Understanding Abandonment Rate
Goals and Metrics
B y M a g g i e K l e n k e
Along with service level and aver-age speed of answer, another com-monly reported statistic of call center performance is abandonment rate. While it can provide some useful informa-tion, it can be misunderstood.
By definition, abandonment rate measures the percentage of callers who select a destination but hang up before the destination is reached. It can be a measure of the percentage of callers who enter an automated attendant or IVR menu, but who hang up before the IVR delivers the call to an agent queue. But most often it is used to measure the percentage of callers who enter an agent queue but hang up before reaching an agent.
In a center where callers are poten-tial buyers of a product or service, loss of the opportunity to talk to that caller could result in a loss of revenue. Even in a situation where direct revenue is not at stake, losing customers due to poor service will have a direct impact on the company’s bottom line. This makes managing the level of abandons a focus in many centers.
The key to managing abandon rate is to determine why callers abandon in the first place and the wait in queue is just one of the reasons. Other possibilities include:
I N S I D E
The Role of Workforce
Management in Business
B y M a r s h a l l L e e , C a r e w i s e H e a l t h , & J u s t i n Z i e g l e r, A m e r i c a n E x p r e s s G l o b a l B u s i n e s s T r a v e l
You are unique. Who else in your organization touches business pro-cess, finance, the contact center, leadership, agents, the executive team, support services, IT, and so much more? Who else sees the associates daily intake and output of time and labor? Who else is in a position to see the impact of indi-vidual, group, and process performance real-time and in simulation? Those in workforce management or workforce planning are most certainly unique. Have you ever wondered how you and your team can help inform and make the big decisions?
Business Performance Management is essential to the function of any
organization. By design, performance management guides an organization toward a goal. The larger progress is measured by metrics which are Key Performance Indicators or KPIs. Ideally your KPIs should be a literal guide to indicate the core successes or failures of performance. For example, a company with a goal of being a competitive claims processer has supporting goals related to quantity, cost, and quality. These subsets of targeted goals are addressed by KPIs. They should define if the organization and individuals are helping to drive success in those areas. You are unique in part because in that organization you could help target each point.
WFM Survey Results . . . 2 WFM Spring Survey . . . 3 Exploring Call Center Turnover . . 4 Numbers
Events Calendar . . . 5 A Pocket Guide to Call Center . . . 6 Terms and Acronyms
It’s All About the Algorithms! . . . . 8 Ask the Workforce Wizard . . . 9 Our Sponsors . . . 9 Comparing Erlang C and . . . 10 Simulation Modeling
Congratulations to our most . . . . 15 recent Certified Workforce
Join SWPP . . . 16
Continued on page 15 Continued on page 12
W F M S u r v e y R e s u l t s
This article details the results of the most recent SWPP quarterly survey on critical workforce planning topics. In this survey, which focused on workforce management for back office functions, approximately 200 call center profes-sionals representing a wide variety of industries participated and provided insight into numerous workforce planning questions.
The largest percentage (45%) of the participants was from large call center operations with over 500 seats, followed by 17% with 201-300 agents. Twelve percent of the survey partici-pants represent centers with 101-200 agents. All types of call center operations were represented in the study, with the biggest percentage representing financial, insurance, and telecommuni-cations industries.
WFM in the Back Office
The survey participants were asked if the work-force management team has been asked to forecast or schedule for non-phone back office tasks in their
organization. Sixty-eight percent of the participants indicated that they have been asked to perform this function. Based on the responses to questions that follow this, it appears that some of these operations may have integrated back office tasks within the call center as agents perform both types of work. In other cases, the back office functions are separate from the call center activities.
Forecasting for the Back Office
The survey respondents were asked if they are currently forecasting volume for the back office tasks and 65% responded that they do, while 34% do not. Forecasting for this type of work can be
challenging as the types of historical data that call centers have relied on may not be easily accessible for non-call activities. It may also rely heavily on agent behaviors (such as remembering to mark on a tick sheet) so may not be accurate for either total or time of arrivals. Handoffs between departments and multi-step processing that can span several hours or even days further complicate the forecasting of this type of work.
Scheduling for the Back Office
When asked if the WFM department is currently scheduling for the back office tasks, over three-quarters of the respondents indicated that they do. This is a higher percentage than those who forecast. Given the sequential nature
of the back office work, the scheduling process is different in that timing of such things as arrivals/departures, breaks/lunches, and other activities may be less critical than when the speed of answer goal for calls demands seconds of response time. Here adherence to schedule details may not be as important as ensuring compliance with the overall total scheduled time in each day, as the work can be done on a more flexible schedule. However, ensuring that there is enough staff to complete the workload within the response time goal is still the objective.
Work Types Tracked
When asked what type of work is tracked for WFM tasks, the responses are fairly dispersed with 27% in Account Manage-ment tasks, 17% each in Billing and Complaints, and 16% in Claims. However, 23% indicated that none of these types matched the types of work they are tracking. As the concept of WFM in the back office continues to evolve and grow, the variety of contacts that the WFM team will be asked to handle is likely to continue to expand.
How Work Types are Tracked
When asked to describe the method that is used to track the back office works types, the responses are quite varied. While a third use
some type of operational system such as claims, complaints, or CRM systems, another 30% use some type of proprietary home-grown tracking system. Only 16% indicated that they are using a vendor-supplied solution and 6% report using agent tick sheets.
SWPP conducts a survey each quarter on critical workforce planning topics. These results will be published in upcoming issues of On Target, as well as published on the SWPP website in the members-only Library section. You may fax this page to 615-352-4204 or fill in the survey online at www.swpp.org.
The focus of this survey is agent scheduling practices.
Responses to the first two questions will allow us to segment the answers by size and type of call center to contrast and compare workforce management practices. Survey results will be completely anonymous.
1. How many agents are in your call center?
Under 50 301 – 400 51 – 100 401 – 500 101 – 200 Over 500 201 – 300
2. What industry do you represent?
Telecommunications Outsourcer Travel Health Care Financial Government Insurance Retail/Catalog Utility
3. Do you survey your agents to see what shifts they would prefer to work?
4. What process do you use for assigning schedules to agents?
Assign automatically through software program Assign through shift bid process
Assign at hiring and don’t change
5. How often are schedules assigned?
Weekly Monthly Quarterly Semi-Annually Annually As needed Once at hiring and never again Never Other _________________________________________
6. Which of the following agent self-scheduling options do you currently offer? (Check all that apply)
Shift bidding (initial schedule) Absence Notification
Request Schedule Swap Request Extra Hours Request Time-Off
Request Schedule Change None of the above
7. Which of these agent self-scheduling options are fully automated (no manual approval or administration required)? (Check all that apply)
Shift bidding (Initial Schedule) Absence Notification
Request Schedule Swap Request Extra Hours Request Time-Off Request Schedule Change None of the above
8. How valuable are the following agent self-scheduling functions to your agents?
Low Medium High Don’t Know
Absence (“Sick-Line”) Notification Request Schedule Swap (With Another
Request Extra Hours Request Time-Off
Request Schedule Change (Trade Scheduled Hours for Unscheduled Hours)
9. What policy restrictions do you have in place for agent schedule change requests? (Check all that apply)
Certain number of requests allowed Leadership approval before submission Must be employed for a certain length of time
before requesting change No Restrictions
10. What is the impact of these restrictions on agent satisfaction?
W F M S u r v e y
Respond and Win!
