[go: up one dir, main page]

WO2025191389A1 - Capteur de compensation d'agent - Google Patents

Capteur de compensation d'agent

Info

Publication number
WO2025191389A1
WO2025191389A1 PCT/IB2025/052216 IB2025052216W WO2025191389A1 WO 2025191389 A1 WO2025191389 A1 WO 2025191389A1 IB 2025052216 W IB2025052216 W IB 2025052216W WO 2025191389 A1 WO2025191389 A1 WO 2025191389A1
Authority
WO
WIPO (PCT)
Prior art keywords
agent
outlier
task
pairing strategy
time period
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Pending
Application number
PCT/IB2025/052216
Other languages
English (en)
Inventor
Chris PARISO
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Afiniti Ai Ltd
Original Assignee
Afiniti Ai Ltd
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Afiniti Ai Ltd filed Critical Afiniti Ai Ltd
Publication of WO2025191389A1 publication Critical patent/WO2025191389A1/fr
Pending legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Classifications

    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04M—TELEPHONIC COMMUNICATION
    • H04M3/00—Automatic or semi-automatic exchanges
    • H04M3/42—Systems providing special services or facilities to subscribers
    • H04M3/50—Centralised arrangements for answering calls; Centralised arrangements for recording messages for absent or busy subscribers ; Centralised arrangements for recording messages
    • H04M3/51—Centralised call answering arrangements requiring operator intervention, e.g. call or contact centers for telemarketing
    • H04M3/523—Centralised call answering arrangements requiring operator intervention, e.g. call or contact centers for telemarketing with call distribution or queueing
    • H04M3/5232—Call distribution algorithms
    • G—PHYSICS
    • G06—COMPUTING OR CALCULATING; COUNTING
    • G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
    • G06Q10/00—Administration; Management
    • G06Q10/06—Resources, workflows, human or project management; Enterprise or organisation planning; Enterprise or organisation modelling
    • G06Q10/063—Operations research, analysis or management
    • G06Q10/0631—Resource planning, allocation, distributing or scheduling for enterprises or organisations
    • G06Q10/06311—Scheduling, planning or task assignment for a person or group
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04M—TELEPHONIC COMMUNICATION
    • H04M2201/00—Electronic components, circuits, software, systems or apparatus used in telephone systems
    • H04M2201/18—Comparators
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04M—TELEPHONIC COMMUNICATION
    • H04M2203/00—Aspects of automatic or semi-automatic exchanges
    • H04M2203/40—Aspects of automatic or semi-automatic exchanges related to call centers
    • H04M2203/402—Agent or workforce management
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04M—TELEPHONIC COMMUNICATION
    • H04M2203/00—Aspects of automatic or semi-automatic exchanges
    • H04M2203/55—Aspects of automatic or semi-automatic exchanges related to network data storage and management
    • H04M2203/555—Statistics, e.g. about subscribers but not being call statistics
    • H04M2203/556—Statistical analysis and interpretation
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04M—TELEPHONIC COMMUNICATION
    • H04M3/00—Automatic or semi-automatic exchanges
    • H04M3/42—Systems providing special services or facilities to subscribers
    • H04M3/50—Centralised arrangements for answering calls; Centralised arrangements for recording messages for absent or busy subscribers ; Centralised arrangements for recording messages
    • H04M3/51—Centralised call answering arrangements requiring operator intervention, e.g. call or contact centers for telemarketing
    • H04M3/5175—Call or contact centers supervision arrangements

Definitions

  • the present disclosure relates to controlling operation of a task assignment system. Particularly, but not exclusively, the present disclosure relates to identifying potential bias within a task assignment system, for example in a contact assignment center, and adjusting operation of the task assignment system to control bias; more particularly, but not exclusively, the present disclosure relates to pairing strategy induced bias in relation to agent outcomes.
  • a task assignment system may assign tasks arriving at a task assignment center to agents available to handle those tasks according to a pairing strategy.
  • the pairing strategy algorithmically determines which of the available agents an incoming task is assigned to.
  • pairing strategies may select an available agent to assign to a task based on, for example, an order in which the agents became available. These pairing strategies may be referred to as “first-in, first-out”, “FIFO”, or “round-robin” strategies. Absent other constraints, tasks of different types within a task queue are distributed to agents assigned to the queue approximately uniformly. Such FIFO or other uniform pairing strategies result in overall task assignment center performance that is mathematically equivalent to selecting agents at random to assign to each task. Optimized pairing strategies seek to pair tasks and agents more intelligently by targeting factors such as balanced utilization of agents whilst simultaneously improving overall task assignment system performance.
  • optimized pairing strategies improve overall task assignment system performance beyond what FIFO pairing strategies provide, optimized pairing strategies may nevertheless introduce non-uniform distributions (i.e., “biased” distributions or simply “bias”) into the operation of the task assignment system.
  • individual agents may be inadvertently penalized with respect to a change in task assignment distribution due to the pairing decisions involving the individual agents made by an optimized pairing strategy.
  • One type of a task assignment center may include a contact assignment center (e.g., a call center) where agents are assigned to contacts (e.g., callers). Therefore, there is a need to improve task assignment systems to account for potential pairing strategy induced bias.
  • a method for bias mitigation in a task assignment system the method performed by at least one processor configured to operate in the task assignment system, the method comprising: identifying, within the task assignment system, a plurality of agents which are paired with tasks according to a first pairing strategy, determining a representative attribute value of an attribute of task-agent pairings involving the plurality of agents over a first time period, identifying an outlier agent within the plurality of agents based on the representative attribute value, the outlier agent having an outlier attribute value indicative of a bias related to the first pairing strategy, and adjusting operation of the task assignment system such that the outlier agent is transitioned from being paired with tasks according to the first pairing strategy to being paired with tasks according to a second pairing strategy.
  • a non-transitory computer readable medium comprising instructions which, when executed by one or more processors configured to operate in a task assignment system, cause the one or more processors to carry out the method of identifying, within the task assignment system, a plurality of agents which are paired with tasks according to a first pairing strategy, determining a representative attribute value of an attribute of task-agent pairings involving the plurality of agents over a first time period, identifying an outlier agent within the plurality of agents based on the representative attribute value, the outlier agent having an outlier attribute value indicative of a bias related to the first pairing strategy, and adjusting operation of the task assignment system such that the outlier agent is transitioned from being paired with tasks according to the first pairing strategy to being paired with tasks according to a second pairing strategy.
  • a device comprising one or more processors configured to operate in a task assignment system and a memory storing instructions which, when executed by the one or more processors cause the device to carry out the method of identifying, within the task assignment system, a plurality of agents which are paired with tasks according to a first pairing strategy, determining a representative attribute value of an attribute of task-agent pairings involving the plurality of agents over a first time period, identifying an outlier agent within the plurality of agents based on the representative attribute value, the outlier agent having an outlier attribute value indicative of a bias related to the first pairing strategy, and adjusting operation of the task assignment system such that the outlier agent is transitioned from being paired with tasks according to the first pairing strategy to being paired with tasks according to a second pairing strategy.
  • a method for bias mitigation in a task assignment system comprising: identifying, within the task assignment system, a plurality of agents paired with tasks by the task assignment system according to a first pairing strategy, determining a plurality of performance values related to a plurality of task-agent pairings involving the plurality of agents over a first time period, determining, based on the plurality of performance values, an outlier agent within the plurality of agents, wherein the outlier agent has an outlier performance value which differs from a representative performance value for the plurality of agents by at least a first threshold amount, transitioning the outlier agent from being paired with tasks according to the first pairing strategy to being paired with tasks according to a second pairing strategy such that a subsequent performance value obtained for the outlier agent when paired according to the second pairing strategy differs from the representative performance value for the plurality of agents by less than the first threshold amount.
