 Okay, it looks like it's 5.15, so I think we'll go ahead and get started. Thank you all for coming. I'm Jan Kemp, Assistant Dean for Public Services at the University of Texas at San Antonio. My co-presenter is William Glenn. He's head of reference services at UTSA. Our briefing today is called Transforming Online Reference with a Proactive Chat System. And today we'll be talking about how our library experienced dramatic increases in the number and complexity of chat questions after we switch to a proactive chat system in 2013. I'll start by giving an overview of the system and its characteristics, and then William will provide some details about some of the system's features. Can everybody hear me okay? Okay, great. So Zopim is a chat system that was created for use by online businesses. It's not one that was in the library world. And some of you, or maybe not, but in 2013 we did a presentation about Zopim here at C&I after we'd had it up for about four months. So now it's been two and a half years, and the increases that we saw after those first four months have only continued and increased in the last two and a half years. As soon as we implemented Zopim, we immediately started receiving large numbers of questions. And in fact, the increase from 2013 to 2015 was 489%. I know, it's incredible. It was to us too, we had no expectation for that. But we don't think that our library is unique. We think that this system would work for anyone. So hopefully what you see today will give you some ideas about that. So actually, full force online reference was predicted in 2005 by Charles Martel, who was then the dean of libraries at California State Sacramento. He said, well, he was talking about the slowdown in the use of physical services in libraries, including reference. And he said, reference transactions will probably remain soft until online reference services appear in full force. However, I think for many libraries, they're still soft. In our case, I think we have seen full force. So maybe we'll tell you a little bit about that. Before you go any further, I just wanted to give you a couple of facts about University of Texas at San Antonio. We're one of the UT system libraries. We're the third largest. We have about 29,000 students. And our physical library is extremely busy. Our main library on a busy day in the fall has anywhere from 9,000 to 10,000 visitors in it. And that's almost a third of our student body. So we feel that we, although we had a lot of students in the library, even back in 2013, they weren't using our reference services, whether in person or chat. And so we had a goal. We wanted to reach more students with reference. The questions that we did get made us realize that there are a lot of students with questions out there. And there's no reason why these students who ask would be so different from the others. And so we started looking around for a chat system that might attract more users. And our user experience librarian, when she went to the ACRL conference in 2013, saw a presentation by the librarians at John Carroll University. And they had just implemented Zopim, and it had doubled the number of chat questions. And maybe even more exciting for them, they had increased the complexity of chat questions to about like 75% of the questions that they received were they considered to be complex. So we decided to go ahead and give it a try. And there's a lot of times in life where you have to say the change didn't happen overnight. But in this case, the change did absolutely happen overnight. On July the 22nd, 2013, we had eight chat questions, and that was very typical for us. The next day, July 23rd, we had five times that many questions. And that was without any change in our staffing, our marketing, or our hours. So we just had to attribute that completely to the new system that we implemented. And what we saw also very quickly was that the majority of these questions were complex questions. They were actually questions from students who were working on topics, who needed help figuring out which database to use, or didn't understand the topic they were working with. And so six weeks later, we began staffing chat with librarians only, which was quite a shock to the system. So what do we think, oops, one, two, five. Okay, so what are the characteristics of the Zocum system? There is a chat box on every page of the library website now, including in Summon. So it's there ubiquitously throughout the library website. Perhaps the most distinctive feature, I think, is its context-sensitive triggers. So what will happen is the libraries who use it have the opportunity to notice which parts of the website, which pages give patrons the most difficulty by looking at web analytics. And so, for example, one place where we have a trigger is the Find Databases page. We notice that when students would go to that, good things did not happen afterwards. And so now what we have is a trigger on that page. And if a student or a faculty member for that matter sits on the Find Databases page for 30 seconds without taking any action, a new chat box pops up in the middle of the page. It's a little bit larger than the one, the regular one. And its context-sensitive, it says to them, can we help you find a database? So it knows that they're on the Find Databases page. And William will show you in a little bit more detail how that works. Another feature is that the chat session continues as the user moves throughout the website. It doesn't go away if they leave. And the chat box can be customized and branded, which is something that we did, which I'll show you an image of it in a few minutes. So what do we, what do I think caused the chat increase? I think that the branded larger chat box, which was so different from what we were using before, really gets their attention. The chat box wording has changed a lot. We used to say ask a librarian, and that's kind of kind of dried. So we moved to something more