UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Simulation modeling as a decision analysis support tool : a case study at the PH&N telephone contact… Hiom, Paul 2000

Your browser doesn't seem to have a PDF viewer, please download the PDF to view this item.

Item Metadata

Download

Media
831-ubc_2000-0211.pdf [ 3.67MB ]
Metadata
JSON: 831-1.0089435.json
JSON-LD: 831-1.0089435-ld.json
RDF/XML (Pretty): 831-1.0089435-rdf.xml
RDF/JSON: 831-1.0089435-rdf.json
Turtle: 831-1.0089435-turtle.txt
N-Triples: 831-1.0089435-rdf-ntriples.txt
Original Record: 831-1.0089435-source.json
Full Text
831-1.0089435-fulltext.txt
Citation
831-1.0089435.ris

Full Text

Simulation Modeling as a Decision Analysis Support Tool A Case Study at the PH&N Telephone Contact Centre by Paul Hiom B.Comm , The University of British Columbia, 1998 A THESIS SUBMITTED IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF MASTER OF SCIENCE (BUSINESS ADMINISTRATION) in THE FACULTY OF GRADUATE STUDIES (Department of Commerce and Business Administration We accept this thesis as conforming to the required standard THE UNIVERSITY OF BRITISH COLUMBIA March 2000 © Paul Hiom, 2000 In p r e s e n t i n g t h i s t h e s i s i n p a r t i a l f u l f i l m e n t o f the re q u i r e m e n t s f o r an advanced degree a t the U n i v e r s i t y of B r i t i s h Columbia, I agree t h a t the L i b r a r y s h a l l make i t f r e e l y a v a i l a b l e f o r r e f e r e n c e and s t u d y . I f u r t h e r agree t h a t p e r m i s s i o n f o r e x t e n s i v e c o p y i n g o f t h i s t h e s i s f o r s c h o l a r l y purposes may be g r a n t e d by the head o f my department o r by h i s o r her r e p r e s e n t a t i v e s . I t i s u n d e r s t o o d t h a t c o p y i n g o r p u b l i c a t i o n of t h i s t h e s i s f o r f i n a n c i a l g a i n s h a l l not be a l l o w e d w i t h o u t my w r i t t e n p e r m i s s i o n . Department of CoflWoerce a n d Sus'^ gss fiA^yruS^TftxVVsO The U n i v e r s i t y o f B r i t i s h Columbia Vancouver, Canada Date Apol 27 .2 . 0 0 0 Abstract This thesis presents a simulation model for the Contact Centre at Phillips, Hager and North Investment Management, Ltd., to examine the impact of staffing, scheduling and process changes to the delivery of client service. The simulation is implemented using Scitor Process simulation and process mapping software as well as Microsoft Excel for reporting and statistical analysis. The results of the simulation give valuable insight into the operation of the contact centre, and provide management at Phillips, Hager and North with a means for testing changes to the Contact Centre without costly service interruptions. i i Table of Contents Abstract ii List of Tables and Figures iv I. Introduction 1 II. The Contact Centre at Phillips, Hager and North 3 A. Background 3 B. Current Contact Centre Structure 4 C. RRSP Season 6 III. The Initial Project 7 A. Forecasting Call Volumes 7 B. The Original Simulation Model 8 C. 1999 Call Volumes 11 IV. The New Contact Centre Design 12 A. Concept 12 B. Improvement Opportunities 13 C. Potential Client Interaction Models 15 V. Methodology 17 A. Literature Review 17 B. Choice of Methodology 18 VI. The Greeter Models 19 A. Mixed Dealer / Client Queue 20 B. Individual Dealer / Client Queue 22 C. Excel Reporting 24 D. Fund Advisor Analysis 24 VII. Results 26 A. Simulation Models 26 B. Fund Advisors Queue 27 VIII. Conclusions 29 A. Questions Answered 29 B. Value Added to the Company and Future Work 31 IX. References 33 X. Appendices A - H 34 iii List of Tables and Figures Figure 1 : Original Simulation Model Screenshot 9 Table 1 : Original Simulation Model Components 9 Figure 2 : Mixed Dealer / Client Queue Simulation Model Screenshot 19 Table 2 : Mixed Dealer / Client Queue Simulation Model Components 19 Table 3 : Service Times for the Mixed Dealer / Client Queue 21 Figure 3 : Individual Dealer / Client Queue Simulation Model Screenshot 23 Table 4 : Individual Dealer / Client Queue Simulation Model Components 23 Table 5 : Fund Adivsor Service Scenarios 25 iv I. Introduction The Phillips, Hager and North (PH&N) Contact Centre project provided three important lessons. First, the methodology for solving a quantitative problem should not be chosen solely on the basis of the mathematical value of the solution; acceptance and ease of implementation are both important factors. Second, the application of simulation methods inherently provides a deeper understanding of the business processes. Finally, a project that is completed with a focus on the greater business need, will invariably uncover further collaborative opportunities with the company. The choice of methodology for this project was heavily influenced by a number of factors that may not be apparent if only the contact centre problem is considered. The choice of simulation as a methodology was influenced by the incidental benefits derived from the transparent nature of the application of simulation. That is, simulation requires the practitioner to thoroughly understand the nature of the clients business, and to communicate that understanding in order to generate results. In addition, the interpretation of results is very clear to the company management, increasing their understanding of the challenges faced in the business area in question. In the case of this project, an additional factor was the introduction of the Scitor Process software as the standard modeling tool within the company. The success of this project has supported the wide-spread use of Process 98 for process mapping and an interest in more widespread application of simulation techniques. The simulation model built for the Phillips, Hager and North Contact Centre required data collection from three sources. The purpose of simulation, and the future direction of client services from the perspective of current management was required to understand the essential purpose and the granularity of answers required from the simulation. Second, an understanding 1 of the processes and the choices made by Contact Centre agents given different circumstances was required to determine the structure and logic used within the simulation. Finally, the application of statistical methods to determine distributions of call volumes, service times and transfer rates was necessary. The order of the data collection was important because it ensured that the results were based on the business requirements and not on the limitations of data. One of the early results of the project were recommendations for changes to data collection that would aid not only the current project, but also further understanding of activities within the contact centre. By the time the model was completed a deeper understanding, not only of the contact centre, but all interconnected departments was achieved. This understanding led to suggestions for future projects that would add value to Phillips, Hager and North, but were not within the scope of the current initiative. Two of these suggestions, a further in-depth analysis of the entire client services area, and a scheduling tool based on application of the initial forecasting model were accepted by Phillips, Hager and North, and represent current projects undertaken within the Centre for Operations Excellence partnership program. 