  • a non-transitory computer readable medium comprising instructions which, when executed by one or more processors configured to operate in a task assignment system, cause the one or more processors to carry out the method of identifying, within the task assignment system, a plurality of agents paired with tasks by the task assignment system according to a first pairing strategy, determining a plurality of performance values related to a plurality of task-agent pairings involving the plurality of agents over a first time period, determining, based on the plurality of performance values, an outlier agent within the plurality of agents, wherein the outlier agent has an outlier performance value which differs from a representative performance value for the plurality of agents by at least a first threshold amount, transitioning the outlier agent from being paired with tasks according to the first pairing strategy to being paired with tasks according to a second pairing strategy such that a subsequent performance value obtained for the outlier agent when paired according to the second pairing strategy differs from the representative performance value for the plurality of agents by
  • a device comprising one or more processors configured to operate in a task assignment system and a memory storing instructions which, when executed by the one or more processors cause the device to carry out the method of identifying, within the task assignment system, a plurality of agents paired with tasks by the task assignment system according to a first pairing strategy, determining a plurality of performance values related to a plurality of task-agent pairings involving the plurality of agents over a first time period, determining, based on the plurality of performance values, an outlier agent within the plurality of agents, wherein the outlier agent has an outlier performance value which differs from a representative performance value for the plurality of agents by at least a first threshold amount, transitioning the outlier agent from being paired with tasks according to the first pairing strategy to being paired with tasks according to a second pairing strategy such that a subsequent performance value obtained for the outlier agent when paired according to the second pairing strategy differs from the representative performance value for the plurality of agents by less than
  • a method for bias mitigation in a task assignment system comprising: collecting outcome data for a plurality of agents which are paired to tasks by a task assignment system according to a first pairing strategy, the outcome data comprising a plurality of outcome values quantifying outcomes of task-agent pairings involving the plurality of agents over a first time period, generating a distributional representation of the outcome data, identifying, from the distributional representation of the outcome data, one or more outlier agents within the plurality of agents, each of the one or more outlier agents having outlier outcome values in relation to an average outcome value of the distributional representation of the outcome data, and causing the task assignment system to transition the one or more outlier agents from the first pairing strategy to a second pairing strategy over a second time period.
  • a non-transitory computer readable medium comprising instructions which, when executed by one or more processors configured to operate in a task assignment system, cause the one or more processors to carry out the method of collecting outcome data for a plurality of agents which are paired to tasks by a task assignment system according to a first pairing strategy, the outcome data comprising a plurality of outcome values quantifying outcomes of taskagent pairings involving the plurality of agents over a first time period, generating a distributional representation of the outcome data, identifying, from the distributional representation of the outcome data, one or more outlier agents within the plurality of agents, each of the one or more outlier agents having outlier outcome values in relation to an average outcome value of the distributional representation of the outcome data, and causing the task assignment system to transition the one or more outlier agents from the first pairing strategy to a second pairing strategy over a second time period.
  • a system for agent outcome optimization within a task assignment system comprising a plurality of agents which are paired to tasks by the task assignment system according to a first pairing strategy
  • the system comprising: a monitoring module configured to: collect outcome data for the plurality of agents, the outcome data comprising a plurality of outcome values quantifying outcomes of task-agent pairings involving the plurality of agents over a first time period; and a control module communicatively coupled to the monitoring module and configured to: generate a distributional representation of the outcome data, identify, from the distributional representation of the outcome data, one or more outlier agents within the plurality of agents, each of the one or more outlier agents having outlier outcome values in relation to an average outcome value of the distributional representation of the outcome data, and cause the task assignment system to transition the one or more outlier agents from the first pairing strategy to a second pairing strategy over a second time period.
  • the above described aspects of the present disclosure provide improvements to task assignment systems by helping to optimize the operation of a task assignment system.
  • Potential biases distortions or skews in performance or outcome — induced on an agent as a result of them being paired to contacts by a particular pairing strategy implemented by the task assignment system are automatically identified and remediated thereby providing an improved task assignment system and improved agent performance/outcomes.
  • pairing strategies used in contact centers have a net positive contribution in terms of performance when performance is considered part, or the entire pool, of agents, individual agents may have a much larger decrease (or even increase) in performance than the part, or entire pool of agents.
  • the invention provides the means to identify individual agents whose performance is affected by the pairing strategy much more than their peers and thus are biased by the pairing strategy.
  • the invention allows for remedial action to be taken for the identified agents.
  • the potential bias may increase load on the task assignment system due to, e.g., non-uniform increases or decreases to an agent’s average handle time, average performance, etc. That is, optimized pairing strategies optimized for one or more metrics may induce potential biases in one or more nonoptimized metrics, which may inhibit optimal operation and functioning of a task assignment system. Identifying such potential biases and adjusting operation of the task assignment system to mitigate such potential biases therefore helps improve the overall operation, functioning, and optimization of a task assignment system.
  • FIGS 1A and IB show block diagrams of contact assignment centers according to embodiments of the present disclosure
  • Figures 2A-2C show block diagrams of a contact assignment center system according to embodiments of the present disclosure
  • Figure 3 illustrates a block diagram of a task assignment system according to an aspect of the present disclosure
  • Figure 4 shows a method for bias mitigation in a task assignment system according to an aspect of the present disclosure
  • Figure 5 shows a method for bias mitigation in a task assignment system according to an aspect of the present disclosure
  • Figure 6 shows a method for agent outcome optimization within a task assignment system according to an aspect of the present disclosure
  • Figure 7 shows a plot of a distributional representation of attribute data according to embodiments of the present disclosure
  • Figure 8 illustrates an example of bias mitigation performed by the systems and methods of the present disclosure.
  • Figure 9 shows an example computing system for bias identification and handling.
  • a typical task assignment system algorithmically assigns tasks arriving at a task assignment center to agents available to handle those tasks.
  • the task assignment center may be in an “LI state” and have agents available and waiting for assignment to tasks.
  • the task assignment center may be in an “L2 state” and have tasks waiting in one or more queues for an agent to become available for assignment.
  • the task assignment system may be in an “L3” state and have multiple agents available and multiple tasks waiting for assignment.
  • An example of a task assignment system is a contact center system that receives contacts (e.g., telephone calls, internet chat sessions, emails, etc.) to be assigned to agents.
  • tasks e.g., contacts, callers, etc.
  • agents receive tasks ordered based on the time when those agents became available.
  • This strategy may be referred to as a “first-in, first-out,” “FIFO,” or “round-robin” strategy.
  • FIFO first-in, first-out
  • round-robin round-robin
  • a performance-based routing (PBR) strategy for prioritizing higher-performing agents for task assignment may be implemented. Under PBR, for example, the highest-performing agent among available agents receives the next available task.
  • BP Behavioral Pairing
  • BP Behavioral Pairing
  • BP targets balanced utilization of agents while simultaneously improving overall task assignment center performance potentially beyond what FIFO or PBR methods will achieve in practice. This is a remarkable achievement inasmuch as BP acts on the same tasks and same agents as FIFO or PBR methods, approximately balancing the utilization of agents as FIFO provides, while improving overall task assignment center performance beyond what either FIFO or PBR provide in practice.
  • BP improves performance by assigning agent and task pairs in a fashion that takes into consideration the assignment of potential subsequent agent and task pairs such that, when the benefits of all assignments are aggregated, they may exceed those of FIFO and PBR strategies.
  • BP strategies may be used, such as a diagonal model BP strategy or a network flow (or “off- diagonal”) BP strategy.
  • These task assignment strategies and others are described in detail for a contact center context in, e.g., U.S. Pat. Nos. 9,300,802; 9,781,269; 9,787,841; and 9,930,180; all of which are hereby incorporated by reference herein.
  • BP strategies may be applied in an LI environment (agent surplus, one task; select among multiple available/idle agents), an L2 environment (task surplus, one available/idle agent; select among multiple tasks in queue), and an L3 environment (multiple agents and multiple tasks; select among pairing permutations).
  • the various BP strategies discussed above may be considered two-dimensional (2-D), where one dimension relates to the agents, and the second dimension relates to the tasks (e.g., contacts, callers, etc.), and the various BP strategies take into account information about agents and tasks to pair them.
  • embodiments of the present disclosure relate to multidimensional BP strategies that account for higher-dimensional assignments.
  • the BP strategy may assign an agent to both a task and a set of actions the agent can take or a set of offers the agent can make during the task assignment.
  • These multidimensional BP strategies may also consider historical outcome data for, e.g., agent-task- actions or agent-task-offers pairings to build a BP model and apply a BP strategy to “pair” a task with an agent and a specific action and/or specific offer set (throughout the specification, the noun and verb “pair” and other forms such as “Behavioral Pairing” may be used to describe two-dimensional pairs and/or triads and other higher-dimensional groupings).
  • the various BP strategies cause the task assignment system to change operations resulting in the task assignment system behaving in a new way.
  • assigning a task to an agent according to a pairing strategy results in technical changes to the operation of the task assignment system (e.g., instructions are sent to one or more switches within the task assignment system to connect the task to the agent) which causes the task assignment system to function in a new improved way.
  • pairing strategies which may be utilized by the task assignment system include a prioritized call routing strategy where priority rules are employed to preferentially route urgent or high-priority tasks, an optimized FIFO strategy, and a random agent selection pairing strategy.
