SUBSCRIBE NOW!

Vol. 12 No. 8 — September 2004

Unfortunately, chat reference has not turned out to be the panacea many of us hoped for. Studies by Joe Janes and others show that the numbers of questions asked on chat reference services are, generally speaking, abysmally low. Janes' Census of Digital Reference finds chat services get a median of just six questions per day and evidence gathered from a broad spectrum of other libraries shows similarly low levels of usage — even among services that have been up and running for several years. Of course, there are exceptions to this rule, with a few services reporting thousands of questions per month. However, most of these reports come from collaborative services made up of many libraries and, on closer inspection, most of the individual libraries making up the collaboratives show the same low levels of use seen in other virtual reference services.

Data gathered over the last few years shows that chat can also be a very expensive way to answer a reference question. First comes the cost of the chat or Web contact software; then the cost of training the staff to use it. And most librarians find it difficult to handle chat and other reference functions at the same time, so managers have to pull staff off the desk and dedicate them to monitoring the chat terminal during all the open hours — regardless of how many questions come in. Finally, when a question does come in, evidence from commercial chat services such as those offered by LL Bean and LandsEnd indicates that it usually take twice as long to answer via chat as over the phone. Add up all the costs, and the cost per question in a chat reference service can be downright horrendous, especially at the average rate of six questions per day.

If librarians had all the money in the world, it might not matter. With unlimited funds, we could offer reference service round the clock through every communications channel — wireless, IM, text messaging, video conferencing, kiosks, etc. Sad to say, we do not have unlimited funds. In fact, libraries operate with tight or even shrinking budgets. Many chat services originally started using grant money and those grants are beginning to run out. Librarians face tough decisions on how best to "shoe-horn" new chat reference services into library budgets already stretched way too thin or whether to abandon the fledgling services altogether. To make the hard decisions, we need to take a hard, honest look at the options for chat reference going forward. What are the alternatives?

PULLING THE PLUG

If you have a virtual reference service that does not live up to your expectations, the first and most obvious solution is to just "pull the plug." A few brave librarians have tried chat and then opted out. A couple more well-known examples are the now defunct chat services at Vanderbilt University, MIT, and Los Alamos National Labs. Perhaps the most interesting case, however, is the Library of Congress — one of the earliest proponents of digital reference services and the defacto flagship of OCLC's QuestionPoint reference service. LC has not exactly closed its chat service, but it has severely curtailed it over the past couple of years. Soon after the Library of Congress moved from its pilot CDRS service to QuestionPoint in June 2002, 11 of the LC's reference divisions were offered chat service for at least 1 hour per day. As of April 2004, eight of those 11 units had closed down their chat reference and were offering only e-mail. All these libraries (and no doubt others that we haven't uncovered) have simply made the decision that the value gotten from chat services did not justify the investment and decided to spend their money and efforts elsewhere. That's one way to handle an underperforming chat service, but it is not the only option.

MAKING IT WORK BETTER

If you don't like the idea of killing off your chat service, then try to improve the way it works. Generally when people talk about improving chat services, they focus on marketing and other ideas for increasing the number of people using the service. However, as we seen, usage is only part of the problem; the other is cost. And if you succeed in increasing usage without reducing costs, you could get the library into a tight space financially pretty quickly, considering what it costs to answer questions virtually and that costs per question do not go down significantly as volume increases. On the other hand, there is probably not a library in the world that could not do a better job marketing itself in general and its virtual reference services in particular.

Marketing works. The experience of services like QandA NJ and KnowItNow in Cleveland clearly indicate that one can get a respectable number of questions with some attention to publicity and a few lucky breaks. And Tutor.com's experience helping libraries to advertise its live homework help service — a specialized form of virtual reference — to school children with pizza parties and the like also indicates that modest amounts of money can have significant impact when targeted to the right audience.

