Skills based contact routing

[0002] This invention relates to call center technology, and more generally, to contact center technology which allows different types of contacts to be serviced by the agent or other contact service employee most suited to service said contact.

[0003] Call center technology is well-known in the art and is critical to many large corporations providing service to their customer base. Call centers typically comprise a plurality of call center agents which service incoming customer inquiries, complaints, etc. A typical example of a large call center might be an airline reservation system, a credit card company, etc.

[0004] Recently, the call centers are more appropriately referred to as contact centers. This change in name is because the call centers may service more than just incoming or outgoing telephone calls. Rather, the call centers may also be involved in servicing other types of agent/customer interaction, such as web chat, e-mail, etc. Ideally, the call center architecture and logistics should provide the best possible service for all of these types of contacts.

[0005] In a large call center, there may be dozens or hundreds of different types of incoming and outgoing contacts. Additionally, complicating things more is the fact that dozens or hundreds of call center agents may have a variety of different skills. These skills permit certain agents to service particular types of contacts better or worse than other agents. In order to insure the best possible service for each contact, it is highly desirable to route contacts to agents best suited to service the particular type of need required by the particular customer that is the subject of the contact.

[0006] Some prior art systems use what is termed “skills routing” in an attempt to match calls with the agents best suited to service those calls. One common technique is to maintain different pools of agents depending upon the particular types of contacts expected. The contacts are then examined and routed to a pool of agents possessing those particular skills.

[0007] An example of the foregoing might be Spanish language. If a contact center anticipated that some callers require Spanish language agents with which to interact, then a Spanish pool might be set up. Calls coming in which indicate a Spanish speaking customer would be routed to the pool of Spanish speaking agents.

[0008] One problem with such a system is that even within the Spanish language speaking customers, there may be different types of services required. For example, there may be Spanish speaking customers calling up with questions about extra charges on their credit card, and there may be an additional group of Spanish speaking agents calling up to report lost or stolen cards. The groups of the Spanish speaking agents to handle these two types of calls may be different, and different agents may have different skills in this regard. In order to solve such a problem, the prior art call center technology would simply create two different pools of Spanish speaking agents. One for handling lost or stolen credit card reports, and another for inquiries regarding credit card charges.

[0009] One problem with the foregoing is that the number of pools of agents required grows rapidly as the number of different contacts grows. For example, a contact center requiring six different types of processes in ten different languages would require sixty pools of agents to handle those calls. Moreover, there is still no guaranty that the best agent will be assigned to the particular contact.

[0010] In view of the foregoing, there exists a need in the art for an improved technique of contact center processing which permits the optimal match between an available agent and a contact to be serviced.

[0011] Ideally, any such implementation should be flexible to allow for the addition and removal of agents, contact types, etc. Moreover, any solution should provide maximum control over the process of matching the best agents to service contacts that they are best suited to process.

[0012] The above and other problems of the prior art are overcome and a technical advance is achieved in accordance with the present invention which relates to processing contact center contacts in a manner which maps the best available agent to any particular contact. In accordance with the technique, each type of possible contact (service), corresponds to an N-dimensional service vector. Each dimension of the N-dimensional vector space represents a different skill required to process that particular type of contact or service. The magnitude of the service vector in any dimension represents the importance of the skill represented by the dimension to the processing of the contact.

[0013] Each agent is also assigned a similar vector, the agent vector representing in each dimension that particular agent's proficiency level in the skill represented by that dimension. The system then calculates a dot product between each agent vector and each service vector, the dot product being termed herein a skill score. Each agent is assigned a plurality of skill scores, one for each type of service that the agent can process. In a preferred embodiment, if an agent can not do at all any particular skill required for a service, then the agent may not be used for that service at all.

[0014] When a call arrives, the system immediately ascertains the type of service required. Thereafter, the call center assigns to process that contact or service the available agent with the highest skill score for that service.

[0015] The above technique insures that the agent assigned to the call is the available agent which has the best skills in the areas that are most important to process that call. An agent with high proficiency in a skill that is extremely important to a particular type of service will be preferably assigned to a particular contact over a different agent with slightly better proficiency in a skill which is much less important to the particular service. The system results in each contact being serviced by the agent whose profile (i.e., set of skills scores) shows that the agent is the best agent available to service that particular contact.

[0016] In various embodiments values assigned to each skill may vary, or the values of each dimension of the service vector and agent vector may be normalized to a common scale. Additionally, contacts may be of any type, including inbound calls, outbound, e mail, web chat, videoconferencing, or any other type of agent user interaction.

[0017] FIG. 1 is a basic flow chart of functional steps of processing a contact in accordance with the present invention;

[0018] FIG. 2 is a table depicting assignments of skills proficiency levels to services and to agents; and

[0019] FIG. 3 is a plurality of service queues for use with the invention.

[0020] FIG. 1 shows a basic flow diagram of the processing of a particular contact at a contact center constructed in accordance with the techniques of the present invention. The contact center accepts the exemplary call at 101 and parses the information available from the call in order to determine the service (contact type). The block 102 may utilize ANI, account identification data entered by a user, or any other information in order to ascertain the type of service required for processing the particular call. Once that is ascertained, the best available agent is located for processing that call at block 103. The call is then transferred for processing at block 104.

[0021] FIG. 3 shows a plurality of service queues, one for each of the different types of services offered by the contact center. In service queue 1, agent 15, agent 20 and agent 4 are shown as available. Each of the other service queues shows a set of three exemplary agents also available to process contacts of the particular type represented by the service in issue.