  • the above described pairing strategies may be classified as optimized (behavioral pairing, performance based routing, prioritized call routing, and optimized FIFO) or FIFO (round robin or random agent selection).
  • Figure 1A shows a block diagram of a task assignment center 100A according to embodiments of the present disclosure.
  • module may be understood to refer to computing software, firmware, hardware, and/or various combinations thereof. Modules, however, are not to be interpreted as software which is not implemented on hardware, firmware, or recorded on a non-transitory processor readable recordable storage medium (i.e., modules are not software per se). It is noted that the modules are exemplary. The modules may be combined, integrated, separated, and/or duplicated to support various applications.
  • a function described herein as being performed at a particular module may be performed at one or more other modules and/or by one or more other devices instead of or in addition to the function performed at the particular module.
  • the modules may be implemented across multiple devices and/or other components local or remote to one another. Additionally, the modules may be moved from one device and added to another device, and/or may be included in both devices.
  • the task assignment center 100A may include a central switch 110.
  • the central switch 110 may receive incoming tasks (e.g., telephone calls, internet chat sessions, emails, etc.) or support outbound connections to tasks via a dialer, a telecommunications network, or other modules (not shown).
  • the central switch 110 may include routing hardware and software for helping to route tasks among one or more subcenters, or to one or more Private Branch Exchange (“PBX”) or Automatic Call Distribution (ACD) routing components or other queuing or switching components within the task assignment center 100A.
  • PBX Private Branch Exchange
  • ACD Automatic Call Distribution
  • the central switch 110 may not be necessary if there is only one subcenter, or if there is only one PBX or ACD routing component in the task assignment center 100.
  • each subcenter may include at least one switch (e.g., switches 120A and 120B).
  • the switches 120A and 120B may be communicatively coupled to the central switch 110.
  • Each switch for each subcenter may be communicatively coupled to a plurality (or “pool”) of agents.
  • Each switch may support a certain number of agents (or “seats”) to be logged in at one time.
  • a logged-in agent may be available and waiting to be connected to a task, or the logged-in agent may be unavailable for any of a number of reasons, such as being connected to another task, performing certain post-call functions such as logging information about the call, or taking a break.
  • the central switch 110 routes tasks to one of two subcenters via switch 120A and switch 120B, respectively.
  • Each of the switches 120A and 120B are shown with two agents each.
  • Agents 130A and 130B may be logged into switch 120A, and agents 130C and 130D may be logged into switch 120B.
  • the task assignment center 100A may also be communicatively coupled to an integrated service from, for example, a third-party vendor.
  • behavioral pairing module 140 may be communicatively coupled to one or more switches in the switch system of the task assignment center 100A, such as central switch 110, switch 120A, and switch 120B.
  • switches of the task assignment center 100A may be communicatively coupled to multiple behavioral pairing modules.
  • behavioral pairing module 140 may be embedded within a component of the task assignment center 100A (e.g., embedded in or otherwise integrated with a switch).
  • Behavioral pairing module 140 may receive information from a switch (e.g., switch 120A) about agents logged into the switch (e.g., agents 130A and 130B) and about incoming tasks via another switch (e.g., central switch 110) or, in some embodiments, from a network (e.g., the Internet or a telecommunications network) (not shown). The behavioral pairing module 140 may process this information to determine which agents should be paired (e.g., matched, assigned, distributed, routed) with which tasks, and, in some examples, with which tasks along with other dimensions (e.g., offers, actions, channels, non-monetary rewards, monetary rewards or compensation, physical resources, proxies for physical resources, etc.).
  • a switch e.g., switch 120A
  • agents logged into the switch e.g., agents 130A and 130B
  • incoming tasks via another switch (e.g., central switch 110) or, in some embodiments, from a network (e.g., the Internet or a t
  • the behavioral pairing module 140 may send one or more instructions to the relevant switch(es) to connect the agent to the task thereby resulting in a change of state in the task assignment center 100A. Therefore, the pairing module 140 causes a direct change in how the task assignment center 100A operates.
  • a switch will typically automatically distribute the new task to whichever available agent has been waiting the longest amount of time for an agent under a FIFO strategy, or whichever available agent has been determined to be the highest-performing agent under a PBR strategy.
  • tasks and agents may be given scores (e.g., percentiles or percentile ranges/bandwidths) according to a pairing model or other artificial intelligence data model, so that a task may be matched, paired, or otherwise connected to a preferred agent.
  • scores e.g., percentiles or percentile ranges/bandwidths
  • a switch In an L2 state, multiple tasks are available and waiting for connection to an agent, and an agent becomes available. These tasks may be queued in a switch such as a PBX or ACD device. Without the behavioral pairing module 140, a switch will typically connect the newly available agent to whichever task has been waiting on hold in the queue for the longest amount of time as in a FIFO strategy or a PBR strategy when agent choice is not available. In some task assignment centers, priority queuing may also be incorporated, as previously explained.
  • a behavioral pairing module 140 in this L2 scenario tasks and agents may be given percentiles (or percentile ranges/bandwidths, etc.) according to, for example, a model, such as an artificial intelligence model, so that an agent becoming available may be matched, paired, or otherwise connected to a preferred task.
  • a model such as an artificial intelligence model
  • Figure IB shows a block diagram of a task assignment system 100B according to embodiments of the present disclosure.
  • the task assignment system 100B may be included in a task assignment center (e.g., task assignment center 100A) or incorporated in a component or module (e.g., behavioral pairing module 140) of a task assignment center for helping to assign agents among various tasks and other dimensions for grouping.
  • a task assignment center e.g., task assignment center 100A
  • a component or module e.g., behavioral pairing module 140
  • the task assignment system 100B may include a task assignment module 150 that is configured to pair (e.g., match, assign) incoming tasks to available agents.
  • a task assignment module 150 configured to pair (e.g., match, assign) incoming tasks to available agents.
  • m tasks 160A-160m are received over a given period, and n agents 170A-170n are available during the given period.
  • Each of the m tasks may be assigned to one of the n agents for servicing or other types of task processing.
  • m and n may be arbitrarily large finite integers greater than or equal to one.
  • a real-world task assignment center such as a task center, there may be dozens, hundreds, etc. of agents logged into the task center to interact with tasks during a shift, and the task center may receive dozens, hundreds, thousands, etc. of tasks (e.g., telephone calls, internet chat sessions, emails, etc.) during the shift.
  • a task assignment strategy module 180 may be communicatively coupled to and/or configured to operate in the task assignment system 200.
  • the task assignment strategy module 180 may implement one or more task assignment strategies (or “pairing strategies”) for assigning individual tasks to individual agents (e.g., pairing tasks with task center agents).
  • a variety of different task assignment strategies may be devised and implemented by the task assignment strategy module 180.
  • a FIFO strategy may be implemented in which, for example, the longest-waiting agent receives the next available task (in LI environments) or the longest-waiting task is assigned to the next available agent (in L2 environments).
  • a PBR strategy for prioritizing higher- performing agents for task assignment may be implemented.
  • a BP strategy may be used for optimally assigning tasks to agents using information about either tasks or agents, or both.
  • BP strategies may be used, such as a diagonal model BP strategy or a network flow (“off- diagonal”) BP strategy. See U.S. Pat. Nos. 9,300,802; 9,781,269; 9,787,841; and 9,930,180.
  • the task assignment strategy module 180 may implement a multidimensional BP strategy that takes into account the next-best action for a task, when the task is assigned to a particular agent.
  • the multidimensional BP strategy may also assign an action or set of actions available to the agent to complete the task.
  • the action or set of actions may include an offer or a set of offers that the agent may present to a customer.
  • a multidimensional BP strategy may pair a task with an agent along with an action or set of agents (e.g., an offer or set of offers) available to the agent to make to a customer, based on the expected outcome of the task-agent interaction using that particular action or set of actions.
  • agents e.g., an offer or set of offers
  • a multidimensional BP strategy goes beyond pairing a task to an agent by optimizing the outcome of the individual interaction between the agent and the task.
  • agent 170A may be more adept at selling sports packages. Therefore, sports packages may be included in agent 170A’s set of offers for some or all task types.
  • agent 170B may love movies and may be more adept at selling premium movie packages; so premium movie packages may be included in agent 170B’s set of offers for some or all task types.
  • a multidimensional BP model may automatically segment customers over a variety of variables and data types. For example, the multidimensional BP model may recommend offering a package that includes sports to a first type of customer (“Customer Type 1”) that may fit tasks of that particular type.