However, just how much can marketing do? Few companies on the Web have more exposure or marketing clout than Google. Yet with over 200,000,000 searches per day, it never has attracted more than a couple of hundred questions per day to Google Answers and, in recent months, the average has dropped down to around 60-70. Granted, you do have to pay for the service, but at an average fee of $15-$20 hardly seems much of a barrier. However, the most compelling evidence that marketing can only help so much is the untimely demise of live reference services such as WebHelp and of the commercial reference market in general. Many of these services spent millions of dollars of venture capital money on marketing. WebHelp even put up a giant, two-story-high neon revolving sign on the busiest street in Toronto, as well as banner ads all over the Web. While those antics may have bought some traffic for awhile, it was not enough to make for a sustainable business model and today all are gone — along with the millions spent trying to market their services. The limited traffic at Google Answers and the demise of well-funded commercial reference services on the Web raise some serious questions about just how much people really need or want reference services online — no matter how well marketed — at least in the ways we have offered up until now. Finally, if marketing could help us increase the use of virtual reference services, couldn't it just as well attract people to traditional reference services?

What about reducing the costs of virtual reference? There are a variety of ways to accomplish this. Look at staff costs first; they constitute one of the biggest expenses in operating a virtual reference service. If you currently staff your virtual reference service separately, consider moving it to the regular reference desk. Although many librarians would not recommend this, the fact of the matter is, some libraries have successfully run a chat service from their regular desk — particularly if desk traffic is light and you only get a few chat questions each day. Another approach is to contract out staff. Both Tutor.com and Docutek offer after-hours and weekend staffing services for virtual reference, and several libraries have asked them to run their virtual reference services altogether to free up their own staff for other work. This could be a very reasonable option — particularly for services not getting a lot of traffic. Because of economies of scale, vendors can often provide virtual reference services much less expensively than individual libraries can do on their own. And these services sometimes employ more experienced virtual reference librarians, because they are answering thousands of questions each week. So the quality may be as good or better than what you can offer yourself.

Another way to reduce costs is to join a consortia. Typically in these arrangements, libraries share reference responsibilities and the cost of software, and some also share other expenses like marketing, access to subject specialists, etc. You still have to help staff the virtual reference desk, but only for a few hours a week — the rest of the schedule is covered by your partner libraries. The consortial model has another advantage. The staff you assign probably won't have to worry about twiddling their thumbs waiting for questions to come in. They will be answering questions coming from everyone in the consortia. Consortia do have their downsides; you may not have much control over the quality of reference others do in your name, and you have to go along with the software and policies that the group has adopted. If you are willing to comply with consortial policies and procedures, however, it can save you some money — and perhaps enable you to provide a better service than you could afford on your own.

Finally, you can save money on software costs. Much has been written comparing the costs and features of the many versions of virtual reference software, and we have neither the space nor the stamina to try to recap it all here. Suffice it to say, that virtual reference software is available in many different price ranges, starting at free and going up to $50,000­$100,000 or more depending on the brand of the software, the size of the system, and the features you want. Some people have argued that free or low-cost systems like AOL, MSN, or Yahoo! Instant Messaging may suffice to do online reference, especially if you only get a few questions a day. However, others claim that the more sophisticated and expensive packages are necessary to do an effective job with database co-browsing and other special reference needs. Study how much you really use and need the special features. No matter what you decide, keep the software costs in perspective.

Overall, however, evidence indicates that staffing costs could be the most important consideration. Look for alternative technologies and strategies that can affect staffing demands.

ALTERNATIVES TO VIRTUAL REFERENCE

Rather than asking the question, "How do I do virtual reference?" — a question answered by "Virtual Reference Desk" conferences, seminars, teleconferences, etc. — we should ask the broader and more fundamental question: "How can I best serve my patrons wherever they happen to be?" Perhaps there are better ways than chat of accomplishing the same purposes.

Answer the Phone

Well over 90 percent of American households have had traditional landline telephone service for years — far greater than the number of households with computers and greater still than those with Internet access. Over the past 10 years, millions of us have been buying cell phones so we can talk anywhere we happen to be. According to the Statistical Abstract of the U.S., the number of people with cell phones has jumped from 22 million in 1995 to over 140 million in 2002. That's over 70 percent of the total adult population — and it is continuing to grow at the rate of 15-20 million new subscribers a year. At that rate, it won't be long before almost anybody older than 10 will be running around with one on their belt or in their purse or backpack.