friendly, chat with us. I believe that the triggers are offering them help at the point of need, which before they would, there were several steps for them to get to the chat box. And I think that it makes sense that users would ask more difficult questions online and more questions online, because that's where the library research is taking place now for the most part. So here is the library homepage on the right. And in the right-hand corner of the bottom, you can see the chat box. So it's quite orange. It says chat with us, and the image there is an icon of our Blue Crew team member. We brand our public services staff as the Blue Crew, and that's Blue Crew Sue. So trying to make it friendly and appeal. I don't particularly like Blue Crew Sue, but students apparently like Blue Crew Sue. So that's fine. And then on the left-hand side is what the chat box looks like when the student has asked a question. You as the student and it's saying, I need to locate an article of the Journal of Visual Communication and Medicine. Can you point me in the right direction? And then Rita is our librarian. And you see they're branding her as a member of the Blue Crew and saying, you know, and then she'll answer their question. This page shows what the triggered chat box looks like. On the right is the triggered chat box. It pops up a little bit above where the other chat box would have been in the lower right-hand corner of the bottom. It's a little bit bigger. And as you can see, as I mentioned with the Find Databases page, the wording there is context sensitive. It says, hi there. Let us know if we can help you find a database. And it's comparing it with on the left-hand side the regular chat box that's sitting there on every page. So as I mentioned, we've experienced a huge increase in chat. And we started off in fiscal year 2013, about six weeks of that 2013 fiscal year were Zopim. The rest was our lower level chat. So 4,600 questions in 2013 to over 20,000 in 2014, which was a full year of Zopim. And then this past fiscal year, 27,039 questions. That's both reference and directional. Everything that came in through chat. So it's continued to grow, too. William has worked constantly to try to make our chat service more efficient. And so in the week of September the 27th, 2015, we had over 1,000 questions for the first time in one week. In fact, we found out that many faculty, or not many, but some faculty have started to encourage their students to get on chat when they're working on a project. And I think what may have happened one morning when we had 34 questions in one hour was maybe something like that. And that was an hour we weren't prepared for, so we had to take some names and call people back, or email them back. So we've also done some interesting comparisons about the complexity of the questions that we received. So the bar on the left is the percentage of complex questions that we receive at the desk. In two sample weeks, February 2015 and September 2015, the total number of questions at the desk was about 420. The total number of questions that were categorized in chat was 1,750, something like that. So just to give you an idea of the numbers. And so 20% of the desk questions were complex, reference questions, and almost 75% of them. And in fact, that just matched what the John Carroll University found out from Zopim, too, when they implemented. And then this is something that's pretty interesting. It looks like the questions are more complex when it's a triggered chat than when it's a question that comes from a patron-initiated chat. So if we prompt them to ask their question, 82% of those questions are reference questions, complex questions, not simple ones, like where's the study room, or can you help me find a known item? And 65.4% of the sample questions in those two sample weeks were complex for patron-initiated. And in fact, the trigger chats are the majority of our chats now, or barely write it. Write it 60% by under 60%, something like that. Yeah, so if we weren't triggering chats, if we weren't inviting them to chat, our chat numbers would be cut in half, or less than half, if I'm not mistaken. So just for fun, I wanted to share with you, because everyone loves reference questions. So here are 10 questions that were asked in a one-hour period on Sunday, October the 25th. William happened to be the lucky person who was working chat at the time when all these chat questions came in. So for the benefit of our audience here in the recording, let me just read a couple, or a few. I'm trying to zero in on a topic from my literature review and find articles from specific journals on the topic. Currently, my topic is cultural variances in female leadership in nonprofit organizations. Question two, I'm looking for books or journal articles, government doc, about the roles of GSEs in operation of CRA. Would you mind helping me find an article on the United States Depression in the 1930s? I have an essay assignment about the benefits of higher education, and we need three academic sources from scholarly databases. And I need help finding a court case where the defendant is facing the death penalty and where the defendant is still alive. So these are pretty, could be pretty meaty questions, depending on where the chat goes from here. And that's just in one hour. So they're not that atypical, or they're not atypical at all. They're atypical. So faculty are asking us questions as well, which was something we didn't see before when we had our other chat. I think they know perfectly well they have a subject librarian, but they know also the subject librarian doesn't work nights and weekends. So here are a couple. A journal asked me to provide keywords for my manuscript. Should I select keywords from established lists to make sure others find my paper? And I'm looking for respondent level data sets for my students to analyze using SPSS. What might you recommend? So you can see why we like to have librarians