  • the multidimensional BP model may recommend offering a premium movie package to a second type of customer (“Customer Type 2”) that may fit a task type.
  • a multidimensional BP strategy may preferably pair a Customer Type 1 with agent 170A and an offer set with a sports package, and a Customer Type 2 with agent 170B and an offer set with a premium movie package. This results in a change in state of the task assignment system 100B by causing connections to be formed between the tasks and the agents based on the pairing determined by the multidimensional BP strategy. Thus, the changes in the pairings ultimate cause the task assignment system 100B to operate in a different manner.
  • a multidimensional BP strategy optimizes the overall performance of the task assignment system rather than every individual instant task-agent pairing. For instance, in some embodiments, a multidimensional BP system will not always offer sports to a Customer Type 1, nor will agent 170A always be given the option of offering deals based on a sports package. Such a scenario may arise when a marketing division of a company running a task center system may have a budget for a finite, limited number of deals (e.g., a limited number of discounted sports packages), other constraints on the frequency of certain offers, limits on the total amount of discounts (e.g., for any discount or discounted package) that can be made over a given time period, etc. Similarly, deals based on sports may sometimes be offered to a Customer Type 2, and agent 170B may sometimes be given the option of offering deals based on a sports package.
  • a marketing division of a company running a task center system may have a budget for a finite, limited number of deals (e.g., a limited number of discounted sports packages
  • a multidimensional BP strategy may account for all types of customers waiting in a queue, agents available for customers, and any other dimensions for pairing such as the number and types of offers remaining, agent compensation or other nonmonetary rewards, next-best actions, etc.
  • a probability distribution may be assigned based on the likelihood that an incoming task or customer type will accept a given offer level based on the agent being paired with the task or customer.
  • the likelihood of a task of Task Type 1 accepting the offer from an average agent is 0%, and the likelihood of accepting the offer specifically from agent 170A is also 0% and from agent 170B is also 0%.
  • the likelihood of the task of Task Type 1 accepting the offer from an average agent may be 30%, whereas the likelihood of said task of Task Type 1 accepting the offer from agent 170A may be 60% and from agent 170B may be 25%.
  • agent 170A, and agent 170B are all assigned to the queue and available, it is possible for agent 170A to perform much higher than the average agent or agent 170B.
  • an output measurement may be attached to each task before and after interaction with an agent.
  • revenue number may be attached to each caller pre- and post-call.
  • a multidimensional BP system may measure the change in revenue and the influenceability of a task based on an offer or a set of offers presented by an agent. For example, a Task Type 1 may be more likely to renew her existing plan regardless of the discount offered, or regardless of the ability of the individual agent.
  • a Task Type 2 may be preferably assigned to a lower-performing agent with a higher cap on discounts in the offer set. In contrast, a Task Type 2 may be more likely to upgrade her plans if she were paired with a higher-performing agent or an agent authorized to offer steeper discounts.
  • a multidimensional BP strategy may make sequential pairings of one or more dimensions in an arbitrary order. For example, the multidimensional BP strategy may first pair an agent to a task and then pair an offer set to the agent-task pairing, then pair a reward to the agent-task-offer set pairing, and so on.
  • a multidimensional BP strategy may make “fully-coupled,” simultaneous multidimensional pairings.
  • the multidimensional BP strategy may consider all dimensions at once to select an optimal 4-D agent-task-offers-reward pairing.
  • the same task may arrive at the task assignment system multiple times (e.g., the same caller calls a call center multiple times).
  • the task assignment system may always assign the same item for one or more dimensions to promote consistency. For example, if a task is paired with a particular action or action set the first time the task arrives, the task will be paired with the same action or action set each subsequent time the task arrives (e.g., for a given issue, within a given time period, etc.).
  • a historical assignment module 190 may be communicatively coupled to and/or configured to operate in the task assignment system 100B via other modules such as the task assignment module 150 and/or the task assignment strategy module 180.
  • the historical assignment module 190 may be responsible for various functions such as monitoring, storing, retrieving, and/or outputting information about task-agent assignments and higher-dimensional assignments that have already been made. For example, the historical assignment module 190 may monitor the task assignment module 150 to collect information about task assignments in a given period.
  • Each record of a historical task assignment may include information such as an agent identifier, a task or task type identifier, action or action set identifier (e.g., offer or offer set identifier), outcome information, or a pairing model identifier (i.e., an identifier indicating whether a task assignment was made using a BP strategy, a multidimensional BP strategy, or some other pairing model such as a FIFO or PBR pairing model).
  • a pairing model identifier i.e., an identifier indicating whether a task assignment was made using a BP strategy, a multidimensional BP strategy, or some other pairing model such as a FIFO or PBR pairing model.
  • additional information may be stored.
  • the historical assignment module 190 may also store information about the time a call started, the time a call ended, the phone number dialed, and the caller’s phone number.
  • the historical assignment module 190 may generate a pairing model, a multidimensional BP model, or similar computer processor-generated model based on a set of historical assignments for a period of time (e.g., the past week, the past month, the past year, etc.), which may be used by the task assignment strategy module 180 to make task assignment recommendations or instructions to the task assignment module 150.
  • a period of time e.g., the past week, the past month, the past year, etc.
  • the historical assignment module 190 may analyze historical outcome data to create or determine new or different offer sets, which are then incorporated into the multidimensional BP model. This approach may be preferred when there are limitations on the number of a particular action set that may be made. For example, the marketing division may have limited the task center system to five hundred discounted sports packages and five hundred discounted movie packages per month, and the company may want to optimize total revenue irrespective of how many sports and movie packages are sold, with or without a discount.
  • the multidimensional BP model may be similar to previously-disclosed BP diagonal models, except that, in addition to the “task or task percentile” (CP) dimension and the “agent percentile” (AP) dimension, there may be a third “revenue or offer set percentile” dimension.
  • all three dimensions may be normalized or processed with mean regression (e.g., Bayesian mean regression (BMR) or hierarchical BMR).
  • mean regression e.g., Bayesian mean regression (BMR) or hierarchical BMR).
  • the historical assignment module 190 may generate a multidimensional BP model that optimizes task-agent-offer set pairing based on individual channels or multi-channel interactions.
  • the historical assignment module 190 may treat different channels differently.
  • a multidimensional BP model may preferably pair a task with different agents or action sets depending on whether the task calls a call center, initiates a chat session, sends an email or text message, enters a retail store, etc.
  • the task assignment strategy module 180 may proactively create tasks or other actions (e.g., recommend outbound task interactions, next-best actions, etc.) based on information about a task or a customer, available agents, and available offer sets. For example, the task assignment system 100B may determine that a customer’s contract is set to expire, the customer’ s usage is declining, or the like. The task assignment system 100B may further determine that the customer is unlikely to renew the contract at the customer’s current rate (e.g., based on information from the historical assignment module 190).
  • tasks or other actions e.g., recommend outbound task interactions, next-best actions, etc.
  • the task assignment system 100B may determine that a customer’s contract is set to expire, the customer’ s usage is declining, or the like.
  • the task assignment system 100B may further determine that the customer is unlikely to renew the contract at the customer’s current rate (e.g., based on information from the historical assignment module 190).
  • the task assignment system 100B may determine that the next-best action is to call the customer (task selection, channel selection, and timing selection), connect with a particular agent (agent selection), and give the agent the option to offer a downgrade at a particular discount or range of discounts (offer set selection). If the customer does not come to an agreement during the call, the task assignment system 100B may further determine that this customer is more likely to accept a downgrade discount offer if the agent follows up with a text message with information about the discount and how to confirm (multi-channel selection and optimization).
  • the task assignment strategy module 180 may apply an Iota (t) parameter to a third (or higher) dimension such as the action or action set percentile or percentile range in a multidimensional BP strategy.
  • the task assignment strategy module 180 may, for example, adjust the action or action set percentile or percentile range (or other dimensions) to skew task-agent-action pairing toward higher-performing actions and imbalance action set availability.
  • the Iota parameter may be applied in either LI or L2 environment and may be used in conjunction with Kappa or Rho parameter, or it may be applied with both Kappa and Rho parameters in an L3 environment. For example, if the task assignment strategy module 180 determines that the expected wait time for a task has exceeded 100 seconds (high congestion), it may apply the Iota parameter so that an agent is more likely to have more relevant actions available to offer, which are likely to be accepted or taken more quickly to reduce congestion and the expected wait time.
  • AHT average handle time
  • Iota parameter may be adjusted to make only less generous offers available.
