October 1997

CTI Implementation For The Help Desk

BY BILL RUDOWITZ

Help desk solutions are used to log calls, track inventory, escalate critical issues, dispatch support, and provide problem resolution and reporting. In theory, help desks (as their name implies) are supposed to help people. In reality, they are only as effective as the support system that is in place. Without CTI, most help desks are significantly less than helpful to the caller, and more aggravating to the staff.

The following example and evolution of CTI help desk integration took place at Magic Solutions in Paramus, New Jersey.

CALL PROCESS
Prior to CTI, callers to the help desk were greeted by one of two call coordinators who would answer the call. During the screening process, the coordinator would identify the calling customer. Once the customer account was identified, the call coordinator would open a new help desk call or locate the existing call if applicable. This process would take 30 to 45 seconds to complete. As multiple calls would enter the help desk, callers were greeted with “Please Hold,” and were placed on hold for an indefinite period of time by the call coordinator.

If the caller was calling to report a new problem, the call coordinator would determine whether the customer’s problem affected all users of the help desk system or if it was a problem of less severity. If the problem affected all users, the call coordinator would log the initial details and then dial or intercom various extensions to determine if an analyst could assist the caller immediately. If no analysts were available, the caller would be given a call number and told that an analyst would call them back as soon as possible. With all other calls, the problem would be logged, the call number given, and the caller told that an analyst would call them back without a specified time.

Thus, the call center was essential-ly a “Call-You-Back center. Invariably, when the analysts would call the customer back, the customer had stepped away from their desk or was on the phone with another call and the analyst would have to leave a voice mail message. When the client returned the call, they were informed that no analysts were available and someone would have to call them back. The call would be incremented in the system and the next available analyst would return the call. This would often become a vicious cycle of call back after call back.

Throughout this process, the help desk manager was required to monitor and manage the workload during the entire day, keeping an eye on the backlog of calls, which had been entered into the system and managing the staff to work the calls as effectively as possible. An escalation server was utilized to assist in this process, but still required a significant amount of human intervention to deal with the lack of direct access to the analysts.

THE GOAL
We sought to improve the process and to increase the level of customer satisfaction. We attended several user group meetings, met with the sales force, and worked with the technical support staff to understand what the obstacles were and how we might address them. The customers’ single largest complaint was the fact that they could never get through to an analyst on the first try. When they did finally speak to an analyst, the issue would either be resolved or the caller was provided the appropriate information. This circumstance caused great strain on meeting the needs of the customers and frustration for the technical staff. The staff simply did not have the tools to deliver the support they were fully capable of providing. The stress generated by both parties in this process only worked to heighten the frustration with each cycle of phone tag. The aim was to provide the customers with several means of reaching support, the primary method being direct telephone contact. Our challenge was to determine how to accomplish this and with which solution.

PREVIOUS ENVIRONMENT
The telecommunications environment included a pair of AT &T Merlin Legend switches. These switches were originally installed in the building in the mid- to late 80s, and were recently linked together via 4-channel tieline cards to accommodate the growth that the company had experienced over the past several years. The Legend switches had 24 incoming telephone and 24 outgoing lines supporting 192 extensions. The physical configuration of the technical support department did not provide the call coordinators with easy access to the analysts. Four-person cubicles were aligned in a linear fashion and were not in an optimum position to be seen from the call coordinator’s location. We had to find an automated means of distributing calls to the analysts in order to deliver them efficiently. LAN ENVIRONMENT The LAN was a fairly flat network running NetWare 4.11 in bindery emulation mode. The hubs were a pair of Synoptic 3000 series running on a 100-mps collapsed backbone. There were a total of 220 network ports in the building, split between 3 floors.

THE SEARCH
The company investigated the options in the marketplace for a solution to this problem. There were two distinct categories of solutions: PBXbased or LAN-based. After discussing the situation with the interconnect provider, it was quickly determined that the Legend could provide the functionality required without replacing the existing PBX. The investment required to make this change for the sake of the help desk was prohibitive.

The other alternative was to look at a LAN-based client/server application. This solution was appealing because it involved PC-based computer software. The concept was simpler to understand because the PBX now became a peripheral device on the network, like a printer or a modem. In this case, the situation dictated that we find a solution that utilized analog station ports. With the large number of LANbased solutions available, our task was to narrow our search and find a solution. The first-place search involved the customer base. We located a customer who had been faced with purchasing a new PBX if they did not implement a LAN based solution. This customer was a large government agency with a significant num-ber of in bound calls that needed timely answering and distribution . They used a helpdesk API (application programming interface) to generate screen pops of client information at the analyst workstations, in addition to providing skills-based call routing and a call back feature. The call back feature allowed callers to leave a telephone number where they could be reached when their request for service reached the top of the queue.

THE SOLUTION
The solution we selected was developed by SynreVoice Technologies in Markham, Ontario. The system consisted of a voice server and an ACD server. Both servers were Pentium 200-MHz machines with 32 MB of RAM and 1-gigabyte hard drives utilizing OS/2 Warp Connect and Dialogic voice processing cards.

THE IMPLEMENTATION
The implementation of this solution was to take place over a weekend, after the voice and ACD servers had been fully configured in advance by SynreVoice. We had several help desk personnel come in early on a Saturday and began to implement the CSR client on several of the analyst workstations.

Once the software was loaded on the desktops, we connected the voice server and ACD systems to the LAN and PBX. The testing went extremely well and required very little modifications. We then had to complete the desktop installations, set up priorities, and appropriately modify the analyst groups and queues according to the specific workflow.