2 II. The Contact Centre at Phillips, Hager and North A. Background Phillips, Hager and North (PH&N) is an investment management firm, formed in 1964 by Art Phillips, Bob Hager and Rudy North. The firm now manages approximately $33 billion dollars in assets, the majority of the assets being pension funds for successful Canadian companies. An area of the business that has been expanding in recent years is the sale of retail mutual funds. Currently, the retail mutual funds business accounts for about 15% of gross revenues, but supporting this type of clientele is more labour intensive than supporting the traditional institutional pension fund clientele. As an illustration, P H & N has approximately 150 institutional clients who represent about 75% of total investment dollars. By contrast, the 10% of assets held by mutual funds clients, are held by about 15,000 different investors. A PHN retail mutual fund client holds either a regular investment account or an RRSP account (or both), with a minimum investment of $25,000 for a regular account and a $10,000 minimum for an RRSP account. The investor has a choice of approximately 10 different mutual funds, such as the Canadian Equity, US Equity, Balanced, Bond or Euro-Pacific Equity Funds. PH&N fund advisors will make portfolio recommendations to the majority of investors by phone or in person, but the investors are responsible for managing their own assets across the available funds. A second group of mutual fund investors purchase P H & N mutual funds through brokers. The brokers are responsible for all client interactions and PH&N transacts only with the brokers. 3 B. Current Contact Centre Structure The customer service representatives at the PH&N contact centre are responsible for all telephone contact with investors that is not related to fund advice. The types of services that are performed by the contact centre agents include: providing information on account balances or the status of transaction requests, providing product information and administrative activities surrounding a mutual fund account, such as change of address requests. In addition, the contact centre is responsible for making outbound phone calls for the client processing department when client transaction instructions are unclear. Currently, for security reasons, all transaction orders are received from clients by mail or fax. It is expected that these restrictions will be relaxed within the next year to allow clients to place buy or sell orders either by telephone or through the Internet. Throughout this study, there were six agents assigned to the Contact Centre. These six agents are responsible for providing telephone coverage from 7 am to 5 pm. The agents are assigned to different shifts; either 7am - 3pm, 8am to 4pm, or 9am to 5pm. The agent shift times are determined to ensure more staffing during the peak demand periods, which occur between 9am and 1 lam. The average daily demand curve is shown in Appendix A. While answering incoming calls is the main function of the contact centre, agents are expected to perform a variety of other duties, such as dealing with walk-in clients, responding to e-mail questions, researching complex questions at the client processing area, assisting fund advisors and completing special projects. The distraction caused by the additional duties makes staffing to meet demand difficult, because the number of agents available to take incoming calls is not constant. If two agents are on the 15th floor researching client questions, and two agents are on lunch, the two remaining agents will probably not be enough to deal with incoming call volumes. 4 However, this may not be obvious management, because the telephone system will report that six agents had been available during the time interval, more than sufficient staffing to deal with call volumes. The contact centre has also seen a recent increase in the number of calls from mutual fund dealers. Mutual fund dealers are allowed to sell PH&N funds, but do not receive any commission from PH&N. The dealers are able to place orders directly through the PH&N computer system, but tend to call fairly frequently to verify that transactions have cleared, or to fix problems with their orders. PH&N has two staff members who support the dealers, but a number of the phone calls come through the contact centre. A survey was completed in April of 1999, which showed that 29% of contact centre calls were from mutual fund dealers. (Seegers, 1999) The survey also showed that the call pattern for dealer calls was heavier during the early morning hours than later in the day, presumably due to the number of dealers located in Eastern Canada. Finally, calls for fund advisors were moved to a separate telephone queue in January of 1999. This queue has a similar call pattern as the contact centre, but has fewer calls, that are much longer on average, 7 minutes, compared to 3 minutes for the contact centre. PH&N currently has six to eight agents assigned to this queue, enough to handle the average incoming call volumes while meeting the service target of answering 80% of the calls in 20 seconds or less. However, the agents also make a large number of outgoing calls each day. If the time spent on these outgoing calls is removed from the available agent time, staffing volumes of six to eight agents tend to be overstated compared to Erlang formula estimates based on the volume of incoming calls. 5 C. RRSP season The daily demand pattern (see Appendix A) is fairly constant, but there is a distinct seasonal pattern with call volumes roughly doubling during the month of February. This period of increased activity coincides with the deadline for the previous years RRSP contributions. During 1998 the RRSP season resulted in phone service times that were unacceptable (see Appendix B). In December of 1998, PH&N commissioned a study to ensure there was not a re-occurrence of the 1998 RRSP service problems. 6 III. The Initial Project The initial project was completed over the period from December 15', 1998 to January 5 , 1999. The goal of the project was to build a forecasting model to predict call volumes for the contact centre, and to use the forecasting model in conjunction with a process simulation to compare process changes to determine there impacts to the contact centre. A. Forecasting call volumes The forecasting model was built using only one year of data because data previous to December 15, 1997 was not available. The data provided call volumes by half-hour interval.1 The initial forecasting model was built using multiple linear regression, with dependent variables representing the twenty half hour periods from 7 am to 5 pm, the twelve months of the year, and the five days of the week. The model captured daily and weekly trends well, but did not adequately capture the RRSP season described in section II.C. To improve the model, an additional variable was introduced to represent RRSP season. This variable increased by 1 for each day from the 1st of January to the 28 th of February. With the introduction of the RRSP variable, the model provided forecasts consistent with management expectations for both day of week and time of day call variations. A further improvement to the model was made by changing the estimation method. The use of maximum likelihood estimation based on a Poission regression model fit the sample data more accurately. The model was fitted using a macro written for Microsoft Excel. The results of the forecasting model are given in Appendix C . 1 Data previous to May of 1998 was collected for one-hour intervals and was divided by two to generate half hour volumes. 