With all of the needs, skills, and preferences defined, the order in which the ACD system selects agents is the next design task. Does it make a better sense to have calls first try agents who are generalists and overflow to specialists as needed? Perhaps the calls should first seek the with limited skills who match this caller's specific needs, overflowing to multi-skilled agents as a second choice.

In this latter case, the single skilled-agents (and even trainees) can be kept busy on the calls they can handle, reserving the highly skilled agents for more complex tasks. The design maximizes the caller's chance of talking to the newest and lowest-skilled person, keeping the long-term multi-skilled agents to reserve for the complex situations.

Alternatively, the multi-skilled generalists can be first, maximizing the opportunity of callers to reach the agents with broader skills, but perhaps not the depth on a single area needed for really complex issues. Then the high-skilled agent provides depth as the single subject specialist and receives calls as transfers from the generalists when they are not able to satisfy the caller's needs. This is a common plan in help desks, for example.

There are two major schedule design types. The first is the "ripples on the pond" design. For example, consider the center that has a four-tiered career development and a skill-based routing process. New agents learn a single skill and take calls of that type only (Type 1). Then they learn a second skill and added that to the first taking both calls types as needed. The third and fourth skills are added as the agent develops in the center.

The routing pattern is like ripples on the pond in the reverse of the agent skill development. Calls for the most complex calls can only be handled by the Type 4 agents, while Type 3 agents can handle any call for Type 1, 2 or 3. Type 1 agents cannot help out any of the other groups since they are only trained on Type1.

Using this design, the center has to be sure it has enough agents trained to be certain it can answer the Type 4 (most difficult) calls effectively. Since no other agents can help this group out if they are understaffed, this group serve as the base staff. Ideally this group should be overstaffed since any spare time they have, could be filled in with calls of nay of other three types.

Once Schedule for this Type 4 group are established, then the call for the type 3 calls is added to the type 4. The schedule for the type 4 group are set in stone from the previous step, and the schedule for the Type 3 agents can be determined. This process utilizes some of the spare time of the Type 4 agents to help out with the Type 3 calls.

A similar process adds the Type 2 calls to Types 3 and 4, and then finally Type 1 calls and agents. Of course, an automated workforce management tool with skill-based routing and scheduling software would do this more easily, but it can be done manually if necessary.

Unfortunately, in centers who use this design, it is common to find that they do not have enough agents to answer the type four calls as this is the most complex and requires the most training and experience. Types 1 and 2 are typically overstaffed with new-hires and low-skilled personnel. No amount of call routing will overcome this mismatch; only training to raise the skills will provide a lasting solution. But at least the design process will reveal the requirements for each skill level clearly so that the training plan can not be set in motion.

The plan above is a relatively straightforward one. However, not all centers are limited to a few skills. It is common today to find centers that have identified twenty or more skills and some have more unique skills than there are agents in the center. This routing and call sorting process looks more like the branches of a tree that ripples on the pond. This design typically use a multitude of telephone numbers and complex menu sorting to define the call type down to this level of details.

In this branching design, the skill definitions are set up to recognize those call types that require a very unique skill held by only a few agents. These agents with the unique skills sometimes have only this skill and answer no other kind of call. But more often, these agents also posses the more general skills needed to answer the basic calls for their business unit, and sometime for other business units.

Some agents have only the general skills, and non of the unique skills. In this design, the calls for the unique skills are dealt with first, as it is again important to ensure that there are enough agents with these skills scheduled to match these call volumes since no one else can take these calls. This agents are added to the pool for the general call types for that business unit much like the ripples on the pond approached described above.

Determining agent schedules must be done for all of the overlapping groups at once, While traditional routing allows each call type to be associated wit h a single group of agents and the staff requirement can easily be determined with the Erlang C Model, skill-based routing defies that model and requires a simulation to determine the combinations of skills and schedules that must be put in place to ensure that all of the calls and media will be handled effectively.

Some automated workforce management systems have these complex simulation packages, but others do not, so this can be differentiating factor in selecting such a system.

The question is whether call center have seen improvements in their performance, customer satisfaction, or cost-efficiency through the use of skill-based routing. And the answer is generally yes, when the design has been done well. For others, the challenges have been too much and they may have returned to traditional routing or a simplified skill-based routing design.

The management challenges of forecasting, scheduling, and tracking the personnel are more difficult then they were in the defined skill or universal agents worlds that preceded them. But the complexity of the businesses, the need for a career development and training plan, and the continuous expansion of both temporary and permanent additions of unique call types demand a response.

Skill-based routing balances the needs of the environments with the desire for economies for scale in agent group size, utilizing every resource in all the ways possible. Saving in the training process help offset the added analytical staff needed to make this skill, based routing process work effectively.

Care must be exercised in the design of the plan, determining the call types that can be used and the sorting method that will effectively route calls to the appropriate agents. A design that result in a high call transfer rate is not an effective solution. With an efficient and effective routing and sorting plan, a reasonable forecasting of workload can follow. That sets up a scheduling process that can match the agents to the callers.

Of course, scheduling is generally a long term issue as most centers do not do a complete change of shifts very often. Managing the day-today variations in the call load and handling the absences of specific agents with unique skill sets is a matter that has its own challenges.

Even the best plan is subject from daily aberrations, new products or services, changes in corporate direction, and so on. The skill-based routing plan must be compare with a procedure to adjust for daily workload and resources shifts as well as the longer-term changes. Otherwise, it quickly becomes out of sync with reality.

#contactcenterworld