Not only will you receive a report of our findings, but you’ll have a chance to win a free SWPP Membership for responding to the survey. Please return to SWPP by September 15, 2014. Congratulations to Robert Brown of Citi, who won a free SWPP Membership last quarter for completing the SWPP survey.
Name ___________________________________________________________________________________________ Company __________________________________________________________________________________________ Email Address ______________________________________________________________________________________
Exploring Call Center Turnover Numbers
One of the biggest staffing problems that call centers face today is staff turnover. Finding and then retaining qualified staff has a huge impact on the bottom line as well as to quality of service.
You should be taking a close look at some of the numbers associated with turnover if you hope to manage it better. The two most important numbers are the rate of turnover and the cost of turnover. This article will take a look at ways to calculate turnover rate in several different ways.
Industry Turnover Statistics
Turnover is at an all-time high in today’s workplace, and some of the overall turnover statistics have significant implica-tions in the call center. According to the United States Bureau of Labor Statistics, workers aged 20 – 24 stay with an organiza-tion only 1.1 years on average (compared to 1.5 years just 15 years ago), and workers aged 25 – 34 stay 2.7 years (compared to over 3 years in the 1980s). Compared to these longevity numbers, call center workers who generally fall into this age group (ages 20 – 34) stay only about one year.
Research studies by callcenterjobs.com indicate that the rate of turnover varies by area of the country, employment factors in a specific region or city, and by industry. There is a much higher turnover rate in routine, order-taking positions or in outbound telemarketing where burnout is high. Turnover is lower in more specialized, higher level jobs and also lower in union environments.
Overall averages for the call center industry as a whole range between 30 – 45 percent, with some centers having almost no turnover, and other centers having turnover in the triple digits.
Like almost any benchmarking numbers, it’s frankly not all that important what the turnover rates are for other call centers or for the industry in general. What’s important is what the turnover rate is in your center and what you can do about it.
Calculating Your Call Center Turnover Rate
You should calculate the turnover rate for your center on a regular basis and in many different forms. You’ll want this number for trending purposes and business case justification for programs to assist with retention.
In calculating this turnover rate, most organizations use an annualized rate to describe the proportion of staff that leaves. The formula for turnover is the number of staff that leave divided by the average number of staff. For the numbers in Table A, the turnover rate would be calculated by dividing 54 (the number of departing staff) divided by the average number of staff positions (81.5), resulting in an annual turnover rate of 66 percent.
This simple calculation gives you a good starting point, but isn’t all that useful by itself.
You’ll want to take a look at your turnover rate in many different ways. Here are some of the variations that will be interesting and perhaps more actionable than just the overall turnover rate.
Internal versus External Turnover
Internal turnover refers to employees leaving for other positions within the company, while external turnover shows employees leaving the organization completely. Both are costly to the call center organization, but obviously some benefit exists to the organization if qualified people are leaving to fill other roles within the company.
The key is to track the exit rate from the center into other positions and build this into your staffing and training plan. One large financial organization has a high percentage of loss to other departments with the call center seen as an easy entry point to the company from which one can springboard to other positions. Rather than bemoaning this fact, they’ve tracked and documented the exodus and gotten extra budget dollars for staffing and training programs for preparing staff for taking other positions in the company.
Voluntary versus Involuntary Turnover
Turnover should also be analyzed to understand the propor-tion of agents who make the decision to leave the company compared to those that are being terminated. If your company has to terminate significant numbers of employees due to poor performance, there may be a problem at the recruiting and hiring phase. The company may be hiring too quickly in order just to fill seats and may not be screening carefully enough for necessary knowledge and skills or evaluating the personality match with the work environment.
Exit interviews should examine those reasons that are unavoidable, such as someone returning to school full-time or relocating for a spouse’s job. These exit interviews can shed light on what reasons are unavoidable and what reasons could have been addressed by the call center.
Turnover Rate by Team
Turnover rates should be calculated for the center, but should also be calculated for smaller defined groups. For example, turnover should be examined by team or supervisor. Calcu-lating turnover rate by team may help pinpoint problems where employees are leaving due to supervisory issues and not because of compensation, job fit, or other factors. Those supervisors or team
B e g i n n e r ’ s G u i d e
Continued on page 5
Month Departing Staff Avg Number of Staff
1 4 75 2 2 78 3 3 80 4 6 80 5 5 82 6 5 84 7 4 85 8 5 85 9 6 85 10 4 82 11 5 82 12 5 80 Total 54 81.5 Table A
Exploring Call Center Turnover Numbers
managers with consistently high retention rates can perhaps serve as “retention models” for other supervisors who can learn from their motivation and management techniques.
Turnover Rate by Call Type
You should also review the turnover statistics to see if there is any correlation with the skill or type of calls being handled. Calculation by call type in one organization revealed nearly double the turnover rate for one particular type of call versus any other one due to its difficulty and higher stress levels. When that center was able to allocate more resources to that call handling group and alter performance expectations slightly to make it a less stressful environment, retention improved significantly.
Turnover Rate by Performance Level
It should be noted that while turnover almost always has a negative effect on the call center, there are some positive aspects. If you are losing personnel that were not performing well, the opportunity to hire replacements can be a positive experience. Bringing in new personnel provides an opportunity to gain new ideas and to obtain some fresh perspectives.
One of the ways to evaluate turnover rate by performance level is to divide your staff into performance groups. The first table illustrates a situation where turnover is troublesome. The top quarter of performers has a turnover rate of over 30%, while only 15% of the poor performers are going out the door.
Turnover Analysis Turnover Analysis 0% 5% 10% 15% 20% 25% 30% 35%
Hire Performers Performers Performers New Low Medium Top
Tur nov e r R a te
On the other hand, the second table represents a turn-over situation that might not be viewed as negatively. In this example, the center is doing a good job of keeping good people, with only 5% of the top performers leaving. Turnover rate is high for the poor performers at 35%, but if you’re going to lose staff, it’s better to have the non-performers leaving.
Turnover AnalysisTurnover Analysis
0% 5% 10% 15% 20% 25% 30% 35% 40%
Hire Performers Performers Performers
New Low Medium Top
Tur nov e r ra te
You should be taking a close look at your turnover numbers if you hope to manage it better.