7 B. The Original Simulation Model The initial simulation model was built using Scitor's Process 98 simulation software. The choice of software was made to support the introduction of the Process 98 as the standard modeling software at Phillips, Hager and North. The software is more difficult to use for simulations than Arena or other simulation-specific commercial simulation packages. For example, Process 98 does not support multiple simulation runs, and some of the data export functions do not work properly. However, Process 98 is much more reasonable financially, for a company that has only a limited number of uses for such an application. Process 98 is also well suited for process mapping activities, the main purpose that it serves at PH&N. The initial simulation model was used to simulate the contact centre during RRSP season given the existing configuration. For this purpose, five different processes had to be simulated; incoming phone calls, outgoing calls, correspondence with clients, walk-in clients, and special service requests that had to be performed on the 15th floor. The forecast discussed in section A above was used to provide the input parameters to simulate incoming calls. The calls were modelled using an exponential inter-arrival rate that varied with time of day, day of the week, month and RRSP season. The Outgoing calls were also modeled from call data, while walk-ins, correspondence and special service requests were estimated from data collected manually by contact centre agents. This data was very limited, but the project time frame did not provide time for further data collection. An important component of the model building was describing and encoding the logic used by contact centre agents to prioritize the tasks available. We decided to make walk-in clients the highest priority because they represent clients who are waiting at the reception area. The second highest priority tasks were incoming phone calls, with outgoing calls, special services and 8 Figure 1 : Original Simulation Model - Scitor Process 98 Screen Shot 3 1 ASA 2.64m Abandoned Calls -X ACD Qwpe i Walk-In C lie rit Outbound Calk J Correspond ence TSF 87.4346 ASA 0.37m "^Correspondence' ^Queue-Table 1 : Original Simulation Model Components Model Component Description Incoming Calls Input Queue for incoming client calls - based on forecast model detailed in section Ill.a. Inter-arrival rates are specified by half hour interval based on day of week, hourly and seasonal factors. Walk-in clients Inter-arrival rates for clients requiring personal assistance. Walk-in clients are the highest priority of all service requests. Outbound calls Contact centre agents make a number of outbound service calls. These calls are given low priority in the model because agents can make the calls when the timing is most convenient. Correspondence Contact center agents are responsible for answering e-mail and regular mail requests. The priority of correspondence calls is below that of outbound calls. ACD Queue Calls waiting for service by a contact centre agent. Statistics are collected on queue length and average weight time (ASA) Abandoned Calls Abandoned call rates were modeled from historical data. The probability of abandonment is based on call waiting time. Contact Centre The Contact Centre resource is modeled with the number of agents required based on the forecasting model and Erlang C staffing projections. Agents (simulation resources) are added or removed based on the time of day and day of week Special Service A small percentage of calls require follow up work by a contact centre agent. This service receives lower priority service than incoming calls, but is higher priority than the correspondence or outbound calls. 9 correspondence left until there were at least 2 agents unoccupied. This approach allowed us to represent the idea that the agents were able to perform the last three tasks at the most convenient time, given current workloads. The completed model could be run for any day of the year, and presents the user with estimated of the expected number of calls, the maximum queue length, expected telephone service factor (TSF), and average speed of answer (ASA) for the contact centre. It also gives users the expected number of outgoing calls, walk-in clients, special service requests and correspondence activities. The average length of time to service each of these requests was not an output of the model, because it was used as an input to generate the model. The service times for each of the tasks were taken from the data available. Incoming calls were modeled with an average three-minute service time and a two-minute follow up time, for a total of five minutes, exponentially distributed. The outgoing calls from historical data averaged about three minutes, also using an exponential distribution. The other three types of service did not have data available. Approximations from contact centre staff and management placed the time to service each of these types of requests at about 15 minutes. For the purposes of this simulation, this value was taken to be reasonable, and these three services were modeled using a normally distributed service time with a mean of 15 minutes and a standard deviation of 5 minutes. The working simulation was tested using two arrangements. First, the contact centre was simulated using the previous years' staffing levels. The resulting customer service level for the month of February proved to be consistent with the poor service level found during February of 10 1998. Secondly, the model was tested with increased staffing and no walk-in clientele2, a scenario that PH&N were thinking of implementing for RRSP season. This scenario resulted in a significant improvement to service levels with the average speed of answer (ASA) improving from 80 seconds to 55 seconds. However, the contact centre would still not make it's service target of TSF80, or 80% of calls answered in 20 seconds or less, unless call volumes were significantly lower than the previous year. For a glossary of telephone call centre terminology see Appendix F. C. 1999 Call Volumes The results of the simulation and call forecaster were well received by PH&N management, however, call volumes for 1999 were significantly lower than 1998. Many factors have been suggested for this decrease, with the two most likely culprits being the general decrease in activity in the mutual fund market, and an improvement in client statements from PH&N resulting in fewer clarifying phone calls. One very positive factor for the call forecaster is that the call volumes are consistently lower than the forecasts, with the day of week, and half-hour interval factors remaining relatively constant. One of the variables that was considered for the original forecast model was a yearly trend. However, the lack of data made the addition of this variable impossible. With the addition of an additional six months of data, a yearly trend variable was introduced in June of 1999. The yearly trend resulted in significant improvements to the accuracy of the forecasts . PH&N management were sufficiently impressed by the results to examine potential staffing and process changes to the contact centre, and ask for further analysis to be performed using simulation. The potential 2 The responsibility for walk-in clientele was to be shifted to another department within the company. 11 designs for the Contact Centre are detailed in section IV, while the methodology and results of the simulations of the potential designs are detailed in section V through VII. 12 IV. The New Contact Centre Design A. Concept The key to the success of Phillips, Hager and North is combining solid financial returns with the best customer service, at reasonable prices. This formula has been extremely successful for dealing with both institutional clients and very wealthy private clients, most of whom were former executives introduced through the institutional business. The addition of the retail mutual fund era has threatened the outstanding customer service that previously defined the culture of PH&N. The sheer volume of clients and the different nature of those clients has forced Phillips, Hager and North to rethink their customer service vision. In the spring of 1999, a committee consisting of the Vice President, Technology Services, the Chief Operations Officer, the Manager of Client Services and the Manager of Advisory Services was formed to examine the manner in which PH&N deals with telephone contact with mutual fund clients. The committee decided that the current service level, represented by a TSF20, (80% of calls answered in under 20 seconds) was not in keeping with the quality of service that PH&N clients should expect. The committee also expressed concern that the number of calls that were reaching fund advisors was not high enough, and finally they were concerned with