Notes from MW11 Discussion Forum: "Designing Around a Single Search: How Discovery Layers Are Changing Library Websites"
Designing Around a Single Search: How Discovery Layers Are Changing Library Websites
ALA 2011 Midwinter Meeting, San Diego
Hosted by MARS Local Systems & Services
Sunday, January 9th, from 10:30 a.m. to 12:00 noon
Discussion Forum Notes
MARS Local Systems & Services committee chair Amy Fry welcomed the audience of around 130 attendees and introduced the three panelists.
Josh Boyer, North Carolina State University, Summon and locally-developed QuickSearch
Beth Bernhardt, University of North Carolina, Greensboro, EBSCO Discovery Service
Johan Oberg, Macalester College, WorldCat Local
Josh Boyer from NCSU (http://www.lib.ncsu.edu/) led off the discussion.
The Libraries of North Carolina State University were awarded a prestigious ALA OITP “Cutting Edge Technology Service Award” following a website redesign project between January and August of 2010. That redesign incorporated a locally-created tool called QuickSearch, which displays article, catalog, library website, and journal list results on a single search result page, along with a “Best Bets” category of locally-relevant resources. Article results are compiled from multiple library databases using ProQuest Summon.
This search and display interface features tabbed searching, which NCSU decided upon based on their hesitation to simply place a single search box on their website. A review of ARL libraries also indicated tabbed searching as a popular method. NCSU wondered whether searchers would choose a tab or use the default and whether a tab, if chosen, would be the most appropriate one.
User testing was completed in a “guerilla” fashion, grabbing students in the library for quick testing on the spot, rather than using a more formal approach. Over half of testers selected a tab (which surprised the researchers), and of those that did select a tab, ¾ selected the most appropriate one for the search at hand. However, users rarely switched tabs from the results page. Searches that tended to work in testing included finding a book by title, finding a recent issue of a journal, and finding the cost for making copies at the library. Problem areas included distinguishing between journals and articles and databases, finding journal articles by subject, scrolling on the results page to all results, and spelling inconsistencies.
Usage tracking of the site finds that 43% of use is given to article exploration, 36% to catalog records, 2% to library website research, and 8% to the Best Bets category. Between August 2009 and August 2010, the library website saw an increase of 17% in traffic overall.
Future usability testing will separate testing for students and faculty. The faculty testing will be more open ended interviewing as opposed to testing a list of tasks.
Beth Bernhardt from UNC Greensboro (http://library.uncg.edu/) was the next speaker.
Historically, students at UNC Greensboro have avoided using the library catalog because of its unappealing interface; students also expressed a preference for a ‘search’ option over use of the A-Z database list. EBSCO Discovery Service provides a much improved look and feel for the catalog. The decision to implement EBSCO Discovery Service was made by a Systems Advisory Committee made up of library public services, technical services, and systems members. The committee trialed both EBSCO Discovery and ProQuest Summon, but chose EBSCO in part because North Carolina has numerous EBSCO databases already in use.
The EBSCO Discovery Service search box is included on the library home page, database page, and on the library’s mobile site, as well as within numerous LibGuides. This takes advantage of the service’s ability to create topic-focused search boxes, highlighting business resources, for example, or health information.
A literature review substantiated the move to a discovery system. Discovery systems tend to be popular with all groups of library users (though graduate students tend to like them less than undergrads and users in general like them more than librarians do); they tend to appear to be easy to use; users like the single search box; and discovery systems seem to increase resource usage.
At UNC Greensboro, about 50% of librarians report using EBSCO Discovery Service with students at the reference desk. The library has seen an increase in the use of collections and databases, and is also tracking an increase in patron-driven acquisitions.
Johan Oberg from Macalester College (http://www.macalester.edu/library/) was the third speaker.
Macalester College uses a shared catalog with other institutions in a consortium, but determined that WorldCat Local would provide a better search experience for their students. WorldCat Local offers a simple search interface that provides access to a broad research experience with content beyond what is owned by a single library. From the WorldCat Local search box, a student may select between local, state, and world content.
Unlike NCSU, Macalester made a decision to reject the tabbed interface model in favor of a single search box, prominently placed on the library home page. This prominent placement has resulted in increased usage (usage far outpacing that at a research university 3x the size of Macalester that places the search box on a secondary page). Placed at the top of the page, the simple search box allows for the devotion of home page real estate to library events and other items of interest.
The library is still unsure what to call WorldCat Local, though students refer to it simply as “Library Search” and have quickly adopted it. The word ‘catalog’ is seldom used by anyone. Faculty have also responded positively, citing the time-saving advantages of the service. Usage has increased 30% since initial implementation in 2008, as has interlibrary loan and overall circulation. The takeaway lesson was ‘Be bold!’ and put the new search front and center!
In the future, Macalester looks toward implementation of a faceted database page, increasing mobile access, highlighting digital content, exploring web scale management, and taking advantage of the WorldCat API to reach students where they are.
A question and answer period ensued.
Q: How did you decide to “brand” your discovery service? Using the vendor name vs. not?
A: Boyer (NCSU) Discovery systems tend not to require vendor branding and instead allow an institution to label by material type – for example “books” or “articles.”
Q: Any thoughts for tracking and recording the “value” of discovery systems beyond full text article views?
A: Bernhardt (UNCG) and Boyer (NCSU) Systems allow for tracking, but comparing data from database to database is like opening a can of worms. More work needs to be done in this area.
Q: Speak to databases that are not included in the indexes of discovery systems? What happens to those, which are often specialized and topic-focused?
A: Bernhardt, Boyer, and Oberg (Macalester) Faculty likely already are aware of specialized databases in their fields. Use LibGuides to drive students to specific databases. Discovery systems perhaps offer most value to undergrads, looking for x number of articles on a topic. Perhaps use discovery system as aggregator of content from several large databases, then trying to push students to database pages. Perhaps use discovery system as starting place – the default place to begin research.
Q: What enhancements are you looking forward to from discovery system vendors?
A: Oberg and Boyer Web-scale for OCLC, APIs to use on home page for better display and visual browsing of material. Search is currently in a primitive state when compared to Google – Google knows that someone searching for a zip code likely wants a map, or that someone searching for “Huckleberry Finn” probably wants the book rather than a book review – discovery search not at that level yet. Search relevancy rankings are not good enough – must use result facets to take advantage of databases and discovery systems – do students use those facets???
Q: What are “Do”s and “Don’t”s for discovery systems and website design?
A: Oberg, Boyer, and Bernhardt Placement on the site is key to driving usage. Get as many people involved with the process to improve “buy-in.” When considering customization of discovery services, be aware of technical time requirements. Oberg suggested looking at the University of Michigan site as an example of using the Summon API with other tools (eg Drupal) to build an interface that is completely customized. Programming time for these custom looks can be significant.