised of more than 40 million items. Our almost 60,000 member user community also makes voracious use of the collections. Each year there are approximately 3 million visits to library facilities, 8 million downloads of content, 1.3 million downloads from the institutional repository, and 1 million circulation transactions. No single platform can currently accommodate all of these materials, their related metadata, and the usage levels from our community.
These statistics alone underscore my claim that principles for a discovery environment need to reflect the context and needs of a user community. Comparative analysis of user survey data further supports this point and indicates the strategic importance of the discovery environment for the University Library. Illinois faculty continue to place greater importance on the role of the library as a gateway to information and express greater support for the role of librarians and institutional investments in the library as compared with national averages. Meeting user expectations for discovery is critical to maintaining the loyalty of the user community to the library.
The collections context and strategic priority for discovery led to exploring multiple commercial discovery options, including WebFeat (2005-2006) and Primo (2011-2014), as well as to developing and deploying Easy Search, a locally-developed and supported broadcast search, search assistance, and recommender system that incorporates commercial discovery systems as targets and data sources (2006-present).
Task Taxonomy and User-Centric Principles
Guiding our discovery work is a taxonomy of user tasks and a set of user-centric principles that emerged over time. The taxonomy and principles were developed from an analysis of a decade of user surveys and over nine years of Easy Search Transaction Logs. Each of the user surveys was analyzed for items that related to discovery and user workflow and data from those items, as well as open-ended comments, were coded for themes and trends. Reports from Easy Search Transaction Logs include descriptive data (e.g., query word length and queries/session) as well as sampling of search sessions that were coded for query type (e.g., known item/topical investigation) and pathway analysis (e.g., which targets did users choose and in what sequence). We have also conducted intermittent surveys, focus groups, and usability studies for Easy Search and other discovery systems.
The taxonomy of user tasks that our discovery environment must support is:
- Locate known item
- Locate known research tool
- Explore topic
- Identify and access library tools/databases for topic
- Identify and access research data and tools
- Identify assistance/library staff
The user-centric principles for discovery as a service are:
- Library users want “my everything” and not absolutely everything. Serendipity is valued but out-of-scope results and recommendations frustrate users and decrease trust in the system. A discovery service must provide relevant results and not extraneous ones – and users must trust the system to do so – if users are going to rely on the library discovery service.
- Enable full library discovery (content, services, and spaces). While most users are searching for known items or topical content, the transaction logs indicated that they search for other library-related topics. In addition, surveys and focus groups revealed that users value learning about complementary services (e.g., liaison librarians) and spaces (e.g., library hours) as context for their topic searches.
- Everything owned, licensed, or otherwise provided should be discoverable. Specialized digital collections in particular often lack adequate metadata or necessary “hooks” to be search targets in a discovery environment. Ensuring that such databases do not “disappear” because they are not integrated in the discovery environment is a critical aspect of full content discovery. When these are locally developed databases, local investments in building out these systems are leveraged into greater impact when they can then be integrated into the discovery environment. For licensed resources, we are increasingly assessing for how — and how well — resources can be brought into the discovery system.
- Don’t mislead. The discovery system should make clear what is not included and must be searched for elsewhere. For this reason alone, we avoided using phrases like “single search” or “one search” in naming and describing Easy Search.
- Decrease the steps from discovery to delivery. Every extra step between the user discovering a desired item and the library delivering it is a potential stumbling block. In reviewing the transaction logs, so many times we would see that a user got within a step or two of retrieving the content it appeared they were seeking but then took a wrong turn or encountered an unexpected barrier. If a link to full-text is in the full record, the discovery system should present that link prominently in the initial display. Discovery should be delivery.
- Utilize consistent labeling that reflects user terminology. Publishers and librarians use specialized and technical vocabularies that provide a great deal of precision; however, such terminology can be opaque and confusing for users. Adopting user language for labeling and using that terminology consistently enables more effective and efficient use of the discovery system.
- Implement adaptive and contextual assistance. Though log files show that users rarely consult general help files, we observed that they would look for assistance in context, exploring multiple paths when faced with unknown results. Thinking carefully about what information is useful in context means providing just-in-time assistance rather than just-in-case and keeping the interface relatively clean and uncluttered of unnecessary detail. As appropriate, offering well-labeled links to relevant LibGuides and contact information for library staff also provides in-the-flow discovery support.
- Always offer a next step. The discovery system should never “fail out” but instead should always provide a pathway forward. The sense of success and self-sufficiency that open web searching provides is often lacking in library systems. Offering support for spelling corrections, suggesting title match options for lengthy keyword entries, and normalizing DOI formatting are all examples of taking user inputted search strings and building on them rather than having the system respond with an error message.
The taxonomy of user tasks and these user-centric principles serve as the yardsticks against which we benchmark our discovery environment. Though these principles are not yet fully realized in our discovery environment, and will likely always be a work-in-progress as information systems and user workflows change, they serve as the touchpoints for our discussions and decision-making and are the foundation for continuing to pursue evidence-based understandings of the needs and practices of our user community. We look to publishers and platforms to partner with us in developing transparent technologies and workflows that enable this vision, directed by the needs of our user community and responding to the challenges we face. By articulating the taxonomy and principles, we have a basis for a shared understanding about the goals of our discovery environment and our decision-making processes.
Note: I want to thank my colleagues Bill Mischo and Michael Norman, with whom I’ve had the pleasure of co-leading a number of discovery-related projects and who encouraged me to formalize our principles, as well as our many colleagues who have worked with us on our numerous task forces and committees. I also want to acknowledge Sarah Hare, now Scholarly Communication Librarian at Indiana University, who was instrumental in the user survey analysis when she was my graduate assistant at the University Library. Further details about the design of Easy Search were recently presented by my colleagues at the CNI 2017 fall meeting.