Not only is the number of phones growing, but their technical capabilities are also improving. Many can now take and receive color pictures, keep track of personal address books and calendars, send and receive e-mail, browse the Web, even serve as portable GPS systems and route finders. In fact, the cell phone is well on its way to becoming a kind of personal information center most of us have on us wherever we go.

Despite the growth of the Internet, the telephone is still the way most of us receive information and get questions answered. While few companies offer chat or any other kind of live communication on their Web sites, almost all can be reached by phone — many 24 hours a day. And not just for simple questions, either. Many companies use the phone and Web site together to help answer questions and provide customer support. And it can work effectively for online databases support as well; both LexisNexis and Westlaw have offered research assistance over the phone 24 hours a day for a very long time now — but neither offers chat.

The telephone remains the communications channel of choice for most of our patrons as well — at least when they can't visit us at the desk. Telephone reference at most public libraries far exceeds anything ever gotten in chat. Just remember those statistics from Santa Monica Public Library that appeared in our previous article. This library routinely handles over 16,000 phone questions per month (over 40 percent of the total reference count) over the phone, while chat is lucky to exceed 100 questions per month. When the Main Library was moved for a brief period, telephone reference for that month shot up by over 4,400 calls, while e-mail questions only rose by 30 and chat by just 25.

Of course, most libraries have offered telephone service for a long time, but we have done little or nothing to emphasize it or even improve how we manage and integrate it with other new technologies and reference tools. What about 24/7 telephone reference? In the past few years, existing late-night telephone reference services in New Jersey and Massachusetts have actually been shut down. We have several dozen statewide virtual reference projects, but none include the telephone.

Some libraries seem to treat telephone patrons as second-class citizens. They may set limits on the number of questions a user can ask over the phone or the length of time a librarian will spend answering them (often three questions or 5 minutes), although no such rules apply to chat. In fact, the average chat runs about 15 minutes — three times as long as what's often allowed on the phone. For example, look at the differential treatment of chat and phone patrons at UCLA Libraries. As mentioned previously, UCLA has a nice chat service that operates 46 hours per week until 9 P.M. most weeknights and it even has a cute little PacMan-like icon for it right on its home page. But take a look at the library's telephone service. For the Research Library, the largest library on campus, telephone reference is only available from 10­11 A.M. — 1 hour — Monday-Friday; and as you might imagine, it's not too easy to get through to them even then. Although some of the smaller libraries have somewhat better hours, none of them are open past 6 P.M. Chat continues until 9 P.M. Also telephone patrons are advised, "Because priority is given to users in the libraries, callers may be asked to hold." Chat patrons get no such warning.

Originally, many librarians got into virtual reference because we felt that most of our patrons were turning to the Web to get their information and we wanted to be there to serve them when they needed help. Back 4 years ago, when most people connected to the Web over dial-up connections, chat was the only way to communicate with them live, and the phone was not an option because the users had their phone lines tied up with the computer. But that is no longer true. Today, with the growth of DSL, cable, and other broadband connections and with the ever-increasing number of cell phones, the number of people who have to use chat to communicate while online is steadily declining and, at least in the U.S., dial-up Internet connections will be largely a thing of the past within the next few years.

Evidence shows that plain old telephone reference could have some distinct advantages over chat technologies:

• Most patrons have easy access to the technology. Most already have one phone, some have two or more. So, no problems with the "digital divide" and no software to download or configure. Most phones work in the same way; no need to worry about whether your patron has a PC or a Mac or what version of what operating system they might use, as with chat.

• Libraries already have the technology in place. No software or hardware to buy; nothing to download or configure. Just let people know your number, then sit back and wait for the calls to come in. Of course, if you want to put in sophisticated call centers, collaborate with other libraries, set up statewide or worldwide projects, or have staff work from home, you will probably need to invest in some additional telecommunications infrastructure. However, unlike some of the chat and Web collaboration software we've tried to use over the past few years, your advanced telephone system will probably work very well, because call center and associated communications tools have de-bugged the system for years.

• Librarians would need no special training. Most of us are pretty comfortable with basic phone technology.

• Questions can be answered a lot faster. Studies at commercial call centers show that the same question takes about half as long to answer over the phone as it does in chat. There is no comparable research for reference work, but still no reason to suspect it would differ.

