Abstract: An automated method for servicing a plurality of work items within committed times. A workflow including two or more work activities is assigned for each of the plurality of work items and a commitment is assigned either to each workflow or to each work item in each workflow. Queued work items are automatically assigned to a next available agent based on an activity state. A short-term predictor provides early detection of work items that are at risk of failing to meet their commitments. A long-term predictor detects backlogs of work items where excessive delays in queue are likely to put work items at risk of failing to meet their commitments. An activity's state can also cause additional reserve and backup agents to be assigned to it, to service work items from its queue before their commitments are missed.

Abstract: In a computer integrated telephony call center where an adjunct host (160) normally controls enqueuing of calls (168) in call queues (120) of an ACD system (101) and corresponding call queues (120′) of the adjunct host and assigning of the enqueued calls to agents, the ACD system assumes (604) control over its call queues upon loss of the control by the adjunct host (e.g., upon failure (600) of the adjunct host). The ACD system flags (602, 606) all calls that are enqueued at the time of loss of the adjunct host control as well as all calls that the ACD subsequently enqueues as being under ACD system control. When the adjunct host regains its control ability, the ACD system stops flagging new calls, call queues of the host processor are cleared (702), and the adjunct host regains control (722) of enqueuing calls in call queues of the ACD and the adjunct host.

Abstract: In a computer-telephony integrated (CTI) contact center, a CTI adjunct (160) enqueues contacts in contact queues (184) of the CTI adjunct, but also causes contacts that are calls (168) to be enqueued as ACD calls in ACD call queues (120) of an ACD system (101), whereby the ACD system and its management information system (MIS 110) provide ACD features to the calls. Similarly, the CTI adjunct enqueues agents in agent queues (185) of the CTI adjunct, but also causes agents (102-104) that have call-handling skills to log into and to be enqueued as ACD agents in ACD agent queues (130) of the ACD system, whereby the ACD system and its MIS provide ACD features to the agents.

Abstract: A call center (100) rejects further calls when the sum of the present time, the anticipated call in-queue wait time (112), and the anticipated call service time (114) for this type (106-108) of call exceeds the closing time of the call center. The call center may redirect the rejected calls to another call center if it determines that the other call center can service them by its closing time.

Abstract: The flow of work items (40) through a workflow process (50) is optimized by repeatedly reordering (FIG. 3) work items enqueued in inbox queues (21) of workflow process tasks (500) to maximize results according to a given business strategy expressed through target times. Each enqueued work item has an associated in-queue rating (IQR 28) that represents the number of queue positions (23) that the work item can be retarded or needs to be advanced to meet its target time. When a work item enters a queue and whenever a work item changes its queue position, its IQR is computed. An optimization function is then performed (404) on the queue to determine an order of the enqueued work items that optimizes a metric of those work items that may fail to meet their target times. The work items in the queue are then reordered (406) accordingly.

Abstract: In a customer contact center (100), a plurality of customers' communications are serviced simultaneously by one resource (120-128) (agent or port). A conferencing function (136) connects the plurality of customers (110-118) to the one resource and/or to each other. When a resource becomes available, a batch service function (140) determines (206-226), for each skill of the resource, the value to the contact center of having the resource presently serve communications needing that skill, and uses the conferencing function to conference (234-236) a plurality of communications needing the skill having the highest value with the resource. The conference connection may be listen-only, listen-and-talk, or listen-only/then listen-and-talk.

Abstract: The alternate destination redirection (ADR) feature (102) of telephone switching systems (101) or an equivalent is used to implement a “post-route” routing architecture having the benefits of a “pre-route” routing architecture in a network ACD (FIG. 1). The ADR feature is administered in the network (100) for individual ACD systems and individual call types at each ACD system to identify another ACD system as an alternative destination for calls of the individual call type rejected by the individual ACD system. The network distributes (302) calls to the plurality of ACD systems (110-112) on a basis (e.g., fixed percentage, round-robin) that does not require the network to know the status of the individual ACD systems. Upon having a call of an individual type routed thereto, an individual ACD system checks (304) the status of the ACD system that is administered as the alternative destination for its rejected calls of the individual type.

Abstract: Call centers (143-145) use an event-driven scheme (207) with a filter (366) to send status updates to an associated network call-routing system (150). The filter ensures that only minimal status changes are not reported to the network call-routing system. Each call center records (310) the status values (360-365) of different splits that it last sent to the call-routing system, and sends a new status update only when a status value of a split has changed (306) by the split's corresponding threshold amount (380-385). On the one hand, in large splits, the number of status-impacting events (such as call arrival, call serviced, call abandon, agent login/logout, etc.) is typically high, but the scale of the change in status effected by each of these events is minimal. On the other hand, in small splits, the number of status-impacting events is typically small but the scale of the change in status effected by each of these events is significant.

Abstract: Selection of a suitable call-center agent (106-108) to handle a call is based on which available suitable agent's handling of the call will tend to optimize call wait times. When a call needing a particular skill becomes available (200), all skills of agents in the agent queue (131-139) corresponding to the particular skill are determined (202, 204). The agent queues corresponding to the determined skills are checked (206) to determine which ones identify only one agent. For each available agent having the particular skill, the number of agent queues which identify this agent as their only agent is computed (208). The available agent having the lowest computed number is selected (210) to handle the call. This minimizes the number of skills that will be left without an available agent to handle subsequent calls, and thus tends to maximize the probability that a next call will also have a suitable agent already available and not have to wait for one to become available, thereby optimizing call wait times.