The roll-out of the new system to the company’s customerbase required some consideration of possible customer reaction. We decided to implement a two-phased approach. The first phase involved queuing up callers for the call coordinators who would still enter the relevant information into the database. The coordinator would then transfer the caller and the call number to the appropriate queue for the next available analyst. The analysts were grouped according to their various skill sets for the system to find the best qualified analyst to respond to the caller’s problem. While callers wait for an analyst, they receive automated, realtime voice updates with their position in queue and the current maximum wait time every 90 seconds. Callers are also given the option to exit the queue and leave a voice message. While on hold, callers hear a combination of music mixed with informational prerecorded messages describing the different services offered by the company. When the call is transferred to the analyst, the information logged by the call coordinator is displayed at the analyst’s workstation. Analysts signal they are ready to take a call by clicking on an icon displayed on their workstation.

This phase lasted for 2 months. During this time, we solicited the opinions of our clients and received positive feedback. Customers felt their time was better spent because they could actually reach an analyst on their first call into the center. We used this information to refine the approach for the second phase. The second phase allowed us to redeploy the call coordinators and allow the customers direct access to the analysts through the CTI solution. This further enhanced the process by eliminating the middle person. Calls were delivered directly to the analyst queues. In addition, the customers now had familiarity with the system and the process and were better prepared to work with them.

The voice server was used to answer all incoming calls, identify the customer, and determine whether the call was a report of a new problem or a follow-up on an existing one. If the caller was following up on an existing problem, he or she would enter the call number. When the caller was transferred to an analyst, the specific call number — with all the call information — was displayed at the analyst’s workstation.

For new calls, the caller was presented with a separate menu. Upon making a selection, the caller was either transferred to an analyst or queued up. The menu had been refined from 4 queues (for the different products that were serviced) down to 2. This was done in an effort to simplify the process. Callers are now asked if all of their workstations are operative or not. If all the workstations are inoperative, the caller is placed in a high-priority queue for service.

The automatic call back feature was reintroduced to allow customers to reserve their position in queue. This reduced the help desk’s 800 number charges and allowed customers to continue with their work while they waited for an available analyst.

THE RESULTS
Since the implementation of the system, the backlog of open calls was reduced from 200 to 50 at any given time. The average time it takes to speak with an analyst dropped from 2 to 4 hours to an average of 10 minutes. This enabled us to prepare service level agreements for customers and measure performance across spans of hours, days, weeks, and more.

Another benefit was the decline in the stress levels of customers and analysts, due largely to shorter wait times and quicker connections. Realtime monitoring allows supervisors to view the number of calls in the queue and take appropriate action. Measurable goals were set for the help desk, and these have been met successfully with the CTI implementation.

Bill Rudowitz is director of Technical Support Services at Magic Solutions, Inc., a market leader in advanced help desk and asset management solutions to more than 4,000 organizations worldwide. A pioneer in the fast-growing help desk market, Magic Solutions is dedicated to providing software solutions that improve customer satisfaction and end-user productivity, while driving down the total cost of support. Magic Solutions’ products include SupportMagic, SupportMagic SQL, and SupportMagic/Web, which provides direct access to the help desk using popular Internet/Intranet browsers. For more information, visit the company’s Web site at www.magicsolutions.com.  

Internet Application For The Help Desk

One of the most important goals in implementing help desk technology is to ease the workload of the help desk technicians. Providing customers with direct access to problem solutions is one method of easing that workload. One approach is to analyze which call types are most common, then publish the solutions to those call types in an area where customers have easy access. The popularity and easy access of the Internet and World Wide Web provide an excellent place to publish these solutions.

et, investing in Internet and Web technology in and of itself does not solve the problem. You must remember that the content within the technology is also important. A useful Web solution should meet a number of criteria. First, the application should allow you to examine the customer base you wish to serve, categorize and author solutions so that the customers understand them, and tailor your solution database towards those customers. Writing the solutions begins with matching a defined problem to a problem defined in your solution database. Then, you should be able to describe the problem in terms that your customers readily understand. Knowledge base search engines are excellent tools for comparing the customer’s problem to your own problem descriptions.

Once the problem solution is defined, its appearance on a Web page becomes as important as the solution itself. Web pages must be both functional and attractive at the same time. Don’t fall into the trap of too much glitz and too little content. Yet, remember that a boring Web page may not hold a customer’s interest, and may even keep them from returning to search the knowledge base in the future. Along with the ability to distribute problem solutions, the Internet also provides a way to distribute applications, documentation, and other supporting files to customers. For example, to support a proposed solution, you may provide the latest documentation or a video to assist with an application. To support another solution, you may provide a program update that a customer can update and install.

The Web also provides an ideal capability to allow customers to keep track of their own call ticket history. This ability is especially helpful for customers with repeat questions. A useful Internet application should allow a customer to review old call tickets to search for previous problem solutions. Finally, a useful Web application should allow quick and easy communication with the help desk itself. Through HTML forms, the customer should be able to submit new call tickets, add memos and additional information to call tickets, or even change the status of call tickets.

Mike Rose is a software developer at Bendata, Inc., the developer of HEAT (HelpDesk Expert Authoring Tool), a leading call management/problem solving software. Bendata is also the developer of HEATLink to the Internet, software which allows Help Desks to author Web pages that link to their own home page and provide the ability for customers to open, view, modify, and close call tickets via the World Wide Web. For more information, visit the company’s Web site at www.bendata.com.