• Librarians will no longer have to fret about misspellings or trying to figure out that cute chat lingo like g2g or brb.

• Even the online research assistance provided through virtual reference could be done just as well using the phone. Both LexisNexis and Westlaw have a long history of providing database assistance over the phone, and thousands of companies use a combination of the phone, the Web, and e-mail to answer complicated questions and offer support services of all sorts.

So, if we are really worried about serving our remote patrons, perhaps we should take a little closer look at a technology we already have — the telephone. It may not be new and sexy, but it does work and it is a tried and true method of providing reference service at a distance — even on Web.

Offer a more effective asynchronous reference service using either e-mail or Web forms in place of chat. Libraries have been doing e-mail reference for quite awhile now and — up until now anyway — it has never gotten much use — even less use than chat. In his Digital Reference Census, for example, Joe Janes found that e-mail or Web forms only accounted for 30 percent of 8,106 questions, while chat accounted for 70 percent. And most libraries have found similar results. In fact, it was the low usage on e-mail reference that spurred many libraries to get into chat in the first place. So don't expect e-mail to do much to improve the usage of your online reference services.

On the other hand, it could do a lot to reduce your expenses. E-mail is a considerably less-expensive proposition than live virtual reference. It does not require expensive software. You probably already have what you need on your computer and, even if you have enough volume to require an e-mail management system, like Altarama or QuestionPoint, these are normally much less expensive than virtual reference software. E-mail is relatively low-tech, so there is little to break or go wrong. It works with almost any computer regardless of brand or operating system. Most important, e-mail reference is much easier to staff and to integrate within existing reference operations than a chat service. You won't need to dedicate staff to sit in front of a terminal and wait for questions to come in as you do with virtual reference. Somebody merely has to check the inbox regularly in between other tasks. And, because people who e-mail questions don't expect an immediate response (otherwise they'd call), staff can work on the question as other duties permit, as long as they make sure to get back to the patron within a reasonable length of time.

E-mail reference also allows staff the luxury of pondering and exploring the question a bit without the pressure of having to deliver an answer on the spot as with chat, phone, and other live reference channels. Although some claim that the reference interview and follow-up can be difficult and time-consuming with e-mail reference, it certainly does not seem much of a problem for Google Answers, which relies entirely on an e-mail-, Web-form-based system and offers no chat, Web collaboration, or live communication of any sort between customer and researcher. So if it's good enough for Google, there's a good chance it could work well for us as well.

Finally, if libraries offered better turnaround on e-mail questions, our patrons just might use it more. Benchmark Portal, a call center research service at Purdue University, found that the companies with the highest customer satisfaction scores responded to e-mails within 3 hours, although most customers said they would be satisfied with a 24-hour response time [http://www.benchmarkportal.com/newsite/article_detail.taf?topicid=31]. But libraries are often way off those marks. A 48-hour or 2-business-day response time is common in many libraries, and some stretch it even further. The Library of Congress requires "5 business days." Google Answers states that most questions are answered within 24 hours and, although hardly doing a "box office" business, it still gets many more questions that most libraries do over the Web. Santa Monica Public Library offers a 2-hour turnaround during regular business hours and 24 hours over the weekend, and it routinely receives more e-mail questions than chat. While offering a 2-hour turnaround may require the staff to hustle a little to meet deadlines, it still would be nowhere near as demanding as on-the-spot answers from a chat reference service.

If you want to reduce costs, but still preserve some reference services online, try e-mail, but keep it on a tight schedule.

What do most people use first to help them find information on the Web? Most people — including most librarians — would probably answer Google. The book-oriented might answer Amazon. While some librarians debated how to catalog the Web, Google built a search engine that, though not perfect, helps most of us find what we want well enough that it has become the one place most people turn to first when looking for information online. Amazon leveraged the capabilities of the Web to build the largest bookstore the world has ever seen. To help people find what they wanted among those millions of titles, Amazon redefined the catalog, adding lengthy book descriptions, reviews, cover art, spell checkers, excerpts, and, most recently, its "Search Inside the Book" full-text feature.