  • the task assignment strategy module 180 may optimize performance by optimizing for multiple objectives or other goals simultaneously. Where the objectives are competing (e.g., discount amount and retention rates), the task assignment strategy module 180 may balance the tradeoff between the two objectives. For example, the task assignment strategy module 180 may balance increasing (or maintaining) revenue with maintaining or minimally decreasing retention rates, or it may balance decreasing (or maintaining) AHT with increasing (or maintaining) customer satisfaction, etc.
  • the task assignment strategy module 180 may implement a multidimensional BP strategy that takes into account agent compensation in lieu of an offer or offer set.
  • the framework is similar to the description above, except that, instead of influencing a customer with an offer or offer set, the multidimensional BP strategy influences the performance of an agent with a compensation that the agent may receive.
  • the multidimensional BP strategy instead of the task-agent-offer set three-dimensional pairing, the multidimensional BP strategy makes a three-dimensional pairing of task-agent-reward.
  • a multidimensional BP strategy may make a four-way pairing of task-agent-offer- reward.
  • Such a multidimensional BP strategy being capable of providing variable agent compensation based on task-agent pairing may lead to better transparency and fairness.
  • some task assignment (e.g., contact center) systems may see a mix of more challenging and less challenging tasks and employ a mix of higher-performing and lower-performing agents.
  • agents of any ability may be equally likely to be paired with more or less challenging tasks.
  • the overall performance of the task center system may be low, but the average agent compensation may be transparent and fair.
  • agent utilization may be skewed, and compensation may also be skewed toward higher-performing agents.
  • a more challenging task type may be preferably paired with a higher-performing agent, whereas a less challenging task type may be preferably paired with a lower-performing agent.
  • this “call type skew” may result in the high- performing agent’ s conversion rate going down and compensation going down.
  • Other factors which may skew an agent’ s compensation include the type of task assigned to the agent, the value of the task, the time spent by the agent in handling the task, and the like.
  • the multidimensional BP strategy determines that the agent should put high effort into a higher value call (higher AHT, higher revenue), it may select a higher compensation.
  • Such a multidimensional BP strategy may maximize an agent’s reward while improving the overall performance of the task assignment system 100B.
  • the historical assignment module 190 may model a multidimensional BP model based on historical information about task types, agents, and compensation amounts so that a simultaneous selection of task- agent-reward may be made.
  • the amount of variation in compensation up or down may vary and depend on each combination of an individual agent and task type, with the goal of improving the overall performance of the task assignment system 100B.
  • the task assignment strategy module 180 may apply an Iota parameter to other dimensions, such as skewing agent compensation to a greater or lesser degree, or to generally higher or generally lower values.
  • the amount and type of Iota parameter applied to agent compensation or other non-monetary rewards may be based at least in part on factors in the task assignment system 100B (e.g., the expected wait time of callers on hold in a call center).
  • a variable compensation may be viewed as temporarily influencing the effective agent percentile (AP) of an available agent to be higher or lower, in order to move an available task-agent pairing closer to the optimal diagonal.
  • AP effective agent percentile
  • TP effective task percentile
  • a benchmarking module 195 may be communicatively coupled to and/or configured to operate in the task assignment system 100B via other modules such as the task assignment module 150 and/or the historical assignment module 190.
  • the benchmarking module 195 may benchmark the relative performance of two or more pairing strategies (e.g., FIFO, PBR, BP, decisioning BP, etc.) using historical assignment information, which may be received from, for example, the historical assignment module 190.
  • the benchmarking module 195 may perform other functions, such as establishing a benchmarking schedule for cycling among various pairing strategies, tracking cohorts (e.g., base and measurement groups of historical assignments), etc. Benchmarking is described in detail for the task center context in, e.g., U.S. Pat. No. 9,712,676, which is hereby incorporated by reference herein.
  • the benchmarking module 195 may output or otherwise report or use the relative performance measurements.
  • the relative performance measurements may be used to assess the quality of the task assignment strategy to determine, for example, whether a different task assignment strategy (or a different pairing model) should be used, or to measure the overall performance (or performance gain) that was achieved within the task assignment system 100B while it was optimized or otherwise configured to use one task assignment strategy instead of another.
  • the benchmarking module 195 may benchmark a multidimensional BP strategy against one or more alternative pairing strategies such as FIFO in conjunction with offer set availability.
  • agents may have a matrix of nine offers-three tiers of service levels, each with three discount levels.
  • the longest-waiting agent may be connected to the longest- waiting task, and the agent may offer any of the nine offers.
  • a high-performing agent may be more likely to sell a higher tier of service at a higher price, whereas a lower-performing agent may not try as hard and go immediately to offering the biggest discounts.
  • the multidimensional BP strategy may pair tasks with agents but limit agents to a subset of the nine available offers.
  • a higher-performing agent may be empowered to make any of the nine offers, whereas a lower- performing agent may be limited to offer only the smaller discount for certain tiers, if the task assignment strategy module 180 determines, based on the multidimensional BP model, that the overall performance of the task center system may be optimized by selectively limiting the offer sets in a given way for a given task-agent pairing. Additionally, if a provider (e.g., vendor) that provides a task assignment system with multidimensional BP strategy uses a benchmarking and revenue sharing business model, the provider may contribute a share of benchmarked revenue gain to the agent compensation pool.
  • a provider e.g., vendor
  • the task assignment system 100B may offer dashboards, visualizations, or other analytics and interfaces to improve overall performance of the system.
  • the analytics provided may vary depending on the relative ability or behavioral characteristics of an agent. For example, competitive or higher-performing agents may benefit from a rankings widget or other “gamification” elements (e.g., badges or achievements to unlock points and score boards, notifications when agents overtake one another in the rankings, etc.). On the other hand, less competitive or lower-performing agents may benefit from periodic messages of encouragement, recommendations on training/education sessions, etc.
  • Figure 2A illustrates a block diagram of an example task assignment system 200A according to embodiments of the present disclosure.
  • the task assignment system 200A may include one or more agent endpoints 211 A, 21 IB and one or more task endpoints 212A, 212B.
  • the agent endpoints 211 A, 21 IB may include an agent terminal and/or an agent computing device (e.g., laptop, cellphone).
  • the task endpoints 212A, 212B may include a task terminal and/or a task computing device (e.g., laptop, cellphone).
  • Agent endpoints 211 A, 21 IB and/or task endpoints 212A, 212B may connect to a Task Center as a Service (TCaaS) 230 through either the Internet or a public switched telephone network (PSTN), according to the capabilities of the endpoint device.
  • TCCaaS 230 is a Contact Center as a Service (CCCaaS).
  • Figure 2B illustrates an example task assignment system 200B with an example configuration of a TCaaS 230.
  • a TCaaS 230 may include multiple data centers 240A, 240B.
  • the data centers 240A, 240B may be separated physically, even in different countries and/or continents.
  • the data centers 240A, 240B may communicate with each other.
  • one data center is a backup for the other data center; so that, in some embodiments, only one data center 240A or 240B receives agent endpoints 211 A, 211B and task endpoints 212A, 212B at a time.
  • Each data center 240A, 240B includes web demilitarized zone equipment 231 A and 23 IB, respectively, which is configured to receive the agent endpoints 211A, 21 IB and task endpoints 212A, 212B, which are communicatively connecting to TCaaS 230 via the Internet.
  • Web demilitarized zone (DMZ) equipment 231 A and 23 IB may operate outside a firewall to connect with the agent endpoints 211 A, 211B and task endpoints 212A, 212B while the rest of the components of data centers 240A, 240B may be within said firewall (besides the telephony DMZ equipment 232A, 232B, which may also be outside said firewall).
  • each data center 240A, 240B includes telephony DMZ equipment 232 A and 232B, respectively, which is configured to receive agent endpoints 211 A, 211B and task endpoints 212A, 212B, which are communicatively connecting to TCaaS 230 via the PSTN.
  • Telephony DMZ equipment 232A and 232B may operate outside a firewall to connect with the agent endpoints 211 A, 211B and task endpoints 212A, 212B while the rest of the components of data centers 240A, 240B (excluding web DMZ equipment 231 A, 23 IB) may be within said firewall.
  • each data center 240A, 240B may include one or more nodes 233A, 233B, and 233C, 233D, respectively. All nodes 233A, 233B and 233C, 233D may communicate with web DMZ equipment 231 A and 23 IB, respectively, and with telephony DMZ equipment 232 A and 232B, respectively. In some embodiments, only one node in each data center 240A, 240B may be communicating with web DMZ equipment 231 A, 23 IB and with telephony DMZ equipment 232 A, 232B at a time.