Continued from page 4
B e g i n n e r ’ s G u i d e
Title Date Location
SWPP/QATC Regional Meeting Aug 21 Staples Call Center, Aurora, CO
Utility Call Center Operational Excellence Sept 8-10 The Florida Hotel & Conference Center, Orlando, FL inContact User Conference (ICUC) Sept 23-25 Renaissance Orlando at SeaWorld, Orlando, FL QATC Annual Conference Sept 24-26 Omni Nashville Hotel, Nashville, TN
SWPP/QATC Regional Meeting Oct 1 IHG Call Center, N. Charleston, SC Call Center Demo & Conference Nov 3-5 Hyatt Regency, Chicago, IL SWPP/QATC Regional Meeting Nov 4 Hyatt Regency, Chicago, IL WFM Virtual Conference Nov 4-7 www.ecrmevents.com Home Agents & Virtual Call Center Summit Nov 18-19 Denver, CO
E v e n t s C a l e n d a r
Thanks to The Call Center School for the ongoing Beginners Guide series. For more information about The Call Center School, please visit their website at www.thecallcenterschool.com.
A Pocket Guide to Call Center Terms and Acronyms
Continued on page 7
There are so many acronyms and calculations in WFM and in the call center in general. We thought it would be nice to share these definitions about some of the most com-monly used terms that you an use as a reference guide. • Adherence Percentage – The percentage of adherence is
typically a measure of the percentage of time that an agent is doing exactly what the schedule dictates. This includes logging in and out on time at the beginning of the day, at break and lunch times and any other times when required. It may also measure time the agent is working in the scheduled work type or skill. (Note: Not all vendors or centers define this the same way.)
• Example = Schedule is 8 AM to 4:30 PM with 30-minute lunch at 12:30 PM and 15-minute breaks at 10 AM and 3 PM. The agent logs in on time but goes to break at 10:10 AM, takes 45 minutes at lunch and takes the afternoon break on time. The agent is 10 minutes out of adherence in the morning and 15 minutes at lunch for a total of 25 minutes out of the 8 hours scheduled = 25 minutes / 480 minutes = 5% out of adherence or achieving a 95% adherence.
• Average Handle Time (AHT) – AHT is the total amount of time an agent is actively involved in the completion of the work for a single call. It starts when the agent answers the call and includes the talk time, any time the caller is placed on hold during the call, and the after call work or wrap up time. Be aware that some ACDs include hold time in the talk time reported while others report it separately. It is an average calcu-lated for a single agent, a team, a call type, or a period of time.
• AHT = (talk time + hold time + after call work) / # of calls
• Example = 14 minutes of talk time + 8 minutes of hold time + 6 minutes of after call work by an agent = 28 total minutes of work/ 8 calls = 3.5 minutes AHT
• Average Speed of Answer (ASA) – ASA is the average wait in queue that callers experience before their call is answered. It is generally measured for a single call type over a period of time such as half-hour, day, shift, week, or longer. The average is calculated using the wait of all the callers in the queue (both those who were answered immediately and those who were not) divided by all the calls offered. There are variations of ASA that discard the wait of calls that abandon or only those that abandon before a set period of time. Some calcu-lations may also ignore some set number of seconds at the beginning of the call (perhaps to leave out the time listening to the answering announcement).
• ASA = Total wait in queue of all callers / total number of offered calls
• Example = 93 calls in an hour wait a total of 27.3 minutes = 27.3 / 93 = .294 minutes or 17.6 seconds • Average Trunk Hold Time (ATHT) – ATHT measures the
average amount of time that telephone trunks are utilized over a period of time. It is used to determine the number of trunks
that are needed to avoid busy signals to incoming callers. Trunk hold time begins when the incoming call starts to ring and includes the recorded announcement time, queue time, IVR interaction time, talk time, and time the caller is placed on hold during conversation. It does not include the after call work time of the agent as the caller has already been discon-nected during that period. If a caller abandons or is totally satisfied with the IVR process and disconnects without talking to an agent, the time on the trunk is still counted.
• ATHT = total ringing + announcement + IVR + queue + talk + hold time / total number of calls
• Example = 75 calls were received during an hour. The total time in ringing is 4 minutes, announcement is 32 minutes, and IVR time is 173 minutes. Only 43 calls required agent assistance and those added 190 minutes of talk time. 4 + 32 + 173 + 190 / 75 = 399 / 75 = 5.32 minutes ATHT
• Compliance – The percentage of time that an agent works compared to the total time that was scheduled. This does not measure if the agent adheres to the exact times scheduled for log in/out, breaks, etc. It focuses on whether the agent put in the total amount of time that was scheduled. This is some-times referred to as conformance. Not all vendors or centers define this in the same way.
• Example = The agent is schedule 8 AM to 4:30 PM with a 30-minutes lunch and 2 15-minute breaks or a total of 7.5 hours of logged in time. The agent arrives late and logs in at 8:15 AM but stays until 4:45 PM in the afternoon. This would be counted as 100% compliance because 7.5 hours was worked even though it was not the exact times scheduled.
• Erlang – An erlang is a measure of traffic density or utilization of a telecommunications trunk, an agent or other server. It is used to calculate the number of servers required to meet a specific service goal. It is named after a Danish mathemati-cian, Dr. Agner Erlang, who created a group of formulas for the Copenhagen telephone company around 1920.
• Erlang = 15 hours of workload during a single hour of time = 15 erlangs
• Example = 75 calls at 300 seconds of average handle time in one hour = 75 X 300 = 22,500 seconds / 3600 seconds in an hour = 6.25 hours or erlangs of workload
• Full-Time Equivalent (FTE) – FTE is a calculation that adjusts part-time workers into portions of full-time workers. This is often used in capacity planning to identify total work hours needed rather than total head count when part of the staff is not full-time.
• FTE = total work hours of a team of staff / the number of hours that constitute full-time such as 40 hours per week • Example = 120 staff work 40 hours per week and 32 staff work an average of 24 hours each week = 4800 hours for
full-time staff + 768 hours for part-time staff = 5568 total hours worked / 40 hours per week = 139.2 FTE (even though there are 152 total head count)
• Longest Delay in Queue (LDQ) – LDQ measures a period and isolates the single caller that waited the longest. There are two versions of LDQ. One identifies the caller who waited the longest to be answered by an agent and the other identifies the caller who waited the longest before abandoning the call.
• Example = During a half-hour, the average wait in queue for callers was 35 seconds, but one caller waited 4.5 minutes to be answered = LDQ is 4.5 minutes • Occupancy – Occupancy is a measure of how busy agents
are during the time that they are logged in and available to handle incoming calls. It includes the talk time, hold time, and after call work (workload) but does not include the time that the agent is idle waiting for another call. Workload is expressed as man-hours of work to be done in an hour of time or Erlangs. In some centers occupancy may also include some of the auxiliary work that an agent may do that directly serves the customers such as outbound calls, emails, chats, etc. It does not include time the agent spends in non-working states such as not ready, on break, etc.