the high staff turnover rate in the contact centre. B. Improvement Opportunities The service target of TSF80/20 is a standard used by most incoming call centres. This standard means that 80% of customers calling will be connected within 20 seconds. For most companies this standard is acceptable, but PH&N has decided that this service level does not reflect the 13 quality of service that they wish to provide to clients. The trade-off to providing a higher level of service is a reduced occupancy rate for Contact Centre agents, and therefore a higher number of staff required to service the available calls. The exact cost of decreasing this service target will be determined later. A second area for improvement was the number of calls transferred to fund advisors. Concerns were expressed that clients were talking to a contact centre agent about account balances or transactions, an activity that naturally leads to sales opportunities. Unfortunately, the contact centre agents are not licensed to offer fund advice to clients, only advisors who have completed the CFC course are allowed to provide investment advice. It is a commonly held belief that any client contact can become a sales opportunity, and this is the approach that PH&N would like to consider. Rather than answering the client's question and then terminating the call, a potential improvement would be to transfer such calls to fund advisors early in the conversation to increase the number of chances for sales of mutual funds. One drawback to this model is the requirement for increased number of fund advisors, who are much more expensive than contact centre agents. The committee is also concerned about the effect that this switch would have on the morale of the fund advisors who traditionally have more education, are more highly skilled and are paid better than contact centre agents. If this group felt that they were performing a task previously performed by contact centre agents without recognizing the potential business benefit, staff problems would probably ensue. The third potential area for improvement is the high turnover rate for contact centre agents. One way to improve the contact centre turnover rate is to make the job more interesting. To do this, more variety will be introduced into the job, by reducing the amount of time spent on the phone queue, and allowing contact centre agents to perform a variety of other tasks, during the non-phone time, such as research for fund advisors. An added benefit expected from this kind of 14 specialization, is an increased focus on incoming phone calls during the designated phone time. Previously, agents determined the best time for outgoing calls or special requests. The choices made by agents were dependant on the current call volumes, and would be inferior to the recommendations for staffing levels provided by a forecasting model. C. Potential Client Interaction Models The PH&N management committee produced two potential models that they wished to evaluate. The two models are very similar. In both cases the following changes were proposed: 1) The service target was adjusted to 95% of calls being answered by a live operator with five seconds. 2) Call flow patterns were altered, with contact centre agents transferring the majority of calls to fund advisors. The service times for transferred calls were reduced, because a transferred call tends to be shorter than a client inquiry. 3) Contact Centre agents would spend only a portion of their day on incoming phone calls. 4) Al l of the non-queue jobs performed by contact centre agents were eliminated. These jobs were assumed to be completed by the contact centre agents not currently assigned to the phone queue. The two models differed in their treatment of the dealer calls. In the first model, dealer calls were answered by the contact centre, but transferred to the dealer representatives in the same manner as fund advisor calls. In the second model, the dealer calls are answered by a separate queue comprised of the dealer service representatives. This model will presumably provide better 15 service for the dealers, but will result in lower occupancy rates, and therefore more agents required overall. V. Methodology A. Literature Review In order to determine the most applicable methodology for the project, recent literature in both simulation modeling and call centre analysis was surveyed. The literature showed that both simulation and queuing theory are used extensively in the understanding of call centres. Whitt (1999) examined the importance of queue wait times to client satisfaction. Whitt stresses the importance of providing waiting time information to the client at the beginning of their hold time. The relatively low volume of calls to the PH&N call centre makes this kind of reporting difficult, but Whitt does highlight the importance of consistency to the provision of good customer service. Whitt (1999-2) also provides a methodology for the calculation of customer wait times in a service queue environment. The application of queuing theory with exponential service and inter-arrival times is relatively straightforward, and would be applicable to the PH&N environment. However, Grossman (1999) highlights the importance of process driven simulation in education about queue behaviour. Grossman suggests that the application of a simple simulation will provide students (or management and staff) with an intuition and greater understanding of the call centre process. In addition, a simulation is a simpler tool for managers to apply after the completion of the project. Sanegre (1998) provided an example of the application of scheduling to call centres, however the use of integer programming required expensive software tools that were unsuitable for the size and scope of the PH&N Contact Centre. Mason, Ryan and Panton (1998) provide support for the 17 use of simulation and forecasting in the scheduling of staff. Their model used heuristic approaches to schedule staff for an airport customs queue. This application provides suggestions for a similar scheduling tool within the Phillips, Hager and North contact centre. Finally, Browning (1998) suggests that the service times for subsequent queues should not necessarily be modeled independently from the initial queue. This work would suggest that an analysis of the greeter model with subsequent queues for both dealers and advice may require dependent call service times. Unfortunately, data was not available to substantiate such an approach, so an assumption of independence was made in service times. Future work in the contact centre should re-examine this assumption. B. Choice of Methodology Just as with the first model, there was significant pressure to use the Scitor Process 98 software for analysis of the two scenarios. Although the questions that had been posed seemed fairly straightforward, there still seemed to be a number of advantages to the use of simulation. First of all, there was some concern over the accuracy of queuing theory and Erlang formulas given the relatively small volume of calls. Secondly, the use of simulation allowed for the graphical representation of the processes being evaluated. This ensured a consistent understanding and interpretation of the results derived from the simulation. Finally, the use of simulation techniques presented an exceptional means for educating the staff who would be impacted about the reasons for the changes that would be occurring within their department. 18 VI. The Greeter Models Figure 2: Mixed Dealer / Client Queue Simulation Model - Scitor Process 98 Screen Shot Mixed Dealer / Advisor Model - July 2, 1999 Agents W o r k i n g 2 Greeter O c c u p a n c y Contact Centre y o x . Contact Centie,-Incoming Cal ls 2 1 2 A S A : 0 sec A n s w e r e d within 3 r ings : 98 .58%| M a x Wai t : 0 . 