  • Each node 233 A, 233B, 233C, 233D may have one or more pairing modules 234A, 234B, 234C, 234D, respectively. Similar to pairing module 140 of task assignment center 100A of Figure 1A, pairing modules 234A, 234B, 234C, 234D may pair tasks to agents. For example, the pairing module may alternate between enabling pairing via a Behavioral Pairing (BP) module and enabling pairing with a First-in-First- out (FIFO) module. In other embodiments, one pairing module (e.g., the BP module) may be configured to emulate other pairing strategies.
  • BP Behavioral Pairing
  • FIFO First-in-First- out
  • a pairing module In pairing agent endpoints to task endpoints, a pairing module (e.g., pairing modules 234A, 234B, 234C, 234D) causes a change to the technical state of the task assignment system 200B.
  • the pairing strategy or pairing model used by a pairing module may result in an improvement to the technical operation of the task assignment system 200B by utilizing the physical resources of the task assignment system 200B in a more efficient manner.
  • TCaaS 230 is shown in Figure 2C as comprising two tenants 250A and 250B.
  • multi-tenancy may be supported by node 233A supporting tenant 250A while node 233B supports 250B.
  • data center 240A supports tenant 250A while data center 240B supports tenant 250B.
  • multi-tenancy may be supported through a shared machine or shared virtual machine; such at node 233A may support both tenants 250A and 250B, and similarly for nodes 233B, 233C, and 233D.
  • Figure 3 illustrates a block diagram of a task assignment system 300 according to an aspect of the present disclosure.
  • the task assignment system 300 comprises a task assignment module 302, a task assignment strategy module 304, a historical assignment module 306, and a benchmarking module 308.
  • the task assignment module 302, the task assignment strategy module 304, the historical assignment module 306, and the benchmarking module 308 correspond to the task assignment module 150, the task assignment strategy module 180, the historical assignment module 190, and the benchmarking module 195 of the example task assignment system 100B of Figure IB.
  • the task assignment system 300 further comprises a monitoring module 310 and a control module 312 communicatively coupled to the monitoring module 310.
  • the monitoring module 310 may also be communicatively coupled to the task assignment module 302, the task assignment strategy module 304, the historical assignment module 306, and the benchmarking module 308.
  • the task assignment module 302 is configured to pair (e.g., match, assign) incoming tasks to available agents.
  • m tasks 314A-314B are received over a given period, and n agents 316A-316B are available during the given period.
  • Each of the m tasks may be assigned to one of the n agents according to a task assignment strategy (or pairing strategy) implemented by the task assignment strategy module 304.
  • the task assignment strategy module 304 may implement a FIFO strategy, a PBR strategy, a BP strategy such as a multidimensional BP strategy, and the like.
  • the task assignment module 302 may change which pairing strategy is employed depending on various factors.
  • the task assignment module 302 may be instructed by the benchmarking module 308 to cycle between two or more pairing strategies in order to benchmark performance.
  • the monitoring module 310 is configured to collect data quantifying the attributes (outcomes, outputs, performance values, etc.) of task-agent pairings involving agents within the task assignment system 300. For each agent within the task assignment system 300 (e.g., the n agents 316A-316B), one or more attributes of task-agent pairings involving that agent are recorded and collated over a specific time period (e.g., 1 week, 1 month. 2 months, 3 months, 6 months, 12 months, etc.). Attribute values for an individual agent or a group of agents may then be calculated from this data (e.g., by calculating an average value for an attribute over a time period).
  • a specific time period e.g., 1 week, 1 month. 2 months, 3 months, 6 months, 12 months, etc.
  • an attribute value calculated for a plurality of agents may be referred to as a representative attribute value.
  • Example attribute values include average outcome, conversion rate, average agent compensation, average handle time, average task type, and average task mix.
  • Average outcome can be calculated for an agent over a time period -> t 2 by recording the outcome (e.g., a positive or negative outcome indicative of whether the agent was able to complete or adequately deal with the task) of each task-agent pairing involving the agent over the time period -> t 2 and calculating the average (e.g., mean, median, mode) from the recorded outcome data.
  • the average outcome may be considered a probability of the agent having a positive outcome.
  • Conversion rate can be calculated for an agent over the time period C -> t 2 by calculating the ratio of the number of successful conversions attained by the agent over -> t 2 and the total number of task-agent interactions involving the agent over ti -> t 2 .
  • Average agent compensation can be calculated for an agent over the time period -> t 2 by determining an average value of the compensation due to the agent over -> t 2 .
  • Average handle time can be calculated for an agent over the time period -> t 2 by recording the handle time of each task-agent pairing involving the agent over the time period -> t 2 and calculating the average (e.g., mean, median, mode) from the recorded handle time data.
  • Average task type can be calculated for an agent over the time period -> t 2 by recording the number of each type of task the agent handles over -> t 2 (e.g., in a histogram) and determining the most frequent task type (e.g., the largest histogram bin).
  • the average task can be calculated as a probability that the agent received a specific task type over the time period -> t 2 .
  • Average task mix can be calculated for an agent over the time period -> t 2 by calculating the proportion of tasks according to task type that the agent receives over the time period ti -> t 2 .
  • the average task mix for the agent over the time period -> t 2 is the vector of values (0.25,0.6,0.15).
  • the average task mix is reported as a single value corresponding to the variance of the vector of values (e.g., the average task mix for the previous vector of values would be reported as 0.04).
  • bias is generally understood as a difference in attribute values for task-agent pairings involving one group of agents when compared to the attribute values for task-agent pairings involving another group of agents. Whilst in other contexts, “bias” may be understood as being dependent on factors such as an individual’s protected characteristic(s), in the present disclosure bias refers to the often hidden perturbation of attribute values for task-agent pairings involving an agent which results from latent characteristics of the pairing strategy used to determine the task-agent pairings.
  • an optimized pairing strategy may increase the likelihood that an individual agent may be paired with tasks having higher expected difficulty and/or challenging tasks (as determined by an optimized pairing strategy from historical data regarding the agent’ s previous handling of tasks). Such non-uniform pairings may inadvertently result in an agent being penalized for changes (e.g., longer average handle times, less variety in assigned tasks, lower conversion rates, etc.). Conversely, an agent may be paired with easier and/or less challenging tasks thereby resulting in the agent receiving potentially skewed outcomes (e.g., higher conversion rates).
  • “bias” may alternatively be referred to as pairing strategy distortion, induced distortion, skew, pairing strategy skew, or perturbation.
  • the present disclosure is directed to detecting and mitigating such bias within task assignment systems. In doing so, the overall operation and performance of the task assignment system may be improved whilst reducing the potential distortion or perturbation to and agent’s outcomes occurring as a result of potential pairing strategy induced bias.
  • Figure 4 shows a method 400 for bias mitigation in a task assignment system according to an aspect of the present disclosure.
  • the method 400 comprises the steps of identifying 402 a plurality of agents, determining 404 a representative attribute value, identifying 406 an outlier agent, and adjusting 408 operation of the task assignment system.
  • the method 400 further comprises the optional step of adjusting 410 operation of the task assignment system if the outlier agent satisfies a return criterion.
  • the method 400 may be performed by a monitoring module and control module (e.g., the monitoring module 310 and the control module 312 of the task assignment system 300 of Figure 3) or any other suitable module(s) within a task assignment system.
  • a plurality of agents within the task assignment system are identified.
  • the plurality of agents are paired with tasks according to a first pairing strategy.
  • the first pairing strategy is an optimized pairing strategy such as a behavioral pairing strategy, a performance based routing pairing strategy, a prioritized call routing pairing strategy, or an optimized first- in-first-out (FIFO) pairing strategy.
  • the plurality of agents are paired with tasks during a first time period according to the first pairing strategy only. That is, for every task that each of the plurality of agents is paired with, the pairing is determined according to the first pairing strategy.
  • the plurality of agents are paired with tasks during the first time period according to the first pairing strategy and one or more other pairing strategies (e.g., different optimized pairing strategies or FIFO pairing strategies for the purpose of benchmarking).
  • the data collected for each task-agent pairing for the purpose of bias mitigation (as described below) relates to the first pairing strategy only.
  • a representative attribute value of an attribute of task-agent pairings involving the plurality of agents over a first time period is determined.
  • attribute values for each task-agent pairings involving the plurality of agents are recorded (e.g., by a monitoring module such as the monitoring module 310 of the task assignment system 300 shown in Figure 3).
  • a monitoring module such as the monitoring module 310 of the task assignment system 300 shown in Figure 3
  • an attribute value for a task-agent pairing is a value which quantifies an attribute of the task-agent pairing such as the outcome, the handle time, etc. over the first time period.