• Occupancy = Workload / total time logged in
• Example = 37 hours of workload / 40 hours of time staff are logged in and available (or Average Positions Staffed) = 92.5% occupancy
• Response Time – Response time is generally used to measure the delay experience of non-call work such as emails, faxes, and white mail. The measurement begins when the item is received by the center and continues until the response has been completed by the agent and sent back to the customer. This makes it somewhat different from inbound calls in that call service is measured only up to the point that the call is answered by the agent, but does not include the time during the call handling.
• Example = An email is sent to the call center by a customer and is received in the email box at the company at 10 AM. The agent opens the email at 2:30 PM the same day, composes an answer that takes 5 minutes and hits the send key at 2:35 PM. The total response time is 10 AM to 2:35 PM or 4 hours and 35 minutes.
• Service Level or Telephone Service Factor (TSF) – Service
level or TSF is the percentage of calls that were answered within a defined number of seconds/minutes in a period. The goal includes both a percentage and a time period, such as 85% of calls to be answered in 30 seconds or less. The total number of calls that are answered in the 30 seconds or less is divided by the total number of offered calls. It does not matter to the formula how long calls wait when it is longer than the goal – the call misses the goal if it waits 1 second longer or 5 minutes longer. There are variations of service
level that discard the calls that abandon or abandon before a set period of time such as the goal. Some calculations may also ignore some set number of seconds at the beginning of the call (perhaps to leave out the time listening to the answering announcement).
• Service Level = Total number of calls answered within the goal number of seconds / total number of offered calls • Example = 83 calls answered within 20 seconds (goal)
out of 110 calls offered = 75% service level
• Shrinkage – Shrinkage is the time agents are paid but are not available to handle customer work. This is primarily used on incoming phone call work but can be applied in other types of work as well. Shrinkage includes such things as breaks, training, coaching, illness, tardy, vacation, non-call work such as research and call backs, and other activities that make the agent unavailable to take calls. It is expressed as a percentage of loss and is used to calculate the total staffing requirement when the workload and service goal are known.
• Shrinkage = Total time unavailable for incoming calls per person / total paid hours per person
• Example = Assume a person works 40 hours per week or 2080 hours per year. Breaks are 30 minutes per day or 6.25% of an 8-hour day. The agent takes 17 days of paid time off (including sick days, holidays, and vacation) per year or 136 hours or 6.5% of the year. Coaching and training = 3% and team meetings are another hour per week or 2.5%. Missing time (or failure to adhere to the planned schedule) is 15 minutes per day or 3.125%. The total shrinkage = 21.375 %
• Workload – Workload is the total amount of work that is measured over a defined period of time. In the case of incoming calls, it is the total talk time plus hold time during conversation plus after call work time. For outbound calls it is the time preparing for the specific call, dialing, listening to busy, ring no answer or ringing that is answered, conversation time and after call work.
• Workload = (Average talk time + hold time + after call work) X total number of incoming calls in a period (such as a half-hour)
• Example = Average talk time is 85 seconds, hold time is 10 seconds and after call work is 22 seconds or a total of 117 seconds per call. 230 calls in a half-hour multiplied times 117 seconds each = 26,910 seconds of work / 1800 seconds in a half-hour = 14.95 hours (or erlangs) of workload.
Note: While these are standard definitions in the industry, there are
variations that are widely accepted or unique to a particular vendor or call center operation. It is more important to ensure that your definitions and calculations are clearly defined, and that all systems and processes are using the same formulae, than that they match specific definitions from outside your organization.
Continued from page 6
It’s All About the Algorithms!B y B o b W e b b , P i p k i n s
Workforce management may be the most under-valued of all contact center solutions. Often, it is bundled and sold as part of an end-to-end solution by vendors who specialize in other software and have no in-depth knowledge of workforce management. Some solution providers minimize the value of workforce management in favor of their area of expertise when in fact workforce management is the cor-nerstone of any contact center operation. Accurate forecasting/ scheduling is imperative for achieving the balance to have your agents consistently in the right place at the right time. There are many vendors who offer workforce management; however, their products are not all equal. The disparity between solutions lies in the algorithms.
Many workforce management systems lack critical functions and flexibility to meet the needs of contact centers; or, their platform is unable to maintain sufficient historical call data to generate accurate forecasts. The most common problems found with workforce management systems are inaccurate forecasting and an inability to generate requirements at the interval level. Both of these problems, coupled with inadequate scheduling algorithms, can prove costly and negatively impact a company’s bottom line.
Workforce management software should use mathematical algorithms for accurate call volume forecasting and scheduling based on data exclusive to each center’s target service levels, fluctuating call volumes, agent skill sets, and “what if” scenario requirements. Accurate forecasting takes into account all the historic dynamics. Systems that use a “simple” weighted moving average can only use thirteen weeks of historical data, which is not enough to provide a statistically valid forecast. Correlated forecasting, which is forecasting for specific events that cause wide fluctuations in the volume of calls that must be processed, can only be performed by the most sophisticated systems which use all historical data.
Algorithms should reflect real-life customer behavior and include curve mapping and pattern recognition. In environ-ments where workloads regularly ebb and flow due to marketing activities and other definable variables, Historical Trend Analysis is the only way to ensure proper staffing because it is the only methodology that can incorporate complex historical trends in its calculations. Without pattern matching to predict different customer behavior for different events, the risk of over- or under-staffing increases dramatically. Historical Trend Analysis not only accurately predicts the continuation of trends, but the more advanced algorithms also incorporate pattern recognition to fine-tune forecasts for special events like promotional mailings or national holidays. Each time a particular event reoccurs, the fore-casted call volume is automatically adjusted to reflect the increase or decline in incoming work caused by comparable occurrences in the past, such as a historical 40 percent drop in volume on the Fourth of July.
An important component of accurate forecasting is having an integrated approach to support multi-skilled issues. It is necessary to have forecasting algorithms that directly calculate requirements in a multi-skilled environment, while avoiding repetitive analytical simulations. A single forecasted set of requirements should be generated for all inter-woven skilled activities, regardless of the type of work being offered, such as email and chat. Recognizing secondary skills and accounting for call overflow to available secondarily skilled agents will help eliminate overstaffing. Forecasts that are based solely on primary skills will generally overstaff, since overflow cannot be considered as a factor.
Overstaffing occurs when abandoned calls are not taken into consideration. Staffing operational costs, which account for 70-80% of your budget, can be severely impacted by overstaffing. For absolute maximum efficiency, your software should have an algorithm that incorporates abandoned calls. Systems that don’t understand abandons will always overstaff your call center. This is like the airplane that takes off with empty seats; you will never have another chance to recover that revenue.