[0022] In operation, the simplest technique of allowing block 103 of FIG. 1 to choose the best available agent is to simply order the available agents in the queues by their skills score with respect to that service. For example, using such a technique, agent 15, agent 20 and agent 4 have the highest, next highest, and third highest skills score with respect to service 1, respectively, as shown in FIG. 3. By utilizing such an ordering technique, the selection of the best agent is simply a matter of traversing the list from top to bottom and picking the first idle available agent. Thus, when a contact is to be serviced, it is immediately apparent to whom the call should be assigned. Although simple and efficient, the invention is not limited to this technique and any of a variety of other techniques may be utilized for selection of the agent with the highest skills score for a particular service.

[0023] The creation of the table shown in FIG. 3 involves four basic steps. The first is the basic definition of the skills required, the second is the assignment of those skills to agents, the third is the definition of the skills, and the skill vector, for a particular service types, and the fourth is the calculation of N-dimensional dot products to produce the ordered table shown in FIG. 3. We address each of these separately below.

[0024] The skills set definition steps involves defining all of the different types of possible skills required for all of the different types of services a particular contact center offers. FIG. 2 shows a line 201 with three exemplary skills, Polish language, knowledge of replacement parts, and franchisee procedures. Each of the skills has a particular level, such as Polish where the skills are defined to be one (the worst) through ten (fluent speaking, reading and writing). The lines 201 and 202 of FIG. 2 indicate for each skill, the different proficiency levels within that skill. Thus, with respect to knowledge of franchise procedures to process callers interested in opening a franchise, there are only two skill levels, one and two.

[0025] The arrangement of lines 201 and 202 of FIG. 2 could be thought of conceptually as defining the N-dimensional vector space within which the process will take place. Each of the N-dimensions is represented by one of the skills.

[0026] Once the skills are defined, the agents are each given a proficiency level with respect to each skill. As shown by lines 207 and 208 of FIG. 2, exemplary agent 3 speaks Polish slightly better than average (6 out of 10), has minimal skills in replacement part knowledge and has top skills in franchisee procedures, scoring a 2 in this category. Lines 207 and 208 also show different skill levels for each particular agent with respect to each skill.

[0027] The third parameter which must be defined is the relative importance of each of the particular types of skills for processing each service. These values of the relative importance of each are shown in conceptual format at lines 205, 205aand 206 of FIG. 2. Each of the skills can be rated in lines 205 and 206 with a number in any range set in advance. Exemplary ranges are shown in the left-hand column of lines 205-206. For example, a skill may have five values, one through five, or may have six values one, five, six, seven, eight and twenty. In the latter case, the highest skill twenty is much better than any of the other skill levels, but skill levels five, six, seven and eight differ only slightly from each other. By assigning the different possible proficiency levels to different particular skills, the system is more flexible. Specifically, the system allows for a situation where a particular agent that has a certain minimum proficiency or better at a particular skill is deemed to be much better than agents of varying proficiency levels in that skill which are below a minimum threshold.

[0028] Finally, the agents skill profile is formed by calculating a vector (dot) product between each agent and each different service vector. This provides a measure of how closely the agents skills match the skills required to handle a particular contact.

[0029] In an enhanced embodiment, an exception is made to the foregoing procedure. Specifically if an agent has zero proficiency in a particular skill, he will not be assigned to process a contact requiring that skill at any level. This prevents an agent from being assigned a contact which requires a particular skill, even at a low level, but which the agent has no ability to process. Moreover, in an additional embodiment, a minimum proficiency level for each skill is defined. Thus, whether or not an agent vector and service vector produce a large dot product, if the agent vector does not have a particular specified minimum proficiency for each particular skill required by a service, the agent will not be assigned contacts that require that service, even if the relevant agent vector and service vector produce a relatively large dot product. These additional embodiments insure that when a contact comes in, the best available agent is assigned to that contact. However, that agent is chosen from a pool of agents each of which has, at the very least, a minimum set of proficiencies in all of the relevant skills required to handle that particular contact.

[0030] Skills may be updated and amended at various times during system operation. Whenever an agent is added, the set of skills possessed by that agent will be entered into the system, and the appropriate information updated. Skills may also be added and updated when different service types are added, or when it is discovered that particular existing services require different skills or skills of differing importance as the system operates.

[0031] The contact center would typically have a supervisory terminal that permits the supervisor to add and amend skills. Also, the magnitude within each dimension can be defined by the supervisor, both for the agent vector and the service vector. One preferred way of doing this is to provide a graphical user interface (GUI) with a slidable indicator. The supervisor can then, for each skill, slide the indicator along a scale to indicate the proficiency level of an agent for a skill, the endpoints of the scale being the highest and lowest levels of proficiency. With respect to the level of importance of a skill for a particular service, a similar system can be used. Alternatively, discreet numbers or levels may be entered by the supervisor.

[0032] As still a further enhancement, the proficiency levels may be named, with relative weights of each level being assigned. For example, a skill may have three values: poor, acceptable, and excellent. The system would allow the operator to enter one of these three, and could then translate the levels into values such as one, two and three, respectively, for purposes of forming the skill score. Additionally, the system can be reconfigured to translate poor, acceptable, and excellent into 1, 3, and 10, respectively. This will cause the agent assignment algorithm to treat acceptable as only slightly better than poor, but excellent as MUCH better than either poor or acceptable.

[0033] While the above describes the preferred embodiment of the invention, various modifications and additions will be apparent to those of skill-in the art. Such modifications are intended to be covered by the following claims.