  • the representative attribute value is a value which represents, or is indicative of, the attribute for the plurality of agents across all task-agent pairings.
  • the representative attribute value may be an average handle time for all task-agent pairings involving the plurality of agents over the first time period.
  • the representative attribute value may be one of an average outcome, an average conversion rate, an average agent compensation, an average handle time, an average task type, or an average task mix.
  • the representative attribute value is determined with respect to task-agent pairings involving the plurality of agents as assigned by the first pairing strategy.
  • the representative attribute value may be an average outcome for task-agent pairings determined by the first pairing strategy.
  • the first pairing strategy may itself influence the representative value, or the proportion of tasks assigned by the first pairing strategy may be small enough to allow for significant random deviations from the “true” representative value (e.g., the “true” average outcome for all agents).
  • the representative attribute value may alternatively be determined with respect to task-agent pairings involving the plurality of agents as assigned by the first pairing strategy and/or one or more other pairing strategies (e.g., FIFO, random pairing, etc.).
  • this helps to reduce or eliminate the above mentioned bias and reduce random variations.
  • a plurality of representative attribute values of a plurality of attributes of task-agent pairings involving the plurality of agents over the first time period are determined.
  • the representative attribute value may be a vector of values with each element corresponding to an attribute value of task-agent pairings involving the plurality of agents over the first time period.
  • an outlier agent within the plurality of agents is identified based on the representative attribute value.
  • the outlier agent has an outlier attribute value indicative of a bias related to the first pairing strategy.
  • an attribute value of the attribute of task-agent pairings involving each agent over the first time period is recorded (i.e., each agent is associated with an attribute value which quantifies taskagent pairings involving that agent over the first time period).
  • the attribute value is one of: an average outcome of task-agent pairings involving the agent over the first time period; an average conversion rate of task-agent pairings involving the agent over the first time period; an average agent compensation for the agent over the first time period; an average handle time of task-agent pairings involving the agent over the first time period; an average task type of task-agent pairings involving the agent over the first time period; or an average task mix of task-agent pairings involving the agent over the first time period.
  • the attribute of the attribute value is the same as the attribute of the representative attribute value such that an outlier agent can be identified from the plurality of agents based on a comparison of the outlier attribute value for the outlier agent and the representative attribute value for the plurality of agents.
  • An outlier agent is an agent within the plurality of agents having an attribute value (an outlier attribute value) which differs from the representative attribute value of the plurality of agents by more than a first threshold amount (i.e., the outlier attribute value differs significantly from the representative attribute value).
  • the attribute value for the outlier agent is determined with respect to task-agent pairings involving the outlier agent as assigned by the first pairing strategy.
  • the first threshold amount is based on a statistical function of a distribution of values of the attribute determined from task-agent pairings involving the plurality of agents over the first time period (as described in more detail in relation to Figure 7 below).
  • the statistical function is a z-score indicative of a number of standard deviations that the first threshold amount deviates from the representative attribute value (which is an average of the distribution of values).
  • the first threshold amount may be at least two standard deviations away from the representative attribute value or at least four standard deviations away from the representative attribute value.
  • the first threshold amount is a predetermined value (i.e., the first threshold amount is static and does not depend on the distribution of attribute values for the plurality of agents over the first time period).
  • the outlier agent is identified within the plurality of agents based on the plurality of representative attribute values, where the outlier agent has at least one outlier attribute value in relation to the plurality of representative attribute values. That is, for a plurality of representative attribute values (a 1 , ci 2 , ... , a n ) and a plurality of attribute values for the outlier agent (o 1 , o 2 , ... , o n ), at least one of the attribute values for the outlier agent o m is an outlier with respect to its corresponding representative attribute value a m (e.g.,
  • at least a portion (e.g., over 50%, over 75%, over 90%, etc.) of the plurality of attribute values for the outlier agent are outliers with respect to their corresponding representative attribute values.
  • operation of the task assignment system is adjusted such that the outlier agent is transitioned from being paired with tasks according to the first pairing strategy to being paired with tasks according to a second pairing strategy.
  • the outlier agent is transitioned from being paired according to the first pairing strategy to being paired according to a second pairing strategy (e.g., by a control module such as the control module 312 of the task assignment system 300 shown in Figure 3).
  • the second pairing strategy is a first-in-first-out (FIFO) pairing strategy (or a round robin pairing strategy or a random agent selection pairing strategy).
  • the second pairing strategy is a modified optimized pairing strategy. More particularly, the second pairing strategy may be modified such that the agent-related data for the outlier agent (upon which the pairing strategy makes a pairing decision) is held back. The modified pairing strategy is then “blind” to the agent data for the outlier agent thereby avoiding any potential bias induced by the pairing strategy.
  • the modified pairing strategy is adjusted such that the assessment of the outlier agent is weighted toward values which would adjust or skew the attribute value for the outlier agent in the opposite direction (e.g., impart a positive or negative bias on the agent).
  • the modified pairing strategy is adjusted such that the outlier agent is assigned a neutral, but not purely random, mix of tasks.
  • the outlier agent may be gradually transitioned to the second pairing strategy over a second time period (e.g., 1 day, 2 days, 1 week, 2 weeks, 1 month, etc.). That is, rather than adjust operation of the task assignment system such that the outlier agent is immediately paired according to the second pairing strategy once the outlier agent has been identified as an outlier, the proportion of pairings involving the first pairing strategy for the outlier agent is gradually reduced over the second time period whilst the proportion of pairings involving the second pairing strategy for the outlier agent is gradually increased over the second time period.
  • the outlier agent can be transitioned to the second pairing strategy over the second time period according to a first target proportion which determines the proportion of pairings involving the outlier agent made by the second pairing strategy at a first target time point during the second time period.
  • 0% of the pairings involving the outlier agent are determined using the second pairing strategy.
  • a target proportion of 100% may be set at time point (which is within the second time period) such that, at time t , all pairings involving the outlier agent are determined according to the second pairing strategy.
  • the proportion is changed according to a linear or non-linear scale factor.
  • the rate of change for a linear scaling factor is constant such that the proportion of tasks determined by the second pairing strategy linearly increases over the second time period.
  • the rate of change for a non-linear scaling factor changes over time (i.e., the proportion non-linearly increases over the second time period). Examples of non-linear scaling factors include logistic scaling, quadratic scaling, cubic scaling, etc.
  • the proportion of pairings determined by each pairing strategy for the outlier agent may then stay at 80% (i.e., until subsequently transitioned back) or increase from 80% to 100% (e.g., during the remainder of the second time period or over a longer time period).
  • the second pairing strategy may not be an optimal pairing strategy for the outlier agent. That is, because the second pairing strategy is FIFO (or random) as opposed to optimized, the outlier agent may not be assigned to the most optimal pairings whilst paired according to the second pairing strategy. It is therefore advantageous to eventually transition the outlier agent back to the first pairing strategy to help ensure that pairings within the task assignment system are optimized. Transitioning the outlier agent back to the first pairing strategy too early may result in a repeat of the condition which originally led to the outlier agent being transitioned away from the first pairing strategy.
  • the outlier agent is transitioned back to the first pairing strategy when the outlier agent satisfies a predetermined return criterion. In a further less preferred embodiment transitioning the outlier agent back to the first pairing strategy occurs after a predetermined period of time (e.g., 1 week, Imonth etc.).
  • step of adjusting 410 if the outlier agent satisfies a predetermined return criterion, operation of the task assignment system is adjusted such that the outlier agent is transitioned from being paired with tasks according to the second pairing strategy to being paired with tasks according to the first pairing strategy over a third time period. That is, a determination is made at the step of adjusting 410 and, if it is determined that the outlier agent satisfies the predetermined return criterion, then the outlier agent is transitioned back to the first pairing strategy. Otherwise, the outlier agent remains having pairings determined according to the second pairing strategy.
  • the predetermined return criterion is satisfied when the outlier agent is determined to have a value for the attribute which differs from a target representative value of the attribute by less than a second threshold amount.
  • the value for the attribute is an average value of the attribute calculated from task-agent pairings involving the outlier agent over a fourth time period (subsequent to the second time period).
  • the target representative value may be an average value of the attribute calculated from task-agent pairings involving the plurality of agents over the fourth time period.
  • the plurality of agents are paired with tasks over the fourth time period according to the first pairing strategy.
  • the second threshold amount is based on a statistical function of a distribution of values of the attribute determined from task-agent pairings involving the plurality of agents over the fourth time period according to the first pairing strategy.