Once scheduling requirements are known, use an algorithm that maximizes the achievable quality of service. Avoid using a simple “hours-net-to-zero” scheduling algorithm. Scheduling systems that use a simple net-to-zero algorithm cannot distinguish between schedules that deliver good and bad service. If under- or overstaffing during different intervals throughout the day nets to zero, you are not truly meeting your service level objectives during those intervals. Wasted labor expenses can occur through over-staffing, while under- staffing results in lost revenue.
Workforce management is based on science and should be approached from a scientific perspective. The best option for purchasing considerations is to choose a vendor who specializes in workforce management and understands its complexities. Starting with a solid workforce management foundation will help ensure you are more likely to reach service level objectives with fewer problems.
Pipkins, Inc., founded in 1983, is a leading supplier of workforce management software and services to the call center industry, providing sophisticated forecasting and scheduling technology for both the front and back office. Its award-winning Vantage Point is the most accurate forecasting and scheduling tool on the market. Pipkins’ systems forecast and schedule more than 300,000 agents in over 500 locations across all industries worldwide. For more information, visit www.Pipkins.com.
Interested in becoming a sponsor? Call Vicki Herrell at 877-289-0004.
It has been our practice here to measure agent conformance to schedule. We recently got a new manager for customer service and she is urging us to move to measuring agent adherence to schedule. Our feeling is that one measure is as good as the other, so why go through the change management exercise? It would be a lot of work to do what she wants. Is it worth it?
I thought we had resolved this issue some time ago but I still encounter contact centers that do, indeed, measure conformance to schedule instead of adherence to schedule. While the two terms sound very much alike, they have very different conse-quences for the contact center.
First, let’s understand what each measure means. If the published schedule has me working from 8:00am to 5:00pm with two breaks and a 30- minute lunch, I should work 8 hours. So, to be in conformance I need to work 8 hours, but I don’t necessarily have to report at 8:00am or leave at 5:00pm. In fact, I could conceivably work from 5:00pm to 2:00am and be 100% in compliance with the schedule.
That’s not the case with schedule adherence. Adherence to schedule means that you started on time, took breaks and lunch when scheduled, and ended your work-shift when indicated. Oh, by the way, you are also in complete compli-ance with your schedule. In a sense, by measuring adherence to schedule you get compliance to schedule for free. But it doesn’t work the other way around.
Schedule compliance is an operations metric that origi-nated in manufacturing. It is a useful measure there because the work to be done need not be done within rigid time boundaries. If I started a manufacturing task several hours later than booked into the weekly schedule, I could always spend a few extra hours on that task “to catch up.”
I’ve noticed some tendency for outsourcers to use schedule conformance. I think it’s tied to the need to be very flexible among multiple different clients and campaigns they are servicing. So, the management team deals with everything constantly in real time; reskilling agents on the fly attempting to leverage paid agent labor hours against shifting demand.
As we are all too aware, you can’t play “catch up” in the contact center. Immediacy changes everything.
Note: This Ask the Workforce Wizard answer is provided
by Bill Durr of Verint Systems. Bill may be reached at firstname.lastname@example.org.
Have a tough question?
Comparing Erlang C and Simulation ModelingB y B a y u W i c a k s o n o a n d R i c K o s i b a , I n t e r a c t i v e I n t e l l i g e n c e
One of the first things that we learn in workforce man-agement is the Erlang C calculation. It is easy to use, is fast in calculation, and is the standard in the call center community for determining staffing requirements. Most work-force management systems use some variant of it and planning spreadsheets use either Erlang C or a simple workload calcula-tion.
It’s available as an Excel add-on for free and it’s what everyone else is using, so it has to be great, yes? Inherent in Erlang C are several assumptions (as is the case of any call center model); assumptions about arrival and handle time distributions and more importantly, assumptions about customer patience. The Erlang C model assumes that your customers do not abandon.
This discussion will focus on comparing our old standard Erlang C with the newcomer, simulation modeling.
We have always been a proponent of simulation modeling to determine headcount required and service expected in contact center operations. Simulation modeling is being used more and more for complex multi-channel or multi-skill operations, because Erlang cannot model operations other than simple single skill inbound phone centers.
One problem with simulation models are that if built right, they are typically not generic, meaning every contact type’s simulation model will be different from the next in its assump-tions. For instance, two call types can be modeled with different call arrival distributions or patience distributions, and hence, even if they were tested on the same volume, handle time, and service goal, they would produce a different staffing requirement. This is a good thing — models should be “tuned” to the contact type they model.
Validation is Key to Accuracy
Because simulation models are best when they are custom built for the operation they are modeling, these models should always be built around a validation process. Calling a model “validated” is simply shorthand for saying that the model has been proven to be accurate, using real-world operational performance data. If you are using simulation in your workforce management system or in your planning models, the model should have a validation graph regularly provided — actual versus model results — to prove to yourself that the model comes close to reality. But the same is true of an Erlang (or any other) calculation.
Let’s go through the exercise of validating both the Erlang C calculation and a simulation model. We performed a simple test: 1. We gathered ACD data from a real-world simple (single
skill) call center.
2. For every interval (hourly in this case) we noted the number of calls received, the number of agent hours available, and the handle times.
3. We fed those parameters into an Erlang C calculator and a custom simulation model.
4. We compared the service achieved versus the service predicted from the two mathematical models. 5. We graphed each hour and graphed a weekly roll up. Below is a graph of a service prediction validation for a banking call center.
This is a great example of a contact type where the Erlang equation fails. Erlang C simply predicts that call center service levels will be much worse than the service achieved. This is a common problem with Erlang C. Because it assumes that there will be no abandons, the equation requires that each and every call — even those with very short customer patience — will add to the service level result.
The simulation model is terrific. It predicts service accurately. Another important test that can be done is to validate that the models predict staff requirements accurately. In this exercise we validate each model to determine the expected FTE cost of the model’s inaccuracy. Here we perform a different simple test:
1. We gather ACD data from a real-world simple (single skill) call center.
2. For every interval (hourly in this case) we note the number of calls, the number of agent hours, and the service achieved.
3. We feed those parameters into an Erlang C calculator and a custom simulation model. We assume that the service achieved is the model’s goal.
4. Compare the staff available versus the staffing require-ment predicted from the two models.
5. Graph each hour and graph a weekly roll up.
The next graph represents staff required for both simulation models and the Erlang C equation for the same call center.
This graph illustrates the cost of a model error. Assuming you were using an Erlang C calculation in your workforce management system or in your planning system, the Erlang model would have overstaffed the operation by over 20%!
