Web analytics in library practice: Exploration of issues
This study examines the use of web analytics in libraries to understand how this tool can be used to interpret users' behavior on the library's website. Evaluating library and information services is important for library management decision-making regarding the quality of library services. Because such decisions-making is time consuming and requires investment of many resources, libraries are constantly looking for efficient approaches that would provide insights into planning and changes to the library services. The study data consist of reports collected by Google analytics on the University of Missouri's (MU) library website and interactive group interviews with the library's usability committee. The poster presents the preliminary findings and discusses the issues of implementing web analytics in a library setting. The study findings would benefit libraries in understanding how web analytics can be used as an evaluation tool for libraries, especially as an addition to the traditional evaluation tools.
Libraries today are competing with availability of instant access to information on the web. In order to attract and retain their patrons libraries make much of their services and content available online. Libraries also face a challenge of dealing with increased knowledge base that on one hand provides the rigor needed, and on the other hand might be limited to use across contexts and cultures (Eldredge, 2006). Studies have been conducted to provide libraries with holistic measures for assessing their services (Nicholson, 2004; Saraf & Mezbah-ul-islam, 2002). These holistic measures ask libraries to look at the user's perspective and incorporate these perspectives into decision making process. There is growing interest into Evidence Based Librarianship (EBL); however, Nicholson (2006a) states that traditional EBL lacks appropriate research articles that librarians can use. Also, the time taken to collect evidence sometimes results in lesser number of publications and hence reduces the power of the traditional EBL. Eldredge (2006) acknowledges the drawbacks and calls for certain fair and truthful practices that should be followed to minimize the downsides of traditional EBL.
Web analytics is a tool that can measure website traffic by constantly capturing online actions of the website visitors (Khoo et al., 2008). Analytic tools provide information about users' navigation behavior, user and page clusters, and possible correlations between web pages and user groups (Eirinaki and Vazirgiannis, 2003). The web analytics packages available today provide increased functionality by presenting data in a visual format (Tyler and Ledford, 2006, p.7; Eirinaki and Vazirgiannis, 2003). Use of web analytics as a tool has been recognized by businesses that seek to improve their internal as well as marketing productivity through an understanding of the user (Jacoby and Luqi, 2007; Sen et al., 2006; Srinivasan et al., 2004). Library and e-commerce websites both want to provide services to the users, or enable the users to fulfill the intended task seamlessly; however, the goals of a library vary from that of the e-commerce websites in terms of their expectations from the visiting users. For instance, an eCommerce website would want to retain users on its website with a goal to convert a visitor to a purchasing consumer who contributes towards the company's revenue goals. On the other hand, libraries want to be instrumental in leading efficiently their users to information. They want their users to quickly find the information they are looking for even if that means a very short visit to the library website.
The data were collected on the University of Missouri's (MU) library website using Google analytics and an ‘interactive group interview’ (Patton, 2001) with six members of the library's usability committee. Google analytics is a free tool provided by Google, and was implemented on the MU library website on March 2007. A preliminary analysis was done of the Google analytics reports of the library and the results were presented to the six members of the library's advisory committee. The preliminary report was based on answering the following questions as guidelines indicated by Kyrnin (n.d.):
Do most people visit your library website at specific times?
Which pages are the most popular?
Do your readers browse more than one page before leaving?
What is the average length of time your readers stay?
Do your readers come from search engines?
What pages are primary exit pages for your site?
Who is linking to your library website?
The presentation also included issues that the researcher faced when analyzing the data, such as same hyperlinks with different labels showed the same visitation percentage. The online catalog link showed ‘nil’ hits, and there were two different links that seemed to be pointing to this catalog. The purpose of the presentation was to stimulate feedback from the library committee about the use and benefits of their Google analytics implementation and to encourage discussion about possible direct and indirect consequences of such an implementation.
Transcripts of the interview were analyzed for themes, at first using open coding followed by axial coding. The researcher went over the data collected multiple times to code each broad category. The second step of the coding involved creating sub-categories for each broad set of categories. The categories and sub-categories were then connected and interpreted with support from the discussions of the interview using Strauss and Corbin's (1990) axial coding. The broad categories identified using open coding were:
content or design (e.g., search terms entered)
service (e.g., problems with access; help with staffing; time of day website in use)
navigation (e.g., problems with access; entry paths of the user)
user behavior (e.g., user activity on website)
library related (e.g. library concerns regarding analytics implementation)
politics and management (e.g., evidence based decisions made in library).
The highlights of web analytics data for library website usage for Spring '08 include:
Use occurs mostly during the course of a semester
Maximum usage between mornings 10 am to afternoon 3 pm
Visit trend is though the same all over but reduces during the latter half of the semester
Data reported involve users from all the library branches
Data also involves staff, faculty and students that access from within the library
Library computers open the library website as default homepage.
The group interview with the members of the usability group revealed that they have used analytics in the past to understand the usefulness of their content. As one of the respondents mentioned “We have used it to see what the higher [rates] are per visit, what else the users are coming through to see, so we could prioritize resources.” The online content of the library is distributed under the different branches of the library and so getting an understanding of the usage of these different sections separately is important. The librarians are also getting useful information from the search terms that are entered on the library website. According to one librarian: “it is been very useful for us to see what terms people are putting in to the search engine on our web page for couple of reasons: one to see the spellings of various databases so we could incorporate the spellings so that it would bring the search directly to what they are looking for.”
The library would like to operate on a service centered paradigm, and would like to ensure smooth access of their resources. The librarians would like to know if the visitors are well equipped to avail the library's online features and would also like to manage their staffing efficiently using proper supporting data. However, there are some concerns about using analytics because of a lack of continuity in the data reported.
In terms of navigations, libraries would like to be know the issues that their online users face and would like to improve users' navigation experience by making changes on the website – “We have a pretty good idea that there are certain pages that just got people dead in their tracks but the problem is those are not on the gateway. Those are inside [the website].” Apart from this they were keen to know how users are actually navigating to their website either by Googling or bookmarking the link. Also, libraries have internal and external users whose behavior they would be interested to know, as stated by one respondent – “Whether they [online users] are coming from library staff machines or people coming from outside the library or it kind of say you want to look at totals. Also you want to look at totals; you also want to look at library staff machines, on-campus view, [and] off-campus view.”
Web analytics can be used to implement various management level decisions. One participant stated that if in case they need to support any decision based on evidence they can do that with the use of analytics. Before, decisions were made without knowledge of actual usage. The librarians also recognize that some political factors influence management decisions.
The preliminary findings from this study indicate that librarians in the study sample recognize the potential of web analytics for gaining useful insight into behavior of their website users and have multiple dimensions of engagement in using it. They are aware that web analytics can improve the quality of services offered to their library users and can add much desired empirical evidence to the process of decision-making. The additional data analysis is still in progress and is expected to address a question about how web analytics tools need to be set up to maximize the quality of information the librarians can obtain about their users' behavior.