For all their success, neither Google nor Amazon offers even the most minimal kind of chat service or instant messaging or phone service, to say nothing of the elaborate Web collaboration tools featured in some virtual reference software. The closest you can come to interacting with a live human being on either site is to fill out a Web form and wait for them to get back to you — and neither service makes that very easy, either. In Amazon, the "contact us" feature is buried many layers down in the site — deep beneath the FAQs; with Google Answers, you have to pay for the privilege.

Both services focus attention and money on improving the self-service functions of the sites. Both have built information systems that provide most people on the Web with the answers they need whenever they need them. The systems are fast, scaleable, and don't require users to type messages back and forth. The answers may not always be perfect — but then, neither are we.

What if we take the money and time spent developing and running chat services on improving our Web sites and information systems? Although library automation vendors have struggled to keep up, even our best library catalogs still look no better than pale copies of Amazon. The commercial databases that cost us so much often have clunky, difficult-to-use interfaces. Too often information on our sites is organized by vendor, instead of by the patron's information needs. We still use bibliographic records designed to fit on 3x5 cards that look like skeletal remains compared to the rich descriptions you find in Amazon or other modern-day catalogs. We've neglected to take advantage of new technologies like collaborative filtering and others that help personalize sites like Amazon. Libraries could do a lot more to make it easier for our patrons to find information on their own.

The problem is, of course, that we don't have the kind of resources that Amazon and Google can bring to the table. Or do we? Google started out as a student project and built itself into the Web's premier search engine with $25 million in venture capital funding. And although it is difficult to get detailed information on how Amazon financed its infrastructure, we do know that the American Booksellers Association and around 1,000 independent bookstores (most even more poverty stricken than libraries) got together and created Booksense.com, an Amazon-like catalog and fulfillment system that allowed even the smallest independents to open up online stores with an inventory of over 2.5 million titles. The Booksense program also includes branding (slogan — "Book Sense — Independent Bookstores for Independent Minds"), a major national marketing program, as well as in-store promotional material and a gift certificate program — good at all participating bookstores. Cost per store — a $375 start-up fee plus $175 per month. Far less than most libraries pay for their catalogs, which don't come with national marketing programs.

The point here is that some of the major improvements we've seen in information retrieval on the Web may cost less than you might think. If libraries got together like the independent booksellers and devoted a small percentage of the resources tied up in virtual reference to improving the self-help features of our Web sites, we might have more to show for it. And unlike our reference services, Web self-service is eminently scalable. It costs Google or Amazon about the same amount of money to handle 6 patrons a day as it does 600 million. But answering questions one by one, as in library reference services, means even small increases in demand require large increases in costs that could quickly deplete the coffers of even the wealthiest libraries.

Let us link forces and apply some of the money and effort we spend trying to get patrons to ask us questions on the Web to make it easier for patrons to find the information they need on their own.

THE BOTTOM LINE

We've looked at the history of virtual reference. We've seen how libraries scrambled to move their reference services to the Web. We've documented what happened — or what didn't happen — once we set up shop there. We've examined how we might increase usage and/or reduce costs of existing chat reference services. We've taken a look at some of the alternative strategies and technologies that might accomplish the objectives we had hoped to achieve with chat-based virtual reference. We've done everything but answer the very fundamental question we set for ourselves at the beginning of this article ... "to chat or not to chat?" And the answer is a great big "it depends."

There are a variety of factors any library should take into consideration before deciding whether to start a new chat-based virtual reference service or to continue one already up and running. First and foremost, what are you trying to accomplish? In the early days, most of us would have answered that we were trying to do a better job helping our patrons looking for information on the Web. But there are other reasons a library might consider virtual reference service. A number of libraries and library consortia have used the technology to deliver specialized reference services — like Spanish-language reference or requiring subject expertise from business, law, and medical librarians — to libraries and patrons who might not otherwise have access to them. A very few libraries, like the Singapore National Library, have looked at virtual reference technology as a less-expensive and more-efficient way of providing remote reference service to smaller branches, including the completely self-service Sengkang Regional Branch. There may be other legitimate reasons, depending on the kind of patrons you serve and the type of information and assistance you want to deliver. There is also one not-so-legitimate reason for getting into virtual reference and that's because somebody says it's the "cool" thing to do or because everybody else seems to be doing it. Unfortunately, that factor has probably played a more significant role in many recent virtual reference start-ups than many of us would like to admit.