The models underlying your workforce management and capacity planning process are very important to the operation. The capacity/strategic planning process is the tool we have to determine our center staffing levels and any model error will have significant cost repercussions to our company.
Note also that the discrepancy between the actual staffing requirements and that predicted by Erlang is inconsistent. Sometimes the error is 8% and sometimes it is 23%. We cannot simply add a fudge factor to Erlang C to make it more accurate.
A Summary of Model Results
We did a small validation survey using ACD data from six different businesses, each representing a different line of business (or contact type). Our goal was to demonstrate error rates of the Erlang C equation and simulation modeling.
We used the same process as above, where we gathered ACD data from each of the businesses, we fed actual interval data into each of the mathematical models, and determined whether the models would predict the service actually achieved in the call center.
Each hourly result from both models was compared to the actual service delivered in the contact center and the error was averaged.
A summary of hourly performance predictions is presented in the table below.
Call Type Avg. Error Sim Avg. Error Erlang C Avg. Abn Rate (%) Avg. SL (%)
Loans 0.01% 27.34% 7.93% 76.45% Member Services -1.02% 30.91% 5.53% 84.70% Preferred Services 2.69% 21.14% 2.93% 73.55% Retail -0.09% -0.93% 1.05% 98.61% Credit Card -4.31% 5.92% 7.23% 55.87% Auto Insurance -1.90% 0.31% 1.32% 87.19% SUMMARY -0.77% 14.12%
There are a few things to note. A positive average error implies that the service expected is lower than the service achieved (actual service minus predicted service). Error in this
direction will imply overstaffing, because for capacity planning, one would need to add more staff in order to achieve the same level of service performance.
Second, the size of the centers tested vary from small (average of 3 FTE per hour) to big (average of 228 FTE per hour). The number of intervals tested ranged from 185 intervals (~3 weeks’ worth) to 4500 intervals (~29 weeks’ worth). While this is a somewhat small-to-medium range sample, its result is in keeping with our experience in modeling contact centers over the last 14 (!) years.
The obvious conclusion, looking at the table is that an Erlang C calculation is accurate only under limiting conditions. Erlang C is accurate only when abandons are very low (~1%). Simulation is not limited by this, as each model uses history to derive a unique customer patience curve.
There is an old saying (often used when discussing golf and your buddy’s terrible shot), “It’s not the tool, it’s the craftsman.” Simulation modeling is such a tool — there is no such thing as a generic call center simulation model that validates unless it has a very good craftsman (a math modeler or operations research professional). Remember, each call center simulation model should be different from the next.
As such, it is very possible to have a simulation model that is poor. But, if the model is developed through a validation exercise, it can be developed accurately. It is very important that every one of your workforce management or planning models is measured for accuracy.
What can we conclude about Erlang C? In the simplest case — when Erlang has the possibility of performing well (single skill, simple center) — Erlang C is still inaccurate most of the time. It is most inaccurate when the call center targets abandon rates greater than 1% or evaluates scenarios with abandons greater than 1%. This, then, implies that the model cannot be used to do any real or interesting what-if analyses.
More significantly, when used in workforce management or the capacity planning process, the Erlang C model will lead to significant waste. Centers will be inconsistently over staffed and center resources wasted.
Note: We are working on a white paper, expanding this analysis that
outlines the differences between Erlang C and simulation modeling, which will be available shortly at www.inin.com.
Ric Kosiba, Ph.D., is a charter member of SWPP and vice president of Interactive Intelligence’s Interaction Decisions Group. He can be reached at Ric.Kosiba@InIn.com or (410) 224-9883.
Bayu Wicaksono leads Interactive Intelligence’s Operations Research department (mathematical modeling). He loves to talk about math, programming, and process improvement and can be reached at Bayu. Wicaksono@InIn.com.
Comparing Erlang C and Simulation ModelingContinued from page 10
Performance management at the agent or macro level centers on reporting from varying sources. Some are provided by workforce management or resource planning; others may come from different teams. There can be a variety of stakeholders in the audience for the reporting. Seldom do all audience members view the metrics and data from identical points of view. These differ-ences in interpretation can lead to discrepancies in how the data is used to manage. Ultimately what makes reporting successful as a part of Business Performance Management is the development and defining of the KPIs that set the scope of what is to be considered success within an organization.
Workforce management is an intersection of theory and practice providing a unique perspective to support both the reporting and definition of the organizational KPIs across an enterprise. Let’s explore ways your team can have a positive impact on and drive Business Performance Management. When a team understands how to facilitate the process of defining and managing to KPIs, the value add is immediate.
Most workforce management teams already support performance management in some way. They track and trend volumes, handle times, and deviations from forecast. In many cases, these same teams produce the scorecards/performance tools by which efficiency or quantitative KPIs are managed. For these teams to evolve to full performance management partners, they first have to understand and work in the departmental scope of influence. Having an expert workforce management team is the foundation of success. Understanding how your own expertise fits into the total equation then lets you view the big picture more clearly.
The most basic scorecard includes metrics around handle time, schedule adherence, and quality. These metrics are at the core of agent qualitative and quantitative stats: are you fast and good? When someone in workforce management understands and masters the most obvious piece – efficiency – they can better understand the full impact of all statistics. A group can be fast but have low quality scores. The appearance is efficient, but in reality the first contact resolution rate is so low there is a higher overall work load. A workforce management team can take this to Quality and Education and see if there is a way to partner to slow down calls in order to reduce volume. The net effect may mean an adjustment to AHT that should reduce volume. Knowing your area in workforce management and using your vantage point allows you to build to beneficial synergies with other teams.
Look at another scenario involving occupancy – service levels are 10% below target. Occupancy rates are hovering around 60%. They have been about 60% for a while now. This scenario sets off a red flag. Management and senior leadership all want to know why service levels are less than optimal while it appears no one is busy! What is going on in workforce manage-ment? We’ve all been in a situation like this. All the relevant data isn’t in the hands of the decision makers. So the immediate
assumption is that schedules must be off. This could be the case. What about adherence? Anyone with the workforce manage-ment view of things can tell you one data set will not provide enough of a story to deliver the needed message. In a workforce management analysis, you would start with adherence scores. If adherence is within acceptable limits then look at scheduling efficiency. If scheduling efficiency is also correct, it may be time to check agent skilling and configurations. In the end, while the “simple” steps may make complete sense to someone in your seat, they may be completely foreign to your average manager, business analyst, or executive.
So how does one take a team that is viewed as mostly “scheduling” and transform it into the pacesetter in performance management?
First, you have to prove you know what you talking about. Don’t just define a problem – show you can innovate a solu-tion. Create reports and dashboards that demonstrate your core expertise. Begin by looking at your direct scope of influence. Does your team do traditional workforce management: staffing, scheduling, forecasting and real-time? Do you handle payroll? Do you do scorecards? Do you do budgets, training, or help with QA? Take stock of those items that you touch. Once you iden-tify all you do, move to the next phase.
