Proactive chat reference: Getting in the users’ space

Jie Zhang; Nevin Mayer

More frequently we find ourselves approached by live customer service representatives in cyberspace offering customer support on the commercial websites we visit. Live customer support chat offers us the opportunity to ask questions and, ideally, to receive quality information in response to our questions about a product or service we are thinking about buying.1

This proactive approach to customer service, commonly used in business, was recently adopted by Grasselli Library at John Carroll University for its chat reference service, which supports 3,500 FTE students and 380 faculty members. By switching to chat software that initiates the offer to help on behalf of the librarian, we experienced success in increasing the visibility of our chat service with our users. This, in turn, increased our patrons’ use of the service.

Starting with a consortium model

Grasselli Library initiated chat reference service in August 2002 by participating in OhioLINK Chat, a multi-type institutional consortium. This consortium was made up of OhioLINK colleges and universities interested in pooling portions of their librarians’ reference time for statewide chat service. Because we are a Jesuit university, in July 2003, we also started participating in the Association of Jesuit Colleges and Universities (AJCU) 24×7 Virtual Reference Project.

While participating in OhioLINK Chat and AJCU 24/7 Virtual Reference, we encountered some scheduling problems. There were breaks during the year when we were unavailable to participate in the service. We also realized that some of the questions we received could be answered more knowledgeably by the students’ home institution.

In 2006, Grasselli Library reassessed its efforts in providing electronic reference services. We decided to discontinue participating in the AJCU project and to continue only with OhioLINK chat. The reasons for this change were low usage of chat by our patrons and the high participation fees AJCU charged for their chat service.

A year later, OhioLINK Chat service was discontinued in favor of the State Library’s KnowItNow24×7, which expanded chat service for all citizens of Ohio and for anyone residing in the state.2 Because we are a private university, we felt our role in statewide chat had become far less essential, so we decided to discontinue participating in the state’s chat service.

Focusing on our own chat service with a reactive approach

To serve our own patrons better, we decided to focus on offering our own IM chat service using Trillium software, which we had started in summer 2005. We installed Trillian on the reference desk computer, and chat was monitored by the librarian on duty. We used Trillian for IM until February 2010, when we switched to Meebo.

Meebo was a more efficient product for virtual reference in that the patrons did not need to create a user ID in order to use it. We embedded Meebo into the “Ask a Librarian” page, which, in turn, was embedded into the various service and resources pages, including the search pages of our EBSCOhost databases.

During the 29-month period from February 2010 to June 2012 that we offered Meebo, we recorded a total of 24,864 reference transactions, with chat transactions accounting for 3 percent (n=708), phone transactions accounting for 9 percent (n=2,209), and in-person transactions accounting for 88 percent (n=21,947). Obviously, chat reference service was not being adequately used.

We believe the poor results largely reflected the reactive nature of the chat software itself. With Meebo, the librarian, whether at the reference desk or at a separate computer in an office, waited for the user to initiate the chat by using the widget. Once the chat came in, the librarian simply reacted to the call by providing the assistance the caller needed.

Switching to a business model (proactive approach)

Because of our continued dissatisfaction with the low number of students using chat service and that Meebo ended its free service in summer 2012, we decided to take a more proactive approach in reaching our users. Towards this goal, we adopted the business-oriented software Zopim, which offers the feature of trigger-initiated chat.3 We hoped that triggered chat would help make students more aware of the availability of person-to-person reference help in the virtual domain.

Zopim is easy to install. It is web-based and can be embedded in all the library pages. The live chat widget displays from the lower-right corner of the screen (Figure 1). Similar to Meebo, or any Meebo-like software, when students need help, they can click on the widget to ask a question.

Figure 1 Display of Zopim widget. View this article online for detailed images.

However, different from Meebo, Zopim does not just sit on the website waiting for the patron to use the widget in order to ask a question. Just as in its use in a commercial environment, Zopim can be programmed to initiate a chat, after the patron has resided on the site for a period of time, with an offer to help on behalf of the librarian (Figure 2). To assure that students have adequate time to orient themselves to the library site, we chose a time period of three minutes.

Figure 2 Zopim triggered and offering help on behalf of librarian.

Once the patron responds to the trigger-initiated question, the librarian on duty is alerted with a pop-up window and a sound signal. The librarian then immediately takes over by continuing the chat as if it were the librarian who initiated the chat. (Figure 3) Through the URLs that appear on the screen, the librarian can view the path the patron is taking. There is also a sound signal each time the user responds with a comment or question.

Figure 3 A chat transcript screenshot.

During slow times, such as mornings and Saturdays, the librarian at the reference desk will handle both chat and in-person requests. However, during busy times, we have had a librarian cover chat from the reference office, and another librarian cover in-person help at the reference desk. When the library is closed, the software allows the patron to leave a message for the librarian.

