The past ten years have seen significant advancement in the field of workforce management. The solutions have expanded their features to meet the ever increasing complexity of modern contact center environments. Features like agent self service, messaging and outsource management modules are all adding more and more value to workforce management solutions. It is exciting to watch the products grow, but we can become distracted by these "new" components and lose sight of the key business drivers that have made workforce management solutions indispensible to contact centers.
There are three important questions that should always be in the mind of analysts or call center managers that are at the core of effective workforce management practices. Effective workforce management is about consistently delivering your targeted grade of service and managing the cost of providing that service. Three questions will keep you focused on meeting these primary objectives and will keep you looking out for your customers and the bottom line of your business.
How much work is there to do?
Your workforce management application must be able to accurately forecast and dimension the upcoming workload. Workforce management applications generally rely on history to create models that can accurately predict volume and average handling time by skill, queue or activity. If you cannot accurately predict and properly dimension the workload by day, week and month, it is pretty unlikely that your service level objectives will be met.
Not all applications are equal in this regard and it would serve you well to understand exactly the core forecasting functionality of your solution or the products you may be considering for your business. The forecasting features must be evaluated against your business requirements. Here are some examples of why it is important to understand the forecasting approach that is applied.
Some forecasting models use moving averages of your history to create forecast for the future. They might use the most previous ten or fifteen weeks of history. You may be able to weight those weeks. Averages have limits. If you are on a steep call curve in either direction an average tends to lag the actual. So you need to be able to build in more aggressive numbers. If the application you are evaluating uses a "static" or average forecasting approach, you need to see how that is going to perform in your business. If you have high volatility this might not be your best option.
Other models may use a "dynamic" forecasting model to create the forecast. This means that rather than using a moving or weighted average of some number of weeks you can specify exactly which weeks you would like to use to forecast say an upcoming like week. For instance you could use the first week of every month for the past ten months to forecast the first week of next month. This could be pretty handy in certain business environs.
Other models enable you to create forecast based on date ranges. Date range forecasting enables an analyst to use the data from the first quarter of last year’s history to forecast this year’s workload and conveniently add ten percent over the top.
What if nothing in your history reflects what you know is going to happen next? If you are company that is growing by acquisition and next week you are going to turn on a volume that is simply not in history, it would be important to be able to either easily manipulate existing data to transform into more relevant data. That data probably exists, but is not in the history. So ideally the ability to upload that data is an important feature for a business managing this kind of growth.
Ideally you should be able to call upon whatever method suits your purpose. Some products support exactly this scenario so it is critical that you understand the limits of the forecasting tools within the applications you have or are evaluating. Accurate forecasting is crucial to customer satisfaction and for creating a schedule that controls the cost of providing great service.
How many people do we need to do the work at a designated grade of service?
Workforce management enthusiasts tend to want to look at this number in several ways. Your accurate forecast tells you how many hours of work there are to accomplish in each daily interval, day, week etc. First it is important to understand how many people we need to staff in order to do the work in pure work terms. The Figure 1 chart shows the forecasted workload and the resulting heads needed to complete that work at a designated grade of service for five days in March. The two curves match reflecting the varying arrival of work into the center by day of week. This, of course, is also broken down by time of day, by day of week by your workforce management solution.
Figure 1
We now know that there are 632 staffing hours required in order to deliver our designated grade of service at the forecasted workload. The 632 hours are the result of using Erlang C or some variation that supports multi-skilled requirements generation. If the average hourly wage in the center is $10, the cost of staffing just the pure work is $6,320 for the week.
Next we need to understand how many people need to be staffed in order to accomplish the work given all of the overhead that our contact center experiences. Most call centers average around 30 % shrinkage but that percentage it is highly susceptible to each type of business and culture and can range from 20 to 50 %. Shrinkage factors include late time, sick time, jury duty, meetings, training, lunches etc. In our average shrinkage scenario that means that to utilize 3 agents who are actually processing transactions we must effectively staff just over 4.
It is important to understand just how much this overhead is costing the center in terms of wages. This is a crucial step in managing cost. It is advantageous to break down shrinkage as two distinct categories, discretionary and non-discretionary. These are the items, respectively, that we can control (meetings, training, vacation time, break time) and the elements we cannot control (late time, sick time, unaccounted for time.)
Figure 2
In this scenario we assume approximately 15 % discretionary shrinkage and about 15% non-discretionary shrinkage. The center has had to staff 30 % more hours on top of the pure work to have enough resources "in the chairs" to deliver our targeted grade of service.