43m Dealer R e p • " S e l e c t " ' ' 1 9 % / 85 > Completed Calls j A d v i s o r Select Dea le r R e p #1 Dealer Advisors 4 0 A s siste d D e aler Calls Fund Advisors V 107 fl A d v i c e A d v i s o r #1 A d v i s o r #4 :  Cal ls J [ ^ A d v ^ H O c c u p a n c y : 22 % A d v i s o r #3 A d v i s o r #5 0 O v e r f l o w Table 2: Mixed Dealer / Client Queue Simulation Model Components Model Component Description Incoming Calls / ACD Input Queue for incoming client calls - based on forecast model detailed in section III.a. Inter-arrival rates are specified by half hour interval based on day of week, hourly and seasonal factors. Contact Centre Queue Target service times for the contact centre queue are set to 95% of calls answered within 5 seconds or less. Contact Centre The Contact Centre provides two services. First, calls that should be handled by either a dealer representative or an advisor are transferred to the correct queue. Second, simple calls are answered within the contact centre. Dealer Advisors Dealer Advisors provide specialized service to mutual fund dealers. The calls are of a very different nature to the typical client calls. Fund Advisors Five advisors are represented. Calls to this area are not queued, instead, the overflow advisor represents multiple calls and provides the maximum number of simulationeous advisor calls. Fund advisor queuing can be simulated separately as required. Statistics Process 98 provides only summary statistics, so data was exported to Excel for reporting purposes. Only simple summary stastistics are shown within the simulation interface. 19 A. Mixed Dealer / Client Queue The model contains three distinct components, a greeter queue (the revised contact centre), a dealer advisor area and a fund advisor area. The initial composition of the model was designed to evaluate the service level for the greeter queue, and the resulting call volumes transferred to each of the two attached areas. Further refinements would provide a greater understanding of the fund advisor area. The greeter area has an automated call distributor (ACD) with an exponential inter-arrival rate based on the forecasting model determined in the initial project (January 1999). Calls are processed on a first-come first-serve basis, with excess calls stored at the diamond-shaped greeter queue object. The number of agents available is modeled based on the Erlang C formula. (Erlang C gives the number of agents required to staff a call centre given the number of calls, a time interval, the required service level and the average call length.) The number of agents used in the model, and Erlang's formulation are described in appendix D. The number of agents is based on both the time of day and the month of the year. The day of the week was found to have only a small impact on the number of agents required, and was therefore omitted for the sake of simplicity. The subsequent route and service times for calls at the Greeter object are determined randomly. Calls are routed to either completion, fund advisors or dealer queues based on historical data. The routing to the dealer queue is based on the results of the Dealer Call Survey (Seegers, 1999). The dealer routing is time of day dependent. Routing of the non-dealer calls in the original model is 90% transferred to the Fund Advisors and 10% completed by contact centre agents. This routing probability is not dependent on the time of day or day of week. These numbers were 20 adjusted to 80/20 and 70/30 in subsequent evaluations of fund advisor call volumes. Service times were modeled as follows: Table 3 : Service Times for the Mixed Dealer / Client Queue Call Type Location Average Service Time Distribution Completed by Contact Centre Contact Centre 3 minutes Exponentially Distributed Transferred to Fund Advisor Contact Centre 1 minute Normal Distribution* standard deviation of 30 seconds Transferred to Fund Advisor Fund Advisor 7 minutes Exponentially Distributed Transferred to Dealer Services Contact Centre 1 minute Normal Distribution* standard deviation of 30 seconds Transferred to Dealer Services Dealer Services 5 minutes Exponentially Distributed * truncated to omit non-zero values The exponential distributions were estimated using historical call data available from the A C D queue statistics from 1998. The Contact Centre service times for transferred times were provided by the Contact Centre management based on call time expectations for the new operating model. Calls transferred to the Fund Advisor queue are allocated to the Fund Advisors in order from #1 to #5 followed by overflow. The overflow advisor represents multiple additional advisors and has an unlimited capacity for calls. This method does not allow for calculation of occupancy for individual advisors or for detailed analysis of any queuing that may be required for Fund Advisors. However, it does allow for a graphical representation of the volume of activity for the Fund Advisor area while the model is running. As each agent becomes busy the color of the rectangle representing the agent will change color. Busy periods are easily apparent when a large number of rectangles have changed to red. 21 This approach also allows us to easily determine the maximum number of agents who would be required at any point of time during the day because the number of simultaneous services by the overflow object provides a count of the maximum number of Fund Advisors required at any point in time. An estimation of the total number of minutes of call time required from Fund Advisors can also be calculated from the simulation output data by summing the Fund Advisor effort times throughout the simulation run. B. Individual Dealer/Client Queues The individual queue model is very similar to the Mixed Dealer / Client Queue with the following exceptions. Dealer calls are not routed through the Contact Centre Queue. Instead, these calls are directed to Dealer Advisors by the A C D . This would probably be achieved through the use of an alternate phone number, because one of the goals of this activity is to reduce the reliance on automated phone systems. In addition, the service time for dealer calls is reduced from 5 minutes to 3 minutes to account for the improved specialization of the dealer agents (per P H & N management). With a small number of agents devoted entirely to servicing the dealers, it is not unreasonable to expect a large improvement in the speed of the service provided. 22 Figure 3: Individual Dealer / Client Queue Simulation Model - Scitor Process Screenshot Two Queue M o d e l - July 2, 1999 Agents Working 3 K X Contact Queue Contact Centre Dealer Queue Max Delal Agents Remaining: 2 ASA: 2 sec Answered within 3 rings: 95.88% 1.88m Agent Occupancy 19% / " Completed Callsj Advisor Selett Dealer Reps Dealer f 191 Advisors l ^ o m p l e t e d Dealer Calls^ Fund Advisors Advisor #1 V Advisor #2 Advisor #4 f Advisor #5 Occupancy: 36 % Advisor #3 1 Overflow Table 4: Individual Dealer / Client Queue Simulation Model Components Model Component Description Incoming Calls / ACD Input Queue for incoming client calls - based on forecast model detailed in section Ill.a. Inter-arrival rates are specified by half hour interval based on day of week, hourly and seasonal factors. Contact Centre Queue Target service times for the contact centre queue are set to 95% of calls answered within 5 seconds or less. Contact Centre The Contact Centre provides two services. First, calls that should be handled by an advisor are transferred to the correct queue. Second, simple calls are answered within the contact centre. Dealer Advisors Dealer Advisors provide specialized service to mutual fund dealers. The calls are of a very different nature to the typical client calls and are queued separately from the Contact Centre calls. Fund Advisors Five advisors are represented. Calls to this area are not queued, instead, the overflow advisor represents multiple calls and provides the maximum number of simultaneous advisor calls. Fund advisor queuing can be simulated separately as required. Statistics Process 98 provides only summary statistics, so data was exported to Excel for reporting purposes. Only simple summary statistics are shown within the simulation interface. 