answer those kinds of questions. And so to conclude my first part of this, I just wanted to say that we're pretty sure that these questions had been out there all the time. I think that because as soon as we implemented Zopam, the numbers jumped right up, and they stayed there for two and a half years. So it's not anything that we had to build slowly and work really hard for us. Just boom. So what I would just like to say is that I think that the proactive chat has lowered the bar of inquiry for our reference questions. And now the library can play a larger role in supporting student success. So we'll begin. Hello. OK, so I just wanted to go through some of the back end of this and give you some additional information. Maybe we'll answer some of your questions you might have in mind. So my name is William Glenn. Before I was at UTSA, I was at Stony Brook University in New York, where I managed the virtual reference systems there for seven to eight years using initially OCLC's question point for quite a few years and then later, SpringShare's LibChat. And I think one year we had a little over 2,000 questions. So we can do that now in about two or three weeks. And so for me coming, I've been at UTSA now a year and a half almost, I interviewed for the job. They told me all of this. I looked it up, did my research, but not until I got there and really experienced the difference between what I'd been doing for seven or eight years and this, did I understand how powerful and transformational this really is. And after a year and a half, I mean, I'm still stunned at how many, not only how many questions we get, but again, the kind of questions we get, the complexity and the feeling that we really are helping students in a way I hadn't felt maybe as much for a while. So I want to go into the triggers a bit more and the statistics, how we use those shortcuts, which are very helpful. And then a really exciting aspect of all of this that we've sort of discovered is mining our chat transcripts. So just to kind of reiterate what Jan showed earlier, this is our fine databases page and here is the chat that has been triggered. So as she said, we have the widget all throughout the website and within Summon. We currently only have triggers on two of the pages. The fine databases page, which we know from analytics is the most visited page after our homepage and also within Summon and the results in Summon. We've thought, as she said, those were really the points of need. People trying to figure out, a lot of our students are still working with the jargon of library. They don't really know what they need if it's a database, a journal, an article, but they've been told databases, so they do tend to go here a lot. So those are two of the... Then we also have a third trigger, which comes on after 50 seconds when one of the librarians has not responded. Our chat rescuer it's called and that's very helpful when we're really, really busy. It's not on a page, it's really part of these triggers. And let's look at the backend. So this is actually, it's pretty simple. There are a lot of different parts there, but really you can set these up. Once you've done one, it's very, very easy to do the others. The three most important parts, and I don't know if you all can see it, but it's like you want the first red circle there, the URL of the page you want. And then the second one is the time length. How long do you want patrons to be on that page before it goes off? Actually, you can see that was 35 seconds. This is for the fine databases page. We've been playing, experimenting a bit with the time length, trying to find... You don't want it going off too much, but you also want to catch them in need. So you can change that very easily. And then down at the bottom, send message to visitor and you just type in the language that you want. And you can adjust these as necessary. The others are kind of conditions, logical sorts of things. If the visitor is doing this, you don't want them. One of those is if they've already initiated chat, you don't want the trigger to go off, things like that. But really once, like I said, you do one, then you can set up the others very easily. And I don't know if there's anything else on that page. We did have them earlier on the e-journals page. Actually, that was removed even before I got there. Do you remember why? Probably just not as much traffic perhaps. They definitely summon our discovery system, gets a lot of that on the homepage of the databases. And then recently, a few months ago, we've revised our accounts page for patrons and we wanted to kind of get some feedback from them. So we easily set up a trigger, set up the time. And with the message after, I don't remember now actually 45 seconds, something we asked them if they needed any assistance and how do you find the new page. And so we left that up there for a while. And then once people started to actually just ask research questions from my accounts page, we realized we could begin to take that off. So the students love it, I have to say. And actually the faculty just this morning when I was on chat, it's great to hear that, faculty members saying, I just want to tell you again how much I love this service. So with the explosion in the number of patrons we're dealing with, the statistics have been very, very important, even more so than when I was doing them before with the chat system. This is a monthly report that is sent from Zopam. It's pretty basic. As you can see it has page views and things, but the two most important numbers there are the ones I circled, which are the total chats. This was from October of last year. That was the busiest month ever that we've had so far. And then over on the right you'll see it says chats completed, chats dropped, chats missed. So we went through all of these categories I did and made sure of what they were. We wanted, you know, I see 200 missed chats and I'm like, oh my God. And so I dug into