We now understand because of shrinkage that we must staff 840 hours this week to meet our service level objectives. We have added 208 hours into the staff plan or $2,080 in payroll expense.
It is easy now to see how crucial it is understood and thus manage shrinkage. It is not cheap! Understanding non-discretionary shrinkage will enable us to better manage agent behavior. This leads us to better agent coaching, management and more effective policies. It also protects the bottom line. Managing the shrinkage we do control, discretionary shrinkage, leads us to our next question.
How can we optimize our schedule against the forecast?
Discretionary shrinkage presents opportunity when it comes time to actually create a schedule for the agents. This may sound a little at odds with the lessons learned so far. We need to discuss culture and schedule optimization to better appreciate the comment.
The traditional representation of workforce management software is that the algorithms optimize the schedules against the forecast, thus saving you thousands of dollars in payroll related expenses. That is a good thing right? Yes, but that is not exactly how most call centers operate.
The cultures of most contact centers result in scheduling practices that usually lean toward satisfying the agents and controlling turnover as opposed to meeting the needs of the business, especially in North America. Most agents are hired into a fixed shift and those are the hours until the end of a bid cycle or until told otherwise. Demand driven schedule optimization then from the perspective of routinely changing start times and changing shift types to meet the needs of the business has been largely eliminated. Optimization takes on a different meaning in this context and it is all about managing shrinkage.
Here is a typical illustration of fixed schedule environment mapped against a forecast or staff plan. Shifts arrive and depart in a typical three wave configuration. We have a total of 65 agents available for scheduling.
Figure 3
You can see that we have over and under staffing due to the fixed schedules of our culture. We can see, using our previous example as simply a reference point, that while the staff plan had a payroll expense of $9,224, the resulting schedule has even more hours than necessary in certain areas and less than required in others. We cannot take hours out where we do not need them, due to our fixed schedule environment, but what we can do is manage shrinkage to maximize the schedule towards the forecasted staff plan.
This is accomplished by using coaching to minimize agent behaviors we cannot control in the areas on non-discretionary shrinkage and optimizing the placement of discretionary shrinkage items like trainings and meetings.
It is essential, before we illustrate this point further, that managers understand the cost associated with the inflexible schedules and be able, if the opportunity presents itself, to develop and adopt better base schedules. The analysts should be presenting this data and analysis to management routinely based on long term forecasting.
Below is our illustration that indicates how we use shrinkage to our advantage to make the most of our fixed schedules. By moving the discretionary shrinkage portion represented by the green line from periods of understaffing to overstaffing we would dramatically improve service in the understaffed areas and we would take full advantage of the cost associated with this schedule and make best use of our resources.
Figure 4
It is easy to see how discretionary shrinkage optimization as opposed to traditional schedule optimization will leverage your overstaffing positions to minimize understaffing. We have not removed hours or expenses from the schedule but we have taken full advantage of the resources at our disposal and increased the level and consistency of service provided to our customers
Summary
Contact centers need to be able to forecast the workload accurately. They need to be able to generate headcounts and understand overhead. By understanding shrinkage and the realities available for schedule optimization they can offer more consistent levels of customer service and take full advantage of the cost associated with any schedule, by optimizing shrinkage.
Answering these three questions using your workforce solution will keep you focused on the essential aspects of your workforce management strategy, delivering service and managing cost. It may help develop policies for minimizing and managing non-discretionary shrinkage and provide intelligent choices for moving discretionary shrinkage to more opportune periods in the schedule.
You will be able to more intelligently evaluate practical alternatives that may move your agent behavior to a more flexible environment that controls cost and delivers even more consistent and higher levels of service, while reducing cost. Even if your agents are fixed for life, understanding the dollar value associated with the existing schedules and managing the opportunity within those constraints is a valuable exercise.
There are plenty of exciting features in today’s modern solutions and they are crucial to expanding the role that workforce management plays in the rapidly changing contact center space. Staying focused on core workforce management fundamentals though might just get you noticed or even entrance into the executive conference room.
About CommunityWFM:
The Workforce Management Software Group, Inc. (WFMSG) was founded in 2005 by a team of workforce management experts. WFMSG personnel were deeply engaged with legacy workforce management technologies and deployment approaches for Fortune 500 companies throughout North America and abroad. This expertise was distilled into the modern, cloud-based design of Community, now available in two performance versions, and our steps-to-success deployment and services models.
Published: Friday, November 6, 2009
I am checking out all the amazing and daily updated content on ContactCenterWorld.com and networking with professionals worldwide
Send To Friends Post On My Wall