23 C. Excel Reporting One serious drawback to the use of Scitor's Process 98 as a modeling tool is the lack of reporting available with the software package. Reports are very rudimentary, and no support is provided for statistical evaluation or multiple runs of the simulation. To address this issue a Microsoft Excel based reporting tool was created. An extract file is created manually from each run of the simulation software in Process 98. The resulting extract file is recorded as a transaction log in an Excel worksheet. This Excel worksheet is the underlying data for a pivot table that allows the user to select a specific day of data and report statistics for each hourly interval of the day. Reports are generated dynamically as the user selects a new day from the set of sample runs extracted from Process 98. The reports are designed to mimic the presentation of call statistics from the ACD. For the two queue arrangement, statistics provided for both the advisor and the dealer queue include: the number of agents working the queue, the number of calls accepted, average talk time, occupancy rate, ASA, the longest delay and TSF. The mixed queue provides a single measure for each of these variables as well as the number of calls and total minutes of calls transferred to each of the other two areas, Fund Advice and Dealers. Examples of a Simulation ACD Queue Statistic Report are included as Appendix E. D. Fund Advisor Analysis An integral question to the new service model is the number of advisors that would be required. The simulation provided some statistics for advisors, the number of calls by hour and the maximum number of simultaneous agents required. Three choices were available to provide these answers, either a new simulation model, a modification to the existing simulation model, or 24 a straightforward application of the Erlang C formulas. Due to time constraints, the lack of historical data for accurate modeling, and confidence generated from the quality of the Erlang staffing predictions in the previous two models, the number of fund advisors was calculated using the Erlang C formulas. An Excel application was created to compare agent requirements given nine different scenarios. The nine scenarios are as follows: Table 5 : Fund Advisor Service Scenarios Scenario # Telephone Service Factor Expected Average Call Time % of Calls Transferred from Contact Centre 1 95% of calls answered in 5 seconds or less 8 minutes 70% 2 95% of calls answered in 5 seconds or less 8 minutes 80% 3 95% of calls answered in 5 seconds or less 8 minutes 90% 4 80% of calls answered in 20 seconds or less 8 minutes 70% 5 80% of calls answered in 20 seconds or less 8 minutes 80% 6 80% of calls answered in 20 seconds or less 8 minutes 90% 7 95% of calls answered in 5 seconds or less 7 minutes 70% 8 95% of calls answered in 5 seconds or less 9 minutes 80% The scenarios were tested for 12 sample days, one from each month of the year. The results show a maximum, minimum and average number of agents required for each day as well as the total number of expected calls. Results are shown graphically for varying call transfer percentages, average call times and telephone service factors. The results of the study will be discussed later. 25 VII. Results A. Simulation Models The results of the simulation models were presented to the Manger of Client Services and a Client Service Representative on July 8, 1999. The presentation focussed on the a comparison of the two queuing arrangements for number of agents required, variability of service time, sensitivity to changes in call volumes and projected occupancy rates. The number of agents required was found to be between 22% and 50% higher for two individual queues compared to the mixed queue. Currently, there are two dealer representatives at Phillips, Hager and North. The simulation model showed that three agents would be required as a minimum to handle dealer calls and still achieve a TSF of 80% in twenty seconds. Even with three full-time agents (7.5 hours each), the variability in service would be high, because the number of calls offered is too small. The mixed queue was much less variable in performance than the two individual queues, because the number of calls is much higher. A comparison of service levels for the Erlang recommended staffing levels above revealed some interesting information. With a service target of 95% for the two-queue model, the Advisor queue achieved the target in 81% of the 120 one-hour time periods evaluated. By contrast, the dealer queue was only able to achieve the TSF95 target about half of the time. The dealer queue was also much more variable in terms of the level of service delivered. By contrast, the mixed queue had an acceptable service level 79% of the time, and the lowest TSF achieved was 77%. 26 Sensitivity analysis was conducted on the results for call volumes and percentage of calls transferred from the greeter queue. The impact of changes in call volumes was as expected, an increase from 250 to 500 calls per day resulted in less than double the number of agents required, and an increase in occupancy rates. However, a decrease in the percentage of calls transferred from the greeter queue had a serious impact on the level of service provided by the queue. Transferred calls were modeled with a shorter service time than completed calls, so a decrease in transferred calls resulted in a decreased service level from the greeter queue, an increased occupancy rate and overstaffing at the advisor queue. Occupancy rates associated with both the mixed and individual queues were low, with the individual queues having an occupancy rate of approximately 20% and the mixed queue having an occupancy rate of approximately 30%. In all cases, the occupancy rate would increase with an increase in call volumes. An occupancy rate this low was not expected by the Contact Centre management, and was the subject of a number of questions. For management, the idea that 30% productivity was the only way to reach the desired service target was very illuminating. B. Fund Advisors Queue Results for the Fund Advisor Queue Analysis are shown in Appendix G. The projected agent requirements for each of the nine scenarios in section VI.D are expressed as a minimum, maximum and average number of agents per ten-hour day. The number of expected calls for each example for each of the twelve months is also shown. Sensitivity to each of the three major variables, call length, % of calls transferred and Telephone Service Factor is also presented in Appendix H. An increase of one minute in call length tends to increase the average agent requirements for a given period by about 10%. An increase of 10% in 27 the percentage of calls transferred increases agent requirements by about 10%, and finally a decrease in the TSF from 95% / 5sec to 80% / 20sec decreases the agent requirements by 15-20%. 28 VIII. Conclusions A. Initial Questions to be Answered At the start of this project, we created a list of questions to be answered for this project to be successful. The questions, along with the answers that we found are given below: Can we reproduce past performance levels to validate the simulation? The initial Contact Centre model produced results that management felt were consistent with past performance. Contact Centre staff and management confirmed that the model structure captured the important elements of the Contact Centre operation and structure. In addition, management from other areas of the firm were able to see some of the difficulties associated with staffing and operating a Contact Centre, and the burden that different requests, such as special projects and walk-in clients, places on the achievement of satisfactory call statistics. What will be the impact of RRSP season on the Contact Centre if no modifications are made? Results were mixed, I believe the analysis was correct, but demand was below expected volumes, leading to service exceeding the expectations of our simulation model. 