Ask yourself “How do I know we are on target as an organization?” What should the data say if we are doing well? What is the desired outcome? For example:
• I have 12 FTE worth of cost to facilitate contacts before cutting into the profit margin.
• I have 3000 calls a week with a mean average handle time of 300 seconds.
• I have a service level of 90% in 30 seconds I have to achieve.
• With my arrival pattern, I have a peak of 14 erlangs 2x a day.
• My forecasted shrinkage is 27% based on all available information.
• I am open 60 hours a week 9:00-9:00 Pacific Time. • With a perfect schedule, I would need 300 hours of
Based on these assumptions, I need to keep cost at or better than 12 FTE and I need to hit my service levels. I have my busi-ness goals. Next I need to define my KPIs. KPIs must be quanti-fiable, measurable, realistic, and relevant. With this in mind:
• Average Handle Time: By Organization / Team / Agent • Forecast accuracy: Is the volume and AHT within a
reasonable range given the available information?
The Role of Workforce Management in Business
Continued from page 1
• Adherence: Is the plan being followed? Are folks where we expect them to be?
• Occupancy: Does our staff utilization rate come close to the workload/staffing for service level?
• Is our cost in line with budgeted FTE?
Once we decide what should be measured we need only define reasonable targets that indicate we are on track. Now you have the ability to use data to quantify and guide the organiza-tion and team members toward a goal. By defining KPIs and empowering everyone to manage to those numbers, you have given the tools needed to help meet the big picture goals. The last piece of your puzzle is education. Everyone must be able to get the same messages from the data so that all goals, targets, and trends are understood. If you do this, workforce manage-ment will have evolved into a partner in total Business Perfor-mance Management.
Doing this may in fact make your job easier in the long run. Not speaking for everyone, but many of us get in an uphill battle just to get others to understand that reasoning for agents (and managers) to follow the agent schedules. By helping the management team and decision makers understand the data provided, and then using it to guide success, you help them turn it into information they can act on. Hopefully with your guidance they will then do so in a way that supports your agents, business, and the workforce management team.
This is simple enough. But remember you are unique. It has been six months since you have delivered an agent, team, and organizational level dashboard, scorecard, or performance management tool (PMT). You have been right so far; in managing to these metrics we have seen continued success. So much so in fact there have been successful sales. Your call volume will go up by 15%. However, high administrative costs negate any new FTEs. What can you do? Are your KPIs still relevant? You know that the cleanest ways to decrease staffing needs are through decreased volume or increased ability to handle volume. Since volume will go up and you can’t add people – someone must find efficiencies. Your team simulates that if AHT can be reduced by 45 seconds and shrinkage reduced by 5% then you should be able to handle the new volume. Now using your view – you become an internal consultant.
Here is where you unlock the power of your unique position, from which you report and inform business performance. From this vantage point you can now guide business direction, process, and efficiency. You use your tools to identify ways to reduce shrinkage and speed up a contact. You suggest shorter team meetings, push-training, refreshed call flow, etc. – You find ways to reduce shrinkage and AHT or run simulations with some changes. No longer is data just delivered. You help chart the course for further success. This is an area where many workforce management teams can add value in ways they do not today;
becoming business performance consultants. No one else can do this as quickly and completely as you can.
To get started you need to capitalize on being unique, but don’t be removed. If your team does not currently have interac-tion with the performance management process get to know the people that do. Get to know the stake holders better. Then start the process by defining scope, goals, KPIs to deliver in reporting. Then when you see KPIs move out of range offer insight to root cause. Include suggestions for issue mitigation. Don’t just report problems or successes, offer insights into solutions or plans to sustain progress. If your team currently provides reports but does not join in on the evaluation or KPI goal setting, reach out to your users and start including “notes” in reports. Provide a brief overview of what workforce management can bring to the table. If your team is already providing reports and interacting in the process you are most likely further along than many out there. There are still items you may be able to expand on. Evolve what you currently provide. Find what would be considered additional value adds.
Advising KPI adjustment can fit into what you already do just by adding a little thought and commentary. There was a group with an adherence goal of 98%. No one was meeting it. It seemed that no matter how many times they coached the agents they could not get down to this “achievable goal.” After a few minutes of research, it turned out the average handle time for a call was around 20 minutes and the group and activity was scaled in such a way that most agents did not have much if any time in between calls. The basic math showed that 98% only allowed less than 10 minutes per day of non-adherence. Breaks and Lunch provided a minimum of 3 opportunities for an agent to be stuck on a call when they should be doing something else. A little reporting later and it was determined that 63% of every break or lunch started while the agent was on a call and there was no allowance built into the tools to deal with this. In short, the adherence goal was made without any thought to the reality of the business. Now the staffing expectations are more readily in line with realistic business goals. This is the power of informing the process not just reporting on it. You are unique – you can see these dots and connect them.
Much has been written along the way by wise folks and a few MBAs. You can find many resources on Business Perfor-mance Management theory. But you are unique. You can see and interpret all of the inputs and data points. Just remember that the success of a business is determined by reaching its goals. You are unique in that you can help draw the map and be the GPS to get to the destination. As long as you remember it’s ok to adjust the route you’ll be fine.
Marshall G. Lee serves as Workforce Manager at Carewise Health. He may be reached at email@example.com. Justin D. Zeigler is Sr. Planning Analyst at American Express Global Business Travel. He may be reached at Justin.D.Zeigler@aexp.com.
The Role of Workforce Management in Business
Continued from page 2
W F M S u r v e y R e s u l t s
This is one of the more challenging elements of WFM for the back office since the ACD tracking and reporting tools we are using for calls are not typically going to be effective for non-call activities. Determining when a back office task arrived, when the agent began the work and ended it, and the total handling time for each is not always as readily available as needed. Getting a data link into the systems that the staff use and ensuring that the data is accurate for WFM purposes can be one of the more difficult parts of a back office WFM implementation.
How Daily Statistics Are Reported
When asked how they report daily statistics from back office work tasks, 23% indicated that they report only by work type, while 10% report only by agent
or staff member. However, two-thirds (67%) of the respondents indicate that they report by both work type and staff member that likely enhances the value of the data to the managers of these back office functions. To a large extent, this capability is a function of the data available from the systems being used to capture the activities.
Verification of Self-Reported Work Items
Without the detailed reporting capabilities to automatically identify the activity levels, WFM in the back office is some-what dependent upon the self-reporting of the staff. When asked how
these self-reported activity levels are verified, 40% indicated that they do not verify the numbers. One-third use system logs while another 20% does a random sampling of those work logs. Tracking activity levels in the back office is a new func-tion in some of these departments and may take some time to become complete and accurate, especially where the process is dependent upon the agent to make some record of their activity. Given the challenges call centers have in ensuring that agents use the ACD work state buttons correctly, it is reasonable to expect similar types of challenges in the non-phone work when the process is dependent upon the agents’ behaviors.