Another helpful feature of the software is the “Chat Rescuer.” This is used in situations when a patron initiates a chat while the librarian is busy helping someone else. Three minutes after the patron asks a question, the Rescuer notifies the patron by saying, “We apologize for keeping you waiting. Our librarians are busy at the moment. We’ll be with you as soon as possible.” When multiple chat requests come at the same time, there is also the familiar feature of canned shortcuts that help librarians respond to standard questions, such as how to find a book.

We launched Zopim at the beginning of October 2012. Because of its ability to trigger automatically with an offer to help, Zopim markets itself to the user. Therefore, we did little to promote the availability of a proactive chat feature to the university community. By the end of March 2013, we had recorded a total of 4,664 reference transactions, of which chat transactions accounted for 21 percent (n=981),4 a remarkable increase from 3 percent; phone transactions accounted for 6 percent (n=297), a drop from 9 percent; and in-person transactions accounted for 73 percent (n=3,386), a decrease from 88 percent. We were very satisfied with this spike in number of reference chat queries.

As to the quality of the questions being asked, we found the initial queries from users to reflect a combination of curiosity and caution. They asked questions such as “Are you a robot,” and “Is this a service or a real librarian?” This response may have been because we introduced the service in the second month of fall semester with very little announcement. However, once the users became more assured that they were indeed talking to a librarian, the questions began to focus on help in discovering scholarly resources for assignments.

Trigger-initiated vs. patron-initiated chats by question type

To discover in more detail whether a proactive business approach to chat actually raises the visibility of online help for our users, we did an analysis of the traditional reference questions that arrived to us through chat. We divided these chat questions into those that were trigger-initiated and those that were patron-initiated through the live chat widget. Of all chat reference queries (n=1,050)5 from October 2012 to March 2013, 70 percent (n=731) were initiated by trigger, and only 30 percent (n=319) were initiated by patrons.

We further analyzed the chat transcripts to find out what types of questions the patrons asked when they responded by trigger-initiated chat, and what types of questions they asked when they responded by the live chat widget. We divided the queries into three well-known categories of question-types: reference/research, directional/technical/policy; and other. “Reference/Research” includes a question such as “I have a research paper to do and am trying to narrow a topic down regarding Iraq and causes of our invasion of Iraq.” “Directional/technical/policy” includes a question such as “I’m having trouble logging into a library database from off-campus. Do you know what password I should use?” And “Other” includes a question such as “Can you turn up the heat in the library?”

Based on the above categories, we determined that of the 731 trigger-initiated chats, 74 percent (n=540) were reference or research-related questions and 25 percent (n=182) were direction-related, while of the 319 patron-initiated chats, 57 percent (n=182) were reference-related and 41 percent (n=132) were direction-related.

These data indicate that for the period between October 2012 and March 2013, our trigger-initiated chats are more reference and research related than are our patron-initiated chats; while our patron-initiated chats are more direction oriented. Clearly, for us, the initiation of a business model with trigger-initiated chat reached students online with complex research questions, who might not have otherwise known where to go for help.


There are considerations that might qualify how well the direct adoption of a web-shopping customer service model from the business world works in an academic library. While a majority of users declining chat service responded with, “No thank you. I’m fine,” there were a few times during the test period when users stated finding the pop-up boxes a distraction to their research. Also, the absence of the ability to co-browse with a patron in the business model may limit its ability to deliver instruction effectively. A more extended study than this one would likely reveal other qualifications.

Nevertheless, despite the limitations, we believe that exploring how academic libraries can effectively adapt the business model’s proactive approach into their chat reference systems is a valuable direction for study. Our experience with a proactive chat model at John Carroll University showed us that there is indeed a readymade market for our services right on our own library pages, and that our users will ask us questions when we approach them.

1. Elmorshidy, A. , “Applying The Technology Acceptance And Service Quality Models To Live Customer Support Chat For E-Commerce Websites. ,” Journal of Applied Business Research 29, no.3 ( 2013 ):589-596 –.
2. “About KnowItNow24×7. ,” State Library of Ohio,
[Full Text] , accessed September 13, 2013. .
3. Zopim,
[Full Text] , accessed September 13, 2013. .

The reason these numbers are not consistent is that the former number (n=981) comes from the CMS records which the librarians input manually; therefore, it is subject to omissions. The latter number (n=1050) is recorded automatically by the software.

Copyright © 2014 Jie Zhang and Nevin Mayer

Article Views (Last 12 Months)

No data available

Contact ACRL for article usage statistics from 2010-April 2017.

Article Views (By Year/Month)

January: 9
February: 20
March: 0
January: 12
February: 26
March: 18
April: 51
May: 42
June: 33
July: 24
August: 11
September: 22
October: 19
November: 10
December: 27
January: 38
February: 29
March: 31
April: 17
May: 28
June: 32
July: 33
August: 47
September: 23
October: 32
November: 25
December: 24
January: 26
February: 34
March: 46
April: 73
May: 130
June: 93
July: 19
August: 27
September: 43
October: 64
November: 35
December: 31
April: 0
May: 56
June: 21
July: 16
August: 11
September: 19
October: 19
November: 25
December: 30