29 How many agents would be required to achieve performance levels, given the current Contact Centre system? The model uses Erlang C formulas, historical call length and forecasted hourly call volumes to estimate the required agents for a given level of demand, see Appendix D. What would be the impact of staffing, scheduling or process changes on the level of service at the Contact Centre? The two process models, along with the fund advisor analysis, addresses the impacts of potential changes in staffing and processes (call flows). A potential project evolving from this work would be the development of a scheduling program to ensure staffing to meet Contact Centre requirements and the evaluation of differing interactions between the Contact Centre, the fund advisors and transaction processing groups. What impact will potential Contact Centre changes have on demand? The process models and the simulation methodology show that simulation can provide useful information to support process changes. However, it will not provide information on changes to demand patterns. One example that was discussed at length, is the extension of Contact Centre hours to further accommodate Eastern Canadian clients. If estimates for the adjusted demand patterns are given, the model can be used to test for required service and staffing levels, but simulation will not provide any clues to the effect that process changes will have on Contact Centre demand. A potential project to model call volumes based on adjusted hours of operation may be of interest for future work. Such a model could incorporate research from a variety of factors external to PH&N to produce call estimates. 30 Would more flexible scheduling practices lead to improved customer service? A future project will address the improvements that can be made in customer service, but the expectation from the simulation model is that the Contact Centre can achieve the required service targets with fewer hours devoted to incoming calls, if the time is allocated more effectively throughout the day. Are there key indicators of daily demand that can be tracked? Are there day of the week, day of the month or seasonal indicators of demand? The forecasting model indicates that day of the week, time of day and seasonal factors are extremely important to the level of demand at the Contact Centre. There was also evidence that the difference between the current forecast and the actual call volumes provides some insight into any short term variations in call patterns that may exist. This information was not required for the simulation, however, if the forecast model is adapted to support a scheduling application such autoregressive factors should be closely examined. B. Value Added to the Company and Future Work The value of the analysis to Phillips, Hager and North is not easily measured in dollars and cents. Unlike the majority of COE projects, the focus is not on the reduction of costs, but instead, on the improvement of customer service levels. To this end, the project has provided decision support and education to PH&N management, and has increased awareness of issues within customer service. Perhaps the best support for the success of the project is the commencement of two new projects, development of a scheduling tool for Contact Centre staff and an analysis of the processes of the entire client service processing area. 31 The scheduling application will allow the application of the forecasting model to the current Contact Centre, with a view to applicability to future Contact Centre configurations such as part-time staffing and an amalgamation with the client transaction processing area. The application allows Contact Centre managers to generate weekly schedules automatically based on the forecasts. These schedules can be easily adjusted to reflect unexpected demand or unusual staffing situations. The client services process analysis will be a sixteen-month COE project with a focus on improving the client satisfaction derived from the client services area. The client services area includes both the processing of mutual fund transactions and the Contact Centre. The project will include mapping of current client services processes, a benchmarking of current client service levels, and recommendations for improvements to processes or department structure, to improve the quality of service provided to Phillips, Hager and North mutual fund clients. One of the reasons PH&N supported the current client services project is the success of the previous project in increasing understanding of the activities within the Contact Centre. The advent of new technologies will have a profound impact on the client service department resulting in an environment of constant change. An ability to adapt quickly to change is required for the client services area (including the Contact Centre) to continue to provide excellent customer service. The application of forecasting, process modeling and simulation methods is expected to play an important role in the continuation and improvement of quality service to PH&N's mutual fund clientele. 32 IX. References Browning, Sharon Guy 1998 Tandem Queues with Blocking: A Comparison Between Dependent and Independent Service. Operations Research, 46.3, 424-429 Grossman, Thomas A. Jr. 1999 Teachers Forum: Spreadsheet Modeling and Simulation Improves Understanding of Queues. Interfaces 29:3, 88-103. Mason, Andrew J., Ryan, Panton David M..1998 Integrated Simulation, Heuristic and Optimisation Approaches to Staff Scheduling. Operations Research, 46.2, 161-175. Sanegre, Rafael 1998 Call Centre Staffing at the Workers' Compensation Board of BC. University of British Columbia. Seegers, Norm 1999 Dealer Call Survey, Phillips, Hager and North. Whitt, Ward 1999 Predicting Queueing Delays. Management Science 45.6, 870-888 Whitt, Ward 1999 Improving Service by Informing Customers About Anticipated Delays. Management Science, 45.2, 192-207 33 Appendix A : Contact Centre Typical Incoming Call Demand Distribution TUESDAY 2/2/99 Plot of Actual Daily Data Vs. Predicted 45.0 40.0 35.0 25.0 20.0 15.0 10.0 5.0 0.0 Actual Predicted Upper/Lower 90% Boundaries 7:30 8:00 8:30 9:00 9:30 10:00 10:30 11:00 11:30 12:00 12:30 1:00 1:30 2:00 2:30 3:00 3:30 4:00 4:30 5:00 AM AM AM AM AM AM AM AM AM PM PM PM PM PM PM PM PM PM PM PM 1/2-hour Period Ending TUESDAY 2/9/99 Plot of Actual Daily Data Vs. Predicted 45.0 , 40.0 35.0 30.0 » 25.0 J 20.0 15.0 10.0 5.0 0.0 Actual Predicted Upper/Lower 90% Boundaries 7:30 8:00 8:30 9:00 9:30 10:00 10:30 11:00 11:30 12:00 12:30 1:00 1:30 2:00 2:30 3:00 3:30 4:00 4:30 5:00 AM AM AM AM AM AM AM AM AM PM PM PM PM PM PM PM PM PM PM PM 1/2-hour Period Ending 34 Appendix B : 1998 TSF and Service Times Percentage of calls answered within 20 seconds for the Month of February 1998 100% 75% 50% 25% Call Volumes by Day for February 1998 (RRSP Season) 600 200 & s& .(* .c* .<* .c!