those a lot and what we discovered were half of those were patrons saying no thanks, I don't need any help. And the librarian doesn't respond. And so it's really not a reference transaction. So that's the largest section of that. There were also sometimes because of the trigger once in a while people will type in, like they're thinking they're gonna search in the search box, but they start, and actually if I may confess, I have chatted with myself on a couple of occasions. You get very busy and I type in climate change and economics and then I hear the chat go off and I'm like, oh my God, not another chat. And I look at it and it's like, oh that's weird, it's the same topic. It's like the same class. That really happened, it's kind of embarrassing. Once in a while, that's a low number but sometimes it does happen. So what we found were the chats missed really only, we only missed 2% of our chats. Genuinely do not respond to a patron. Which I find stunning because when we were doing 2000 a year at Stony Brook, we were trying to keep it under 10% of missed chats. And now with such an enormous volume, I was so impressed with our librarians that and it gets really busy sometimes but we don't miss that many chats, realistically. Then I also looked at chats dropped, that sounded dire. So that turns out, those are really reference transactions though. Zopam basically I think is when the patron says the last thing in a chat and they immediately leave, they call it a dropped chat, I guess because we didn't finish it or they didn't do something but I've looked through those and those are legitimate chats. So really when we're trying to figure out how much each month we're doing, it's the total chats only minus the chats missed. And then we do have offline messages, you'll see which come in overnight, we handle those just like live answers, questions or email reference. We'll do those first thing in the morning. There's not too many of those as you'll notice. Chats on until 11 at night, starts at nine in the morning most days but we do get some in the middle of the night. So in addition to that monthly report, Zopam sends us very detailed files each month that we can go through and use to create things like this. This is our spreadsheet, this was October 2015 and I was able to figure out the average number of chats per hour to see when are the hotspots, the red being the most. You'll see Thursday from 11 to 12, we were averaging 19 questions in one hour and then the orange is a little less and then the yellow and then the white. As you can see Saturday, not a real popular day so we only have one person on most of the day but this has been incredibly helpful it also allowed us to expand because I was seeing enough questions between 10 and 11 that we were like, oh, let's try nine and sure enough it's not as much but you can see that's 10 questions an hour, a couple of days and now we're about to expand to midnight because we get so many questions between 10 and 11. You see a lot of those are eight, nine questions an hour from that time period. And we can do other things, Zopim you can also just look within Zopim I've been able to verify a lot by seeing what they send us and then you can use their search mechanism to go between very specific time periods and I can see how many chats there are and I can read them to see what were the missed chats, what were the dropped chats. You mentioned, or I think you mentioned you have librarians, do you have librarian staff for the night to midnight? We do, yes. Professional librarians? Yes, all of them are professional librarians. Three of them are part-time librarians? We have three part-time, well, now four part-time we just, she starts today. But they actually, the part-time librarians handle about 50% or a little more of the questions because they're doing primarily chat. The subject specialists do X number a week because of everything else that they're working on. So another great way to help is to have these shortcuts and I will say having used question point and lib chat, these shortcuts in Zofum are fantastic. You're in the middle of the chat, you start typing the word and it populates and boom, you send it off like that. You don't have to think about it. I don't know, I remember the canned messages I had to go and do a drop-down menu and choose it but this is super easy when it's very busy and even when it's not, they can be very useful. You can see just, this is a sampling. We are doing about 125 shortcuts and I went through and reviewed all of them. Some are not used that much. Some are used a lot but if, and we actually eliminated some that weren't being used at all but you can see the first one's canceling a study room. We'll get a fair number of questions. There's a link to information for one of the subject specialists. So we try to refer a lot. It's chat, it's hard to accomplish some of things we've been able to do at the desk in the past and you really wanna make sure that they're able to come in and talk with someone in person. So we do refer as much as we can. Points of view, academic search completes. So some of our popular databases will have links that we send to them. Policies, I see alumni, borrowing acts, things. Where are the book drops? And then things like anything else which actually help a lot. What I would call conversational shortcuts. Hello, how may I help you? I'll be with you in a moment. I'm helping other people right now. There's different things. Is there anything else I can help you with? These come in very handy when you're handling multiple people at the same time. And these are even easier to set up than the triggers. I mean, give it a name, type in the message, create shortcut. And you can create those on the fly, which we've done in some cases. Weird network problem. Or the classic one, a middle school who discovered our website in our chat service. And suddenly we were flooded with middle schoolers asking all kinds of interesting