Whatever your purpose for considering chat-based virtual reference, the next critical question you need to answer is whether chat reference is the most effective way of accomplishing that purpose. Virtual reference is not an end in itself. It is a tool — a method of accomplishing a particular purpose. And it may not always be the best method. For example, if you want to do a better job serving patrons on the Web, you may well question the value of chat technology. With the growth of DSL, cable, and other high-speed connections, a growing number of people no longer need chat to communicate with each other online. A combination of phone reference and basic page-pushing or co-browsing technology, or a combination of phone reference and e-mail, may prove more effective and efficient for answering questions online. The Singapore National Library came to that decision when it designed its "reference kiosks" for the self-service Sengkang Branch. The kiosks feature a computer monitor with co-browsing capabilities coupled with a standard telephone, so the patron and librarian can talk while online, but with no keyboard and no chat. Originally, the system had a built-in videoconference facility so patron and librarian could see each other during the interaction, but that was dropped when it turned out that patrons did not like "seeing their own faces on the screen when using the service."

Also, significant contingents of the population are apparently not interested in chat as a communications tool, no matter how good the service behind it. These include lawyers and doctors and almost anybody else without a lot of time on their hands. LSSI had some abysmal failures with chat services designed especially for lawyers. Whatever the facts, there are clearly people who do not take to chat, and if that group populates the market you're trying to reach, chat may not be the most effective strategy. Even when dealing with a population of high school and college students known to use chat and IM heavily — at least among their friends — chat may still not work for effective reference service. Perhaps phone or a combination of phone and e-mail would work just as well if not better and certainly cheaper. Or maybe you should follow Google's and Amazon's model and invest in better Web self-service, while beefing up traditional reference services to handle those who still couldn't find what they needed on your site.

Maybe the communications channel doesn't matter. Maybe the problem is that your patrons know little about your existing reference services and what you can do for them. Maybe you don't really need chat at all, just better PR. Of course, there are situations in which some sort of virtual reference technology clearly is the best and most cost-effective solution to a particular problem. But we can only reach that conclusion after considering all the various ways of solving our problems. Too often in the past few years perhaps, we've jumped to the conclusion that virtual reference was the only solution.

You must also consider the impact of the service on your staff. Can you afford to hire new staff to handle the increased work load? If not, how do you propose to fit it into people's existing responsibilities and what's likely to suffer as a result? Will staff need a lot of additional training and practice to become proficient with the technology, or do they already have most of the needed skills? Just as importantly, what is your staff's attitude towards the new service? Are they enthusiastic and committed to seeing it succeed, or do they think of it as an administrator's pet project? Sometimes, in our rush to implement virtual reference over the past few years, we've forced more than a few librarians behind chat terminals who have no interest or desire to be there. And that can prove frustrating for the librarian and patron alike.

Of course, if you don't have the capacity or the enthusiasm among your own staff, you could always consider hiring your service out to a reference contractor in whole or in part. Before you choose that option, consider the costs of outsourcing compared with the costs of doing it in-house and the potential effect on the quality and character of your reference services. Here again, the staff impact can vary markedly across different methods of offering the same service. E-mail reference, for example, may be much easier to integrate into existing reference operations than setting up a whole new chat service that may require considerable training and continuous monitoring. Phone service has its own requirements, while improving Web self-service may reduce the workload on the reference staff. The important thing to remember here is that staff are a critical factor in any reference service. You need to weigh the effect the service may have on them just as carefully as you would any of the other factors discussed.

You also need to do some serious thinking about whether your reference staff, your library, and your library's budget are adequate to handle the amount of reference traffic you expect to get. Can your service scale to meet the potential demand? Is it sustainable over the long haul? A lot of us jumped into virtual reference in the early days with starry eyes, thinking that all we had to do was open up shop on the Web and in no time at all we'd be right up there with WebHelp — handling 20,000 questions per day. Of course that never happened, and it is a very good thing it didn't, because with an answer time of 15 minutes per question (which we now know is pretty normal for chat reference), standard staffing calculators show that it would take a crew of 252 reference librarians working 24 hours a day to handle that level of demand. Since most librarians simply refuse to work 24 hours a day (slackers!), we'd actually need to hire three 8-hour shifts of 252 librarians each, or a total of 756 brand reference staff, to handle the level of business many of us had hoped we'd get. There's no library — nor even group of libraries — at present that could have easily ramped up for that level of staffing, at least not without securing significant additional budget and other resources.