Monitoring for Schedule Adherence
When asked how the team monitors adherence to the planned schedules, 57% indicated that they use a
vendor-provided WFM system. Excel spreadsheets are used by 18%, while 7% use a home-grown prorietary system, leaving 18% who use some other method. Given the sequential nature of back office tasks, if this is the only work this person and team is doing, adherence to schedule details may not have the impor-tance that it does for inbound calls where the service goal is in seconds. This may be a situation where we need to “pick our battles” and put less emphasis on exact timing of breaks and more on the quality of the data recorded to ensure accurate forecasting of the workload to be performed.
Staff Effectiveness Factors
Survey partici-pants were asked to identify what factors they use to deter-mine the effective-ness of the staff. Respondents were able to make more than one selection
if appropriate and the average number of choices was two. Of all of the responses, one-third use supervisor observation and another 27% utilize the annual performance review. Self-reported productivity measures such as tick sheets were chosen by 20% and rework rates by 8%. This is another significant chal-lenge of managing the back office operation.
Agents Handling Both Phone and
Back Office Tasks
The survey participants were asked if their inbound phone agents also performed back office tasks. Almost two-thirds (62%) indicated that they do have mixed responsibilities but 35%
responded that their inbound call agents do not handle back office activities. In some centers, agents switch back and forth in blocks of time so that they can get a break from one type of work and do another for a while and then be available during peak load periods for call handling.
Percentage of Back Office Work
When asked what percentage of the back office work the inbound call agents perform, more than half (54%) indi-cated that it is small – less than 10%.
However, nearly one-quarter (23%) handle between 11-25% of that work while 18% handle between 25-50% of the back office tasks. Only 5% indicated that their inbound agents do more than 50% of the back office work.
Workforce management for the back office is evolving rapidly. As customers choose a variety of communication methods in addition to phone calls and as organizations begin to realize the benefits of managing staffing in non-call depart-ments, this will continue to grow and change. It is interesting to see how many of the respondents are already deep into the forecasting and scheduling for these tasks but those who are not yet doing them can reasonably expect the demand to increase in the near future.
Implementing back office WFM is not a small undertaking. It will require finding a reliable mechanism to identify the work-load (both volume and handle time). However, once that hurdle is overcome, the benefits of improved manageability of these functions is significant and well worth the effort in most cases.
• The caller is interrupted by something more urgent. • The delay announcement answered the caller’s question. • A predicted wait announcement indicated the wait is
longer than the caller wants to experience at that time. • An alternative offered in the delay announcement was
chosen (e.g., website).
• A “virtual queue” offering allows the caller to key in a telephone number for a callback but the call remains in queue even though the caller has hung up.
Analysis of the time the caller spent in the queue before abandoning can provide useful data. Most phone systems can provide a report that gives the breakdown of when people abandoned at various intervals such as at 5, 10, 30, 60, and 120 seconds. These times can typically be set for the intervals that would be most helpful in understanding caller behavior. If, for example, a significant percentage of callers abandon at 5 or 10 seconds into the queue, the length of the wait is probably not the primary driver. It is common to see the number of aban-dons early in the wait be fairly low but at some point, the curve takes a sharp upward direction. This is the point that indicates caller tolerance for wait has been exceeded and can be a good indicator of what the service level or ASA goal might need to be. (Remember, of course, that if the tolerance level appears to be about 40 seconds, the service level will be some percentage in the number of seconds so you might want to set it a bit shorter than 40 seconds. ASA is the average, so a lot of callers will wait longer than that goal, so adjust accordingly.)
Setting a goal for abandonment rate is a slippery slope. It is not a mathematically predictable number as it is totally a func-tion of human behavior and will change due to several potential conditions. For example,
• The urgency of the need/desire
• The alternative options available to solve the problem • Time of day (caller might not be willing to wait when
calling on their break from work rather than in the evening at home)
It is not uncommon for abandonment rates to vary by time of day and day of week as well as in response to what-ever stimulated the call. The behavior around a surge of calls responding to a time-limited marketing offer is one example. Another would be the flood of calls that would be received at an electrical service provider the moment the lights go out. In the latter case, if the delay announcement tells callers that the company is aware of the problem in a certain area, some portion of the callers will hang up rather than wait for an agent. They are satisfied and these are “good abandons.”
The typical Erlang-based models are not effective for predicting abandon behaviors. There are some models that accept a prediction that you specify and then plan the agent staffing model assuming those calls won’t have to be answered. There are also simulation systems that can be used to take abandonment into account, but they too depend on the human prediction of what the abandon behavior will be. There really isn’t an accurate way to predict staffing models based on aban-donment rates, so turning your goal into a service level or ASA is commonly required to generate schedules.
If you over-predict abandons and callers stay in the queue, you will see a significant negative impact on the service level due to the extra work. If you under-predict abandons, you may have more staff than you need and deliver a better than planned service level. Determining the right approach to planning for abandons is something that every WFM team should undertake. Educating the entire team on the implications of the options available is an important part of that project.
Maggie Klenke is a Founding Partner of The Call Center School. She may be reached at Maggie.firstname.lastname@example.org.
Continued from page 1
Continued from page 14
Congratulations to the Industry’s
Newest Certified Workforce
Robert Adagbedu, AscendOne Corporation, CWPP Jud Beall, BCD Travel, CWPP
Philip Kuhn, ServiceOntario, CWPP Andrew Simmons, ICBC, CWPP Elina Weckstrom, Cygnific, CWPP
These individuals have passed all three CWPP tests and successfully presented a project to a panel of certified workforce professionals. Congratulations!
For more information about how you can earn this prestigious designation, please visit our website at www. swpp.org/certification.
Visit us on the web at www.swpp.org
Society of Workforce Planning Professionals
6508 Grayson Court
Nashville, TN 37205
Join the Society of Workforce
Become a part of an organization designed specifically to facilitate professional development of workforce planning professionals. SWPP provides its mem-bership with a variety of benefits, including this quarterly newsletter, regional networking meetings, online forums, an annual conference, and more.
Membership in SWPP is available to anyone in the workforce planning or related profession.
There are three distinct levels of membership in SWPP: individual/associate membership, site membership, and corporate membership. One of these memberships is right for you! Membership costs vary with the type of membership. Prices for each membership are as follows:
• Individual/Associate Membership $295 USD • Site Membership (up to 3 members) $595 USD • Corporate Membership (unlimited number of members) $4995 USD Membership applications are available on the web at www.swpp.org. Still have more questions? Call us at 877-289-0004 or email us at email@example.com. We’d love to hear from you!