Abstract: Selection of a call-center agent (106-108) to handle a call is based on which available agent's handling of the call will tend to optimize call-center performance criteria such as efficiency (e.g., minimize per-call handling time) or derived benefit (e.g., maximize revenue). Each agent has a service profile for each type of call that they handle. A service profile (400-402) comprises present values of a plurality of service metrics, such as proficiency, profitability, customer satisfaction, and agent satisfaction. When a call of a particular type becomes available, the present values of the service metrics of the service profile (400-500) of that call type of each agent who is available to handle the call are combined (304) into a score according to one of a plurality of formulas which corresponds to that call type, and the agent with the best score is assigned (306) to the call.

Abstract: Selection of a call for handling by a call-center agent (106-108) is based on which call's handling by the available agent will tend to optimize call-center performance criteria such as efficiency (e.g., minimize per-call handling time) or derived benefit (e.g., maximize revenue). Each agent has a service profile for each type of call that they handle. A service profile (400-402) comprises present values of a plurality of service metrics, such as proficiency, profitability, customer satisfaction, and agent satisfaction. When an agent becomes available, the present values of the service metrics of the service profile (500-502) of that agent for each type (e.g. skill) of available call handled by the agent are combined (610) into a score according to one of a plurality of formulas which corresponds to that call type, and a call of the type for which the agent has the best (e.g., either highest or just-sufficient) score is assigned (612) to the agent.

Abstract: In a skills-based ACD, an available agent is reserved and assigned to handle calls needing a "rare" skill of the agent and is prevented from handling calls needing a "common" skill of the agent even if calls needing the common skill are waiting to be handled, if not reserving the agent for the calls needing the rare skill would deprive those calls of the last available agent, even if no calls needing the rare skill are available for handling. The agent is reserved only if target performance criteria, such as average speed of answer, for handling the calls needing the common skill are being met. In case of more than one agent being available to handle calls that need the rare skills, one of these agents is selected and reserved and the other agents are freed to serve calls needing "common" skills.

Abstract: In a call center (100) of a business, if it is determined (314-318 or 320) during the handling of a call involving a party who is a customer of the business that the party is at risk of being lost as a customer to the business, an identifier of the party, such as ANI, is captured (300) and stored (322) along with an "at risk" indication. The determination may be made either automatically by detecting (314-318) that the party terminated the call while on hold, or manually (320) by the agent handling the call or a service observer whereupon the agent or observer pressed an "at risk" feature button (112) on their terminal (105, 110).

Abstract: Call-center (FIG. 1) performance is improved by assigning different service-time objectives (222) to different types of calls or to call queues (21) for different types of calls, and then selecting (212), for an agent (25) who has just become available (200) to handle a call, a waiting call that is farthest along in exceeding its assigned service-time objective. The objectives represent limits on the amount of time that calls should spend waiting for agents before being handled. For example, a video call may need to be serviced within tens of seconds of its arrival, and a voice-only call may need to be serviced within minutes, while e-mail may need a response within hours of its arrival. Relative distance of calls from their assigned service-time objectives is preferably determined by determining (206) the calls' present or anticipated wait times and computing (210) weighted percentages of the assigned service-time objectives that are represented by the present or anticipated wait times.

Abstract: The maximum wait time for callers in a call center (FIG. 10) is lowered by selecting, for an agent (25) who has just become available to handle a call, a highest-priority waiting call that would most likely wait the longest if it were not selected at this time. Anticipated wait times are computed for the calls at the heads of the non-empty highest-priority call queues that correspond to the agent's skills or splits (202-210). The anticipated wait time of a call is computed as the call's present (elapsed) wait time plus the average rate of advance of calls in the call's queue (210). The call with the longest anticipated wait time is then selected first and is assigned to the available agent for handling (212-216). The process is repeated each time that any agent becomes available.

Abstract: A call-distribution function (150) of an ACD system (101) improves the equity of distribution of calls to agents (106-108) by basing the distribution on the agents' individual occupancies. Illustratively, determining an agent's occupancy involves either determining (304) how many calls the agent has handled within a predetermined time interval, or determining (305) how much time of a predetermined time interval the agent has spent on handling of calls. The occupancy is periodically re-determined, and a next call is allocated to the presently least-occupied one of agents who are available to take the call, or to an agent to whose occupancy the allocation will be most beneficial, such as an available agent to whom allocation of the call will maximize convergence of actual and target occupancies.

Abstract: In an automatic call distribution (ACD) system, an improved estimated waiting time arrangement derives a more accurate estimate of how long a call that is or may be enqueued in a particular queue will have to wait before being serviced by an agent, by using the average rate of advance of calls through positions of the particular queue. For a dequeued call, the arrangement determines the call's individual rate of advance from one queue position to the next toward the head of the queue. It then uses this individual rate to recompute a weighted average rate of advance through the queue derived from calls that preceded the last-dequeued call through the queue. To derive a particular call's estimated waiting time, the arrangement multiplies the present weighted average rate of advance by the particular call's position number in the queue. The arrangement may be called upon to update the derivation at any time before or while the call is in queue.