Low volume splits or skills

Last Updated : Aug 30, 2013 |

Very small resources, for example, 2-3 agents, have special needs. With BSR, you can maintain a balance of wait times across a network of call centers. However, for very small splits or skills, wait times for each call can vary significantly.

To see why this is the case, let us take an extreme example of a split with a single agent logged in with one call active and no call in the queue. Average call handling time is 3 minutes. Now, if a new call arrives in the queue, that call can be answered almost immediately or the call can be in the queue for more than 3 minutes. The variation in wait times is perhaps 5-180 seconds.

In general, the fewer agents logged in to a split or skill, the greater the variability in wait times because agents become available less often. BSR favors large resources, steering calls away from smaller resources when there are no available agents or wait times are not the best in the application. This tendency helps reduce the possibility that an individual caller can have a disproportionately long wait at a small resource.

If your network includes very small splits or skills, you have three options:

  • If the operation is not badly affected by a small percentage of calls having variable wait times, use BSR across the network.

  • If your principal concern is that a call does not wait in a queue while an agent is available elsewhere, use BSR, but write primary vectors at smaller locations to perform rapid look-ahead attempts to other resources once the call is queued. Rapid LAI vector loops use the interflow-qpos conditional, which is an enhancement to LAI.

  • If you want agents to answer each call quickly, use the following configuration: Do not deliver or queue calls directly to the very small resources. Deliver or queue all incoming calls to larger resources and use BSR to balance the load across the larger locations. Ensure most of the larger locations perform rapid look-ahead attempts to more than one of the smaller resources. In this way, the members of the very small resource become an extension of the agent pool at one of the larger call centers.

In any network, do not have several large resources poll or make look-ahead attempts to a very small resource. As the status at the very small resource changes infrequently, frequent polls to that resource is a waste. A very small resource must receive look-ahead attempts or be polled only by other small resources or by one large resource.