> .c* ,riJ> Appendix C : Forecasting Model Parameters Parameter Coefficient (Constant) 1.5176371 7:30 AM 3.55508639 8:00 AM 4.5807055 8:30 AM 9.08447608 9:00 AM 9.86439599 9:30 AM 11.9842406 10:00 AM 11.1849796 10:30 AM 11.4433152 11:00 AM 11.4603624 11:30 AM 11.5201882 12:00 PM 11.3755729 12:30 PM 9.39722385 1:00 PM 9.44908547 1:30 PM 8.34066778 2:00 PM 7.81572302 2:30 PM 4.51624091 3:00 PM 4.2162905 3:30 PM 3.0997361 4:00 PM 3.23029848 4:30 PM 1.2180699 5:00 PM 0 MONDAY 2.93647461 TUESDAY 1.62985181 WEDNESDAY 0.66846191 THURSDAY 0.29849788 FRIDAY 0 TREND 0.18586921 M1 7.78702927 M2 9.34592325 M3 7.60459001 M4 5.47465075 M5 3.45938521 M6 1.19271492 M7 2.02372773 M8 1.59465515 M9 0 M10 1.4241821 M11 0.37936705 M12 0.75946989 The number of calls for any half hour period is calculated by (Constant + Half Hour Period + Day + Month + RRSP Trend )/2 Appendix D : Required Agents and Erlang C Formulation Number of Agent Hours Required by Half Hour Period by Month Month 1 2 3 4 5 6 7 8 9 10 11 12 7:30 4 5 4 3 3 3 3 3 3 3 3 3 8:00 4 5 4 3 3 3 3 3 3 3 3 3 8:30 5 5 4 4 4 4 4 4 4 4 4 4 9:00 5 5 4 4 4 4 4 4 4 4 4 4 9:30 5 6 5 4 4 4 4 4 4 4 4 4 10:00 5 6 5 4 4 4 4 4 4 4 4 4 10:30 5 6 5 4 4 4 4 4 4 4 4 4 11:00 5 6 5 4 4 4 4 4 4 4 4 4 11:30 5 6 5 4 4 4 4 4 4 4 4 4 12:00 5 6 5 4 4 4 4 4 4 4 4 4 12:30 5 5 4 4 4 4 4 4 4 4 4 4 13:00 5 5 4 4 4 4 4 4 4 4 4 4 13:30 5 5 4 4 4 3 3 3 3 3 3 3 14:00 5 5 4 4 4 3 3 3 3 3 3 3 14:30 4 5 4 3 3 3 3 3 3 3 3 3 15:00 4 5 4 3 3 3 3 3 3 3 3 3 15:30 4 5 4 3 3 3 3 3 3 3 3 3 16:00 4 5 4 3 3 3 3 3 3 3 3 3 16:30 4 5 3 3 3 2 2 2 2 2 2 2 17:00 4 5 3 3 3 2 2 2 2 2 2 2 # agent hours req'd 46 53 42 36 36 34 34 34 34 34 34 34 *daily variations are small and were excluded for the sake of simplicity in the model. February staffing represents average staffing levels for mid-February Erlang C The Erlang C distribution statistic is used for telecommunications applications to measure inbound waiting queues (immediate service). The Erlang C formula is based on the following assumptions: • Calls are served in order of arrival • Call arrival rates are exponentially distributed • Blocked calls are delayed • Holding times are exponential 37 Appendix E : Simulation ACD Queue Statistic Report Simulation ACD Queue Statistics Date : Friday, January 15, 1999 Queue Type : Two Queues Advisor Queue Period Ending Agents Calls Accepted Average Talk Time Average Occupancy Rate ASA Longest Call Delay TSF 5 sec # TSFs to Fund Advisors Fund Advisor Minutes 8:00 AM 3 17 79 12% 0 0 100% 13 83 9:00 AM 3 20 79 15% 0 0 100% 14 97 10:00 AM 3 33 87 26% 2 50 97% 26 245 11:00 AM 3 25 80 19% 0 0 100% 20 201 12:00 PM 3 33 80 24% 0 0 100% 27 260 1:00 PM 3 33 158 48% 16 192 79% 17 111 2:00 PM 3 32 128 38% 28 207 78% 22 132 3:00 PM 3 32 77 23% 0 0 100% 22 146 4:00 PM 3 27 59 15% 0 0 100% 20 130 5:00 PM 3 19 81 14% 0 0 100% 16 98 Total 271 93 23% 5 207 94% 197 1503 Dealer Queue Period Ending Agents Calls Accepted Average Talk Time Occupancy Rate (Avg) ASA Longest Delay TSF 20 sec 8:00 AM 3 10 149 14% 16 157 90% 9:00 AM 3 15 145 20% 61 426 80% 10:00 AM 3 22 208 42% 68 279 59% 11:00 AM 3 19 181 32% 91 294 37% 12:00 PM 3 26 269 65% 361 862 15% 1:00 PM 3 14 163 21% 69 506 79% 2:00 PM 3 6 43 2% 1 7 100% 3:00 PM 2 6 131 11% 18 110 83% 4:00 PM 2 6 143 12% 42 186 67% 5:00 PM 2 0 0 0% 0 0 0% Total 124 184 24% 121 862 57% 38 Appendix F: Telephone Call Centre Terminology A C D Automated call distributor. Transfers incoming calls to agents and collects statistics. A S A (Average Speed of Answer) shows the average time a caller had to wait before being answered. To calculate the ASA, the total time-before-answer is divided by the number of answered calls. Calls Abandoned - % the percentage of accepted calls that were abandoned. Calls Abandoned - Avg Wait divide the queue's wait-time-before-abandoning by the number of abandoned calls. Calls Answered the number of incoming ACD calls answered by the agents Duration of NonACD Calls the total time of incoming calls timed from call answer to final release. Hold Time (HDCP) Total time the agent had ACD calls on hold during the reporting period. Hold time begins when the agent presses the Hold key and ends when the call is resumed or when the caller hangs up. Not Ready (PCP) is calculated by dividing the total Pre-/Post-Call Processing time by the number of ACD calls answered by the agent. Talk Time (DCP) shows the agent's Talk Time divided by the number of answered incoming calls. If no calls were answered, this field is starred. Talk Time (DCP) sometimes called Direct Call Processing time, the time spent on incoming ACD calls, from initial answer to final release, minus any Hold time (HDCP). Target Answer Time (TAT) shows the target time within which an agent should answer a call. Target TSF % (Telephone Service Factor) gives the overall service goal for the queue. Answering 80% of calls (TSF %) within 20 seconds (Target Answer Time) is a common industry standard. Wait Time measures the time the agent is available and waiting for calls. 39 Appendix G : Fund Advisor Queue Statistics - ACD Output Data ACD Queue A c t i v i t y P e r i o d i c T o t a l s Date : 07/15/1999 Day o f W e e k . . . : Thursday Queue G r o u p . . . : PH&N V a n c o u v e r / A d v i s o r s P e r i o d Avg Agents R e q u i r e d C a l l s C a l l s ASA Longest C a l l s Abandoned TSF T o t a l E n d i n g A v a i l a b l e Agents A c c e p t e d Answrd Ans Delay No. % Avg Wait % T r a n s f e r s ACD-DN : 6205 Queue N a m e . . . . : A d v i s o r s 8 30 7 6 2 1 0 18 0 18 0 0 00* * * * * 100 0 9 00 8 4 3 4 0 08 0 10 0 0 00% 100 0 9 30 8 6 2 2 0 04 0 04 0 0 oos .... 100 0 10 00 8 5 2 2 0 29 0 48 0 0 008 .... 50 0 10 30 8 5 1 1 0 04 0 04 0 0 00% .... 100 0 11 00 8 6 2 2 0 31 0 54 0 0 00% .... 50 0 11 30 8 3 1 1 0 06 0 06 0 0 00% 100 0 12 00 9 3 1 1 0 12 0 12 0 0 00% .... 100 0 12 30 9 7 3 3 0 15 0 28 0 0 oos .... 66 0 14 00 10 2 1 1 0 06 0 06 0 0 001 .... 100 0 14 30 9 3 1 1 0 20 0 20 0 0 00% .... 100 0 16 00 11 7 2 2 0 13 0 20 0 0 00% .... 100 0 16 30 10 7 1 1 0 06 0 06 0 0 00% .... 100 0 17 00 8 6 2 2 0 13 0 20 0 0 00% .... 100 0 SUB-TOTAL ( for 62 05): 24 24 0:14 0 0.00% **** 87 Appendix H : Sensitivity to Transfer rate, Call length and TSF TSF 95 % CallsTsfd 70% 5 sec Length of Call 7 Agent Hours Max Calls January 64 7 February 77 8 March 57.5 6 April 54 6 May 50 6 June 45 5 July 46 5 August 45.5 5 September 43 5 October 45 5 November 44 5 December 45 5 % CallsTsfd 80% Length of Call 7 Agent Hours Max Calls 69 8 84.5 9 64 7 58 7 54 6 50 6 52 6 50 6 45.5 6 50 6 46 6 47.5 6 % CallsTsfd 90% Length of Call 7 Agent Hours Max Calls 74 8 92 10 67 8 63.5 7 58 7 53 6 54.5 6 54 6 49.5 6 53 6 51 6 52.5 6 TSF 95 % CallsTsfd 70% 5 sec Length of Call 6 Agent Hours Max Calls January 58.5 6 February 69 7 March 53.5 6 April 48.5 6 May 46 5 June 42 5 July 43.5 5 August 43 5 September 39 5 October 42 5 November 40.5 5 December 41 5 % CallsTsfd 80% Length of Call 6 Agent Hours Max Calls 63.5 7 76.5 8 57 6 54 6 49 6 45 5 46 5 45 5 43 5 45 5 43 5 45 5 % CallsTsfd 90% Length of Call 6 Agent Hours Max Calls 68 7 82 9 63 7 56.5 7 53.5 6 47.5 6 50 6 49 6 45 5 48.5 6 45.5 6 46 6 41 Appendix H : Sensitivity to Transfer rate, Call length and TSF (cont.) TSF 80 % CallsTsfd 70% 20 sec Length of Call 7 Agent Hours Max Calls January 48.5 5 February 59 6 March 44.5 5 April 40 5 May 36.5 4 June 34 4 July 36 4 August 35 4 September 31.5 4 October 34 4 November 32 4 December 34 4 % CallsTsfd 80% Length of Call 7 Agent Hours Max Calls 54 6 66.5 7 48 5 44.5 5 41 5 36 4 37 5 36 4 35 4 36 4 35.5 4 35.5 4 % CallsTsfd 90% Length of Call 7 Agent Hours Max Calls 58 6 72 8 53.5 6 47.5 6 43.5 5 39 5 42 5 40.5 5 36 5 40 5 36.5 5 38 5 % CallsTsfd Length of Call Agent Hours Max Calls Percent of calls transferred from the Contact Centre to the Advisor Queue Length of the call and post-processing time for the Advisor only Projected hours of phone time required per day in the month Estimated maximum number of calls to be handled by advisors at one time 42 

Cite

Citation Scheme:

        

Citations by CSL (citeproc-js)

Usage Statistics

Share

Embed

Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                        
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            src="{[{embed.src}]}"
                            data-item="{[{embed.item}]}"
                            data-collection="{[{embed.collection}]}"
                            data-metadata="{[{embed.showMetadata}]}"
                            data-width="{[{embed.width}]}"
                            async >
                            </script>
                            </div>
                        
                    
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:
https://iiif.library.ubc.ca/presentation/dsp.831.1-0089435/manifest

Comment

Related Items