questions. The classic one, are you a robot? We get that from time to time. Although less now. We used to get that a lot in the beginning. So, created a little message to send to our middle schoolers and send that off. Actually came up again just recently, the night that an assignment was due. One of the only database that could be used went down and we were getting flooded. So, the librarian just created that instantly and was able to handle. So, we're again getting multiple questions. So, those are really nice, very useful. And then my, yeah, I think this is super exciting, mining the chat transcripts. I didn't really just think about it at first, but it's really fascinating because it's both qualitative information and quantitative because we have 27,000 plus chats a year. What are our patrons talking about? What are they wanna know about? What are they complaining about? And actually it was Jan who said, the first time I'd really started to click with me was, hey, let's look at all of the noise complaints. What are people saying? Because they would use chat a lot, makes sense. They don't wanna get up and go tell on somebody or things like that. So, we were able to go through and using the search mechanism and Zopim be able to look up noise, loud, quiet, various terms for a time period and very precisely figure out what people were saying. We went through all the ones we could find and you'll see there. Can someone please tell the people next to the quiet study area on the second floor to turn down their videos or music? Third, four quiet zones, some, yeah. How do I submit a noise complaint? How can I file a noise complaint? Study room 23 on the second floor. So what we found was, and actually, you know, I knew this was gonna happen. I debated when to put these slides. I'll go back. We were able to very precisely figure out where the noise was happening. And it was basically two areas. Our fourth floor area, which was not enclosed and was hard to keep quiet. And then we kind of realized that our third floor quiet zone was surrounded by study rooms as opposed to the other side, which was surrounded by faculty offices. So, switch, we did that. We moved the quiet zone from the fourth floor down to an enclosed area on the second floor. But having a map, you know, and being able to see specifically, I mean, they were giving it, because we would ask them, you know, can you give us a call number range or whatever would help us? So that was, and after that, it was suddenly like, oh, hey, we can look up all kinds of things. Because I'm also on our web presence team and our discovery team, where we're investigating, you know, how are patrons using the websites? How are they finding resources? What problems? That was the big thing we're, because also because of live call, we had, you know, we could see that people were frustrated across all the levels with accessing online resources. Well, what does that mean? So we started going in and you can see this is the search thing mechanism. And it's very easy. Keyword, it worked really well too. Date range, time range, which can be very helpful at times. And that's usually what we're doing to find eBooks. So we've been analyzing what are all the issues that people are running into with our eBooks? Which platforms? What are they trying to do when they have a problem? What titles are they trying to access? Streaming video, we did the same thing. So now we're keeping track of these. I'll show you that in a second. And then also this is how we find, you can see we can look for the different kinds of chats, triggered, and things like that. So that's one way we're able to determine percentages, like how complex something is. Or chats by a person, when, you know, I can go through and look for people. And also the chat transcripts are just fantastic for training purposes. You know, it's just wonderful to have these reference transactions that people can study to see, you know, what's a good job of doing this. And I review our chats on a regular basis to pick up on things, trends, and we use the search feature. If I notice a couple of things, and I'm like, uh-oh, this book title's having issues. Let me go in and sure enough, oh my God. There was, you know, because they happen just infrequently enough sometimes, it's just an assignment usually. And then all of a sudden you see there's 14 chats about the same eBook. Well, boom, we can go in and make sure it gets straightened out. We had that recently, database issues. What are courses? We've also, we see weird assignments. The one, you have to do a search on men and the culture, but it has to be from a print journal, you know, all 37 we have left. And the poor kids, you know, these kids, they had no idea how to search for something within a print journal. So when we started to see that, we're like, we better talk to this professor. Is this really, you know, do they, you know, and we contacted them. And we can also see what kinds of questions are coming in based on a course or a program. We, for the psychology librarian, we did, we did research on psychology just that term. We looked up psych info and other databases and resources. How many times were they coming up? What were people saying? So it's been a fantastic source for us. And then this is just us tracking ebook problems. We have the date and time. You can copy and paste that pretty easily. The visitor number and the date and time allow us to go and look at the actual transcript. And we found a way to put the transcript in the spreadsheet, but it was really clunky. So, but yeah, the question, the title, the provider and what we did. And so that's, this is from September 22nd through November 9th. These were some ebook questions using different terms to try to find what people are looking for. So on that note, data mining work, that's all I have. And so we would love to get questions from you all. Okay everyone, well thanks so much for all your questions. Thank you.