Today's call center environments are becoming increasingly complex by handling multi-lingual calls, and supporting diverse clients, products, or services. Efficient labor management dictates that agents in such environments should be cross-trained in multiple skilled activities. The need for a Workforce Management system that can comprehend and support these multi-skilled agent environments becomes increasingly critical to overall call center performance and profitability. Some ACDs are able to intelligently route these diverse calls to multi-skilled agents. But to realize any advantage from this ability, the Workforce Management system must be able to intelligently schedule agents to optimally meet skilled requirements.

ACDs that support skill-based routing can present calls that arrive in different queues (that require different call handling skills) to the same multi-skilled agent. Skill set scheduling is the ability to optimally schedule multi-skilled agents to meet an integrated set of multi-skilled requirements. Call centers using multi-skilled agents will operate in one of two environments: all agents are cross-trained, or some agents are cross-trained while others only handle a single type of call. The matter is complicated when more than two skills are represented and agents are cross-trained in varying skilled combinations.

Figure 1.  This figure illustrates how the ACD will automatically route calls to agents whose skilled call handling capabilities have been defined within the ACD.

For example, one agent may speak English and Spanish, while another might speak Spanish and German, and yet another English and German. A Workforce Management system must be able to account for agents possessing multiple skills when developing requirements, and before producing a schedule for either of these multi-skilled call center environments. Any Workforce Management system that cannot recognize these abilities will encourage overstaffing. For example, an ACD recognizes that an agent is able to handle calls from a queue offering English speaking callers and a queue offering Spanish speaking callers. Each queue represents a different call handling skill. As the calls originate, the ACD will route them from both of these queues to the agent. A Workforce Management system that does not support skill-based scheduling might schedule the agent to fulfill the requirement of one position for English and also schedule another agent to fulfill one position for Spanish. However, since the agent handles both types of calls, the service levels for both queues are beneficially affected, resulting in a higher quality of service than was intended, and a correspondingly higher cost of operations because of the staffing overage. It is this failure to allow for an increased occupancy level by multi-skilled agents that causes the overstaffing.

Forecasting algorithms, such as the popular Erlang-C model, are used to blend the projected call volumes, handling times, and service level objectives of different skill-based queues, and arrive at an integrated set of requirements for all skilled activities. Another industry standard forecasting algorithm, Merlang^®, will additionally consider busies and abandoned calls when generating a forecast. These two factors affect the number of calls that a queue will contain at any time, and if not taken into account, will indicate an incorrect number of calls in the queue. The incorrect number of calls handled, when used with historical data as the basis of a future forecast, will represent an inflated call volume. A forecaster generating requirements based on this inflated call volume will then be misled, and inevitably overstaff. Service level objectives must be specified for each skilled set of queues and may be defined for any period of time. Workforce Management systems should allow you the flexibility to define service objectives for as short as five, or ten-minute intervals. This will provide maximum control over call handling and staffing concerns, should your call center operation require it.

Another important consideration for a Workforce Management system is the way in which the ACD will route calls to multi-skilled agents. There may not be an even distribution of call types arriving for a multi-skilled agent. For example, perhaps the agent is eligible to receive calls from three queues providing English speaking callers, and only one queue providing Spanish speaking callers. Assuming equal call volumes in all queues, the agent will receive more English calls than calls requiring Spanish. Any Workforce Management system, while generating requirements that fails to take this imbalance into account will again result in overstaffing. This situation is addressed by recognizing the concept of primary and secondary (tertiary, etc.) skills. In the above example, English would be specified as a primary skill and Spanish would be defined as a secondary skill. These skilled team eligibilities are prioritized to allow the scheduler to better emulate the mixture of calls that the agent is likely to receive and fulfill the requirements accordingly. Workforce Management systems that do not recognize the concept of primary and secondary skills will also have problems when working with a call center that has a mixture of multi-skilled and single-skilled agents.

After generating requirements, the Workforce Management system must schedule agents to fulfill the requirements generated for each set of skilled call types. Some Workforce Management systems attempt to resolve the skill-set scheduling problem by mimicking the agent to queue relationship defined in the ACD. This allows the scheduler to meet requirements based upon the ACD's understanding of agents' skills. It does not however provide any visibility as to how the requirements for each skill have been met, and whom is doing what. By grouping agents as skilled teams that share common skills and handle calls from common queues, and associating those teams with ACD queues, a Workforce Management system is better able to determine skilled requirements and provide the agents to meet those requirements. A team may handle calls from a set of common queues, and more than one team may handle calls from a single queue. Individual service level objectives are defined for each team. Staffing requirements are generated for each skilled team, and the number of agents available to work in a skilled team is always known at any time.

Figure 2.  This figure illustrates the possible relationships between queues and teams. One or many queues may service one or more teams.

There are many advantages to this team grouping scheme. It provides clearer visibility of call center staffing requirements and issues. Without this important concept, call traffic reporting can only be based on individual or aggregate queue traffic, and reporting on more than one group of agents that handle traffic from the same queue is not possible. Without the team concept weekly schedules for agents can only indicate the hours that the agent is expected to be at work, not the skilled activities that the agent is expected to perform. In this case, if an agent called in sick, the call center manager would not be able to know which of the skilled call types would be affected by the absence.

The Workforce Management system will schedule multi-skilled agents to skilled team activities for each day of the scheduling period. To achieve an optimal schedule over the scheduling period, agents must have the flexibility to work at more than one skilled activity during the same day. This necessitates daily eligibility rules that constrain the types of skilled activities to which the agent may be assigned for each day of the week. Another important property of skill set scheduling is a Workforce Management system's ability to integrate off-line or non-demand activities with inbound call handling duties. Requirements are generated for non-demand activities, and evaluated and fulfilled by the scheduler according to agent eligibilities when producing a schedule.

Additional adaptive functionality is required when a Workforce Management system supports a skill-set scheduling solution. Reporting capabilities need to inform the call center manager of metrics for commonly skilled groups of agents. Deviation reports must be constructed for these teams of agents, giving agent team level visibility to a common set of skills. Daily Assignment reports for agents need to reflect each daily skilled assignment, informing both the agent and supervisor as to what call handling skills are expected of the agent during each day of the scheduling period. Real-Time Adherence must also support skill-set scheduling. Real-Time reports must reflect agent adherence to each skilled activity. These reports show the call center supervisor which activity the agent was scheduled to perform and which activity the agent is currently performing. Supervisors then have continually updated visibility of the performance of each call skill type.

Skill-set scheduling, if fully implemented, can help reduce overstaffing in complex multi-skilled environments. The complexity of the skill-based environment will dictate the required flexibility of the solution. Any skill-set scheduling solution must be flexible enough to handle the skilled dictates of your unique call center operation.