Telephonic communications – Centralized switching system – Call distribution to operator
Reexamination Certificate
1999-12-15
2003-09-02
Matar, Ahmad F. (Department: 2642)
Telephonic communications
Centralized switching system
Call distribution to operator
C379S265060
Reexamination Certificate
active
06614903
ABSTRACT:
FIELD OF THE INVENTION
The invention relates generally to call centers or other call processing systems in which voice calls, e-mails, faxes, voice messages, text messages, Internet service requests or other types of communications are distributed among a number of service agents for handling.
BACKGROUND OF THE INVENTION
Call centers distribute calls and other types of communications to available service agents in accordance with various predetermined criteria. Existing call centers suffer from a number of drawbacks. For example, as will be described in greater detail below, such centers are generally unable to provide satisfactory techniques for branded service level processing, dynamic agent pooling, and dynamic use of predictive algorithms.
With regard to branded service level processing, the concept of branded customer service is generally understood to include the consistent delivery of a particular level of service in order to reinforce customer perception of the brand and cultivate long term loyalty. In existing call centers, there is a growing need to make sure that work tasks are performed in a targeted range of time, with minimal “over-service” or “under-service” of the end customer. However, existing call centers often simply assume that a shorter wait time is always preferred over a longer wait time, i.e., there is no concept of an over-service condition.
Although such an approach can be adequate if all customers are considered equally important, it fails to provide an appropriate tiered structure of customers or types of work that vary in urgency. For example, a conventional call center is generally not configured such that each call or other work task may start service within a pre-set range of time with minimal over-service or under-service. Each agent is assumed to be able to handle one or more types of skills, but the actual state or potential state of over-serving or under-serving on each of these skills is generally not known each time the agent becomes available. In addition, there is generally no clear understanding as to whether adding service at another call center site would improve the ability to handle the call or other work task in a more desirable amount of time.
Conventional call centers are also unable to provide an adequate metric for characterizing branded customer service. Such call centers generally use either Average Speed of Answer (ASA) or Percent in Service Level (PSL) as metrics for planning and reporting on their operations. However, each of these metrics can be problematic in that it can appear “good” while in fact masking poor operations. For example, a good ASA may be comprised of many zero wait times and a few unacceptably high wait times. As another example, two call centers may each report a PSL of 80% of calls answered in 20 seconds, but one of these two centers may have answered all of those calls in 0 seconds and the remaining 20% in 60 seconds, while the other center may have answered the 80% between 5 and 20 seconds and the remainder within 40 seconds. Clearly, the second call center is doing much better at delivering a consistent service level, but the PSL metric of 80% in 20 seconds does not reveal the true differences in performance. These conventional metrics thus allow an under-service condition to be hidden effective branded service level processing requires a metric that penalizes for both over-service and under-service conditions, while also rewarding for service provided within a defined branded range. Such a metric is generally not available in conventional call centers.
With regard to agent pooling, it is apparent that the number of agents working on a particular skill needs to be increased in order that actual or potential caller wait time problems, e.g., too long a wait, are cleared as rapidly as possible so that operation returns to the branded service level. Similarly, the number of agents working on a skill needs to decrease as the service state drops to an over-service condition in order to avoid such a condition.
Conventional call centers rely primarily on manual intervention to change the number of agents working on any particular skill. Agents are added to improve performance or removed when performance is better than required. Although some existing call centers automate much of this process, such centers nonetheless fail to allow for dynamic, predictive changes in agent pooling for both under-service and over-service conditions.
With regard to predictive algorithms, it is known that call center operations can be dramatically improved when such algorithms are used to predict consequences of actions. These algorithms and their corresponding predictors can be very accurate when call volumes are sufficiently high, and decisions made using these predictors in this situation can be very effective and beneficial. For example, a prediction can affect whether a call center allows more or fewer agents to handle calls for a particular skill. Insufficient accuracy can therefore lead to undesirable under-service or over-service conditions. It is therefore very important that the predictor be accurate. Unfortunately, as call volume decreases below a sufficient level, predictors can become less accurate.
Decisions made with a less accurate predictor may still be effective enough to have a valuable impact. Eventually, however, with a low enough call volume, the accuracy of the predictive algorithm becomes suspect to the point that the predictor should be disregarded and decision making should be based on other factors. Conventional call centers, however, have been unable to provide suitable techniques for determining whether a given predictor is of suitable accuracy for use each time a decision is to be made based on that predictor. In addition, such call centers fail to provide adequate alternative decision factors for use when a predictor is not sufficiently accurate to be used. Low call volume situations are often encountered at night or on weekends in a call center. Such situations have been worked around using, e.g., time of day/day of week branching in vectors which govern call flow. However, this approach is generally not useful if a low call volume situation does not regularly occur under specific time of day/day of week conditions.
As is apparent from the foregoing, a need exists for techniques for providing features such as branded service level processing, branded service metrics, dynamic agent pooling and dynamic use of predictive algorithms in a call center.
SUMMARY OF THE INVENTION
The invention provides methods and apparatus which improve the processing of calls or other communications in a call center. More particularly, the invention in an illustrative embodiment provides techniques for implementing branded service level processing, branded service metrics, dynamic agent pooling, and dynamic use of predictive algorithms in a call center.
In accordance with one aspect of the invention, a call center is configured to determine which of a number of designated service states is associated with a particular skill or type of communication supported by one or more agents of the call center. A particular one of the states represents a branded service level, while other states represent over-service and under-service conditions. If the particular skill or type of communication is determined to be associated with a service state other than that corresponding to the desired branded service level, a communication processing function of the call center is adjusted so as to return the skill or type of communication to the desired branded service level state.
In accordance with another aspect of the invention, a branded service metric may be used to characterize the performance of the call center with respect to the desired branded service level. The branded service metric preferably provides a certain maximum score for handling a communication within a specified branded range, and a lower score for handling a communication with a particular amount of over-service or under-service. The branded service metric may also
Flockhart Andrew Derek
Foster Robin H.
Kohler Joylee E.
Mathews Eugene P.
Avaya Technology Corp.
Matar Ahmad F.
Nguyen Quynh
Ryan & Mason & Lewis, LLP
LandOfFree
Methods and apparatus for service state-based processing of... does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Methods and apparatus for service state-based processing of..., we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Methods and apparatus for service state-based processing of... will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-3049135