At an average rate of six questions per day, the effect of chat services on existing reference staff has been pretty negligible. But next time we may not be so lucky. We need to think carefully about the level of demand and amount of traffic we might reasonably expect to see if a proposed reference service succeeds. We need to know in advance where we will get the budget, staff, and other resources needed to handle success. In some cases, we may find the cost of answering questions using a particular reference technology simply too great; great success might bankrupt us. If that's going to happen, it's much better to know about it up front than several months or years down the road when you find yourself running out of money and facing the public relations nightmare of cutting off thousands of patrons from a service they've come to depend upon. Remember, it was not lack of traffic that killed most of the commercial reference services, it was lack of a sustainable business model to handle the traffic they got. Libraries are no different.

Finally, and perhaps most importantly, how do the costs and benefits of building and operating a chat reference service compare with the costs and benefits of other things the library could do with that same money? All libraries operate on pretty limited budgets; when we choose to do one thing, we usually have to choose not to do something else. Under those circumstances, librarians must keep their priorities straight to insure that they get the maximum mileage out of each dollar spent. So just how much is chat reference worth to us, or more importantly, how much is it worth to our patrons? Suppose that a library will have to cancel 50 current serial titles in order to afford a new virtual reference service? Is that trade-off worth it? If you answered yes, would it still be worth it if your chat service only averaged six questions per day? If not, how many questions would you need to justify the cost? And what if a large percentage of your chat patrons could have easily used other methods of contacting you — like phone or e-mail — and just chose not to do so? The cost may not always involve canceling materials. It could mean Sunday hours your library is not open or classroom instructions your librarians do not offer or Web sites that don't get built or updated, or any of a thousand of other levies both major and minor. But there's nothing unusual in that, virtually everything we do in a library comes at a price, the question each of us has to answer is whether that price is worth the value we received in return in comparison to the price and value of other projects, services, or collections we may have had to forgo to pay for virtual reference.

These are the critical questions librarians should ask themselves about virtual reference. There is no one right answer to any of them. Each is something individual librarians must figure out for themselves based on their own circumstances, the needs and characteristics of their patrons, and the staff and resources available. Nor are there any easy answers. There are often many different ways and technologies that could accomplish a similar purpose. It takes careful and deliberate consideration to figure out what is most appropriate and cost-effective for a given library and a given set of patrons.

The purpose of this paper has not been to argue for or against chat, but rather to suggest that libraries should approach the whole issue of virtual reference with much more careful and deliberate consideration. In the past few years much of the profession seems to have been gripped by a sort of "irrational exuberance" (to borrow a phrase from Alan Greenspan) about the prospects of virtual reference. Many of us saw it as a kind of panacea for all that ailed us. And for awhile there, it seemed that no matter what the problem was, or who the patrons were, or what we were trying to accomplish — the answer was always "chat."

The answer is not that we should now discard chat and go chasing after the next new technology to attract the attention of the reference cognoscenti. The answer is that we all need to give the whole issue of how best to provide reference services careful thought and analysis. We must weigh the costs and benefits of any new approach. We cannot be wooed by the technology or become terrified that the rest of the world would pass us by if we did not adopt the latest thing. Our funds are too limited and our reference and basic library services far too important to squander money on services that don't work. So let's try to be more careful and intentional as we look for the best ways to provide reference and library services in the future. Our patrons and the communities we serve deserve no less.

Linda Arret is an independent consultant specializing in digital reference services. She has extensive experience as a reference librarian in academic and government libraries, including 25 years' experience in leading and implementing systems-based reference services at the Library of Congress, where she developed generations of digital reference services, was the project coordinator for the early and pilot stages of QuestionPoint, and helped coordinate NISO's first steps toward standards in digital reference. Linda can be reached at linda.arret@verizon.net.