By Eddie Neuwirth - Sr. Product Manager, Discovery Services
In the last several years, it has been great to see libraries investing in improving access to library content. This has been accomplished via implementing discovery services, launching website redesigns and mobile applications, and applying new reference services and collection development models. Yet for most libraries the same level of innovation has not been applied to the library’s link resolver, which plays a critical role in process of “delivery” – linking users to the content and resources they’ve discovered. Unfortunately, link resolver technologies have been slow to evolve and, as a result, linking remains one of the biggest stumbling points for today’s researchers.
In our previous post, we discussed how linking to content via the library link resolver is key to a successful discovery experience. Failed links have real consequences for libraries, including user frustration, and, even worse, user abandonment of library resources altogether. Despite these consequences, most libraries are employing linking tools reliant solely on technology (OpenURL-based linking) that is roughly 12 years old.
Meanwhile, accompanying link resolver interfaces have not kept pace with expectations for the ways that modern library users interact with online content. So as libraries look to improve discovery to compete with Google and other online services for users’ attention, they must make sure that they are not only focused on improved discovery engines and services, but improving the linking experience for their users as well.
Link Reliability and User Success
OpenURL link reliability (or lack thereof) is a well-documented issue for libraries, so it is a bit surprising that libraries do not always evaluate linking and link resolvers with the same scrutiny they do with other aspects of discovery. Various research studies on the issue of reliability of OpenURL links have concluded that OpenURL links regularly fail between 5% and 30% of the time. 
Jason Price and Cindi Trainor provided an analysis of OpenURL reliability of 360 Link and Ex Libris’ SFX, the two most popular link resolvers in use by libraries. Price and Trainor found a 7% error rate for OpenURL links in 360 Link and a 30% error rate for users of SFX. 360 Link also exhibited higher reliability than EBSCO’s link resolver during tests conducted as part of the NISO IOTA project and reviews by Marshall Breeding.  Regardless of which link resolver outperforms the other, the fact remains that link resolver error rates are generally too high to meet users’ expectations. If one in 4 or one in 15 Google links failed, users would likely lose some confidence in Google, too.
There are many common causes for OpenURL link failures, including: incomplete or missing source metadata; knowledgebase errors; and a given link target’s ability to work well with inbound links. Libraries that want more reliable OpenURL linking should work with a link resolver vendor who is vigilant with updating metadata, maintaining relationships with publishers and database providers, proactively testing links, and providing ongoing support to customers. According to the 2012 study by Breeding (commissioned by the National Library of Sweden), 360 Link ranked highest in customer satisfaction among commercial link resolver vendors as well as ranking highest in accuracy and comprehensive of the knowledgebase. 
In the end, a link resolver is only as good as the knowledgebase and link connections that inform it. And, in order to achieve the highest levels of linking reliability, ProQuest employs a dedicated team of linking and content experts who work to address the continuous volatility in OpenURL-based linkers. The team also ensures that the link resolver is working with an authoritative, enriched, up-to-date and centrally provisioned knowledgebase. The 360 Link team pushes changes, corrections and additions out to our customers via a seamless SaaS model with no effort required by the library.
For example, in 2013 alone, our team had to add 70 new link targets, which impacted links to more than 1,000 databases. This does not take into account the updates required to maintain linkers working for more than 5,000 databases, which is an unrealistic workload for a typical library staff to keep up with.
However, despite its strengths, 360 Link is not immune to general challenges facing all link resolvers, including links that don’t always work, and link resolvers often are a point of confusion for end users.
User Experience Determines Linking Success
Despite the work done by the 360 Link team and ProQuest data experts to provide a superior link resolver experience for libraries, additional research suggests that reliability of links served up by link resolvers is really only half the challenge when it comes to providing a positive linking experience for users. Numerous research studies focusing on the usability of link resolver “menus” – the intermediary page presented to users after they click on a desired article – have determined that link resolvers too frequently do not get users to the content desired, even when the link displayed is working correctly. Put another way, even with reliable links present, the library link resolver can still be a failure point.
Bonnie Imler and Michelle Eichenberger studied undergraduate interaction with the SFX link resolver menu at Penn State University, and found that nearly 25% of users who visited the SFX link resolver menu page never actually clicked on the appropriate link that was supposed to direct them to the desired article.  It’s worth noting that the SFX user interface experience this research evaluated has not changed much in the 12 years the product has been in libraries.
Earlier research on SFX by Jina Choi Waikimoto, David Walker and Katherine Dabbour at the California State University system found that out of 188,000 SFX menu uses, 48% of the time (nearly 90,000 times!) users closed the menu window without clicking anything.  This research suggested that low link resolver click-through rates could be attributed to the simple fact that presenting users with an intermediary link menu goes against their expectations by forcing them to take extra steps to get to the full text. Yet this is how most libraries ask their users to navigate to content.
To address the issue of extra clicks between end users and the full-text content they are after, 360 Link introduced its pioneering “one-click” feature in 2006, which gives libraries the opportunity to bypass the link menu and lets the OpenURL linker work invisibly behind the scenes to the user. Recently copied by some link resolver vendors, this approach provides users with the illusion of a direct path to an article, but is only partially effective due to the inherent unreliability of OpenURL linking and other usability factors. Failed Open URL links and usability issues, when coupled with a one-click feature, can frustrate users by leaving them at dead ends.
Suspect reliability and other library preferences make some libraries uncomfortable with a one-click approach to linking. But whether one-click linking features are employed or not, most library users still must navigate some sort of hybrid pathway to get to the full text they desire – sometimes they are taken directly to content, sometimes they are directed to a link resolver menu, and sometimes they are left at a dead-end. It is obvious that link resolver menus must become more intuitive and seamless for users to minimize confusion and failure points.
The ideal link resolver experience is one that offers the most direct and reliable access to full text, yet also offers additional, intuitive pathways to full text for users, in the event the links fail. In this context, the ideal link resolver menu must be highly customizable to provide libraries with opportunities to promote other library services including direct help for end users at the very point they need the help the most.
In the next month we will be introducing significant innovations for 360 Link, designed to eliminate the most common barriers to successfully linking users to library content. Stay tuned for an upcoming announcement and further details about these advancements and, in our next blog post, we will focus on what we’re doing at ProQuest to help libraries deliver the most reliable and modern linking experience for end users.
 Jason S. Price and Cindi Trainor. "Digging into the Data:Exposing the Causes of Resolver Failure." Library Technology Reports 46.7 (October 2010): 15-26.
,  Marshall Breeding. “Knowledge Base and Link Resolver Study.” National Library of Sweden. (2012) http://184.108.40.206:11349/dokument/Knowledgebase_linkresolver_study.pdf
 Bonnie Imler and Michelle Eichenberger. “Do they ‘Get it?’ Student Usage of SFX Citation Linking Software.” College and Research Libraries. (September 2011) 454-463.
 Jina Choi Waikimoto, David S. Walker and Katherine S. Dabbour. “The Myths and Realities of SFX in Academic Libraries.” Journal of Academic Librarianship 32. no.2. (2006) 127-36.