  • the statistical function may be a standard deviation or z-score of the distribution of values (e.g., the second threshold amount is at least two standard deviations of the distribution of values or at least four standard deviations of the distribution of values).
  • the outlier agent is gradually transitioned to the first pairing strategy over the third time period according to a second target proportion.
  • the second target proportion determines a proportion of pairings involving the outlier agent made by the first pairing strategy at a second target time point during the third time period.
  • the second target time point may be determined based on a second transition rate.
  • the second target proportion and the second transition rate operate in a similar fashion to (but may have different values than) the first target proportion and the first transition rate described in detail above. As such, the skilled person will appreciate that the above description of the first target proportion and the first transition rate are applicable to the second target proportion and the second transition rate.
  • Figure 5 shows a method 500 for bias mitigation in a task assignment system according to an aspect of the present disclosure.
  • the method 500 comprises the steps of identifying 502 a plurality of agents, determining 504 a plurality of performance values, determining 506 an outlier agent, and transitioning 508 the outlier agent.
  • the method 500 further comprises the optional step of transitioning 510 the outlier agent if the outlier agent satisfies a return criterion.
  • the method 500 may be performed by a monitoring module and control module (e.g., the monitoring module 310 and the control module 312 of the task assignment system 300 of Figure 3) or any other suitable module(s) within a task assignment system.
  • identifying 502, determining 504, determining 506, transitioning 508, and transitioning 510 are similar to the steps of identifying 402, determining 404, identifying 406, adjusting 408, and adjusting 410 of the method 400 in Figure 4.
  • a plurality of agents paired with tasks by the task assignment system according to a first pairing strategy are identified within the task assignment system.
  • the first pairing strategy is an optimized pairing strategy such as a behavioral pairing strategy, a performance based routing pairing strategy, a prioritized call routing pairing strategy, or an optimized first- in-first-out (FIFO) pairing strategy.
  • the plurality of agents are paired with tasks during a first time period according to the first pairing strategy only.
  • the plurality of agents are paired with tasks during the first time period according to the first pairing strategy and one or more other pairing strategies (e.g., different optimized pairing strategies or FIFO pairing strategies for the purpose of benchmarking). This process can occur as described with respect to the step of identifying 402 in Figure 4.
  • a plurality of performance values related to a plurality of task-agent pairings involving the plurality of agents over a first time period are determined.
  • performance values are determined from data collected for task-agent pairings involving the plurality of agents over the first time period.
  • the plurality of performance values is one of: a plurality of average outcome values; a plurality of average conversion rate values; a plurality of average agent compensation values; a plurality of average handle time values; a plurality of average task type values; or a plurality of average task mix values.
  • the plurality of performance values are determined from task-agent pairings involving the plurality of agents according to the first pairing strategy only, or according to the first pairing strategy and/or one or more other pairing strategies.
  • an outlier agent within the plurality of agents is determined based on the plurality of performance values.
  • the outlier agent has an outlier performance value which differs from a representative performance value for the plurality of agents by at least a first threshold amount.
  • the outlier agent has a performance value which is an outlier performance value if it differs from the representative performance value.
  • the performance value for the outlier agent is determined with respect to task-agent pairings involving the outlier agent as assigned by the first pairing strategy.
  • the outlier performance value is one of: an average outcome of task-agent pairings involving the outlier agent over the first time period; an average conversion rate of task-agent pairings involving the outlier agent over the first time period; an average agent compensation for the outlier agent over the first time period; an average handle time of task-agent pairings involving the outlier agent over the first time period; an average task type of task-agent pairings involving the outlier agent over the first time period; or an average task mix of task-agent pairings involving the outlier agent over the first time period.
  • the representative performance value may be an average of the distribution of the plurality of performance values. That is, the representative performance value is a value which represents, or is indicative of, the distribution of the plurality of performance values.
  • the representative performance value may be an average handle time or an average outcome for all task-agent pairings involving the plurality of agents over the first time period.
  • the representative performance value is determined with respect to task-agent pairings involving the plurality of agents as assigned by the first pairing strategy only or with respect to task-agent pairings involving the plurality of agents assigned by the first pairing strategy and/or one or more other pairing strategies (e.g., FIFO, random pairing, etc.).
  • the first threshold amount may be based on a statistical function of a distribution of the plurality of performance values.
  • the statistical function is a standard deviation of the distribution of the plurality of performance values (e.g., a z-score).
  • the first threshold amount may be at least two standard deviations of the distribution of the plurality of performance values or at least four standard deviations of the distribution of the plurality of performance values.
  • the first threshold amount is a first predetermined value.
  • the outlier agent is transitioned from being paired with tasks according to the first pairing strategy to being paired with tasks according to a second pairing strategy such that a subsequent performance value obtained for the outlier agent when paired according to the second pairing strategy differs from the representative performance value for the plurality of agents by less than the first threshold amount.
  • the second pairing strategy is a first-in-first-out (FIFO) pairing strategy (a round robin pairing strategy or a random agent selection pairing strategy).
  • the second pairing strategy is a modified optimized pairing strategy as described above in relation to Figure 4.
  • the outlier agent may be gradually transitioned to the second pairing strategy over a second time period according to a first target proportion.
  • the first target proportion determines a proportion of pairings involving the outlier agent made by the second pairing strategy at a first target time point during the second time period.
  • the first target time point may be determined based on a first transition rate.
  • the outlier agent may be transitioned in the same way as described in relation to the adjusting 408 step of the method 400 in Figure 4.
  • a subsequent performance value i.e., a performance value obtained from task-agent pairings involving the outlier agent during a further time period after the outlier agent has been transitioned, or after transitioning has begun — differs from a representative performance value by less than the first threshold amount.
  • the representative performance value is the representative performance value determined from task-agent pairings during the first time period.
  • the representative performance value is determined from task-agent pairings for the plurality of agents during the further time period mentioned above.
  • the outlier agent is transitioned from being paired with tasks according to the second pairing strategy to being paired with tasks according to the first pairing strategy over a third time period if the outlier agent satisfies a predetermined return criterion. That is, a determination is made at the step of transitioning 510 and, if it is determined that the outlier agent satisfies the predetermined return criterion, then the outlier agent is transitioned back to the first pairing strategy. Otherwise, the outlier agent remains having pairings determined according to the second pairing strategy.
  • the predetermined return criterion is satisfied when the outlier agent is determined to have a performance value which differs from a target performance value by less than a second threshold amount.
  • the performance value of the outlier agent is an average performance value calculated from task-agent pairings involving the outlier agent over a fourth time period.
  • the fourth time period is a period subsequent to the outlier agent being transitioned to the second pairing strategy.
  • the target performance value is an average performance value calculated from task-agent pairings involving the plurality of agents over the fourth time period (and determined according to the first pairing strategy).
  • the target performance value is a predetermined performance value.
  • the second threshold amount is based on a statistical function of a distribution of performance values of task-agent pairings involving the plurality of agents over the fourth time period.
  • the statistical function may be a standard deviation of the distribution of values (e.g., a z-score).
  • the second threshold amount is at least two standard deviations of the distribution of values or at least four standard deviations of the distribution of values.
  • the outlier agent may be gradually transitioned to the first pairing strategy over the third time period according to a second target proportion.
  • the second target proportion determines a proportion of pairings involving the outlier agent made by the first pairing strategy at a second target time point during the third time period.
  • the second target time point may be determined based on a second transition rate.
  • Figure 6 shows a method 600 for agent outcome optimization within a task assignment system according to an aspect of the present disclosure.
  • the method 600 comprises the steps of collecting 602 outcome data, generating 604 a distributional representation of the outcome data, identifying 606 one or more outlier agents, and causing 608 the task assignment system to transition the outlier agents.
  • the method 600 further comprises the optional steps of determining 610 if an outlier agent satisfies a return criterion, maintaining 612 the pairing strategy for the outlier agent if the return criterion is no satisfied, and causing 614 the task assignment system to transition the outlier if the return criterion is satisfied.
  • the method 600 may be performed by a monitoring module and control module (e.g., the monitoring module 310 and the control module 312 of the task assignment system 300 of Figure 3) or any other suitable module(s) within a task assignment system.
  • outcome data is collected for a plurality of agents which are paired to tasks by a task assignment system according to a first pairing strategy.
  • the outcome data is collected by a monitoring module such as the monitoring module 310 of the task assignment system 300 shown in Figure 3.
  • the outcome data comprises a plurality of outcome values quantifying outcomes of task-agent pairings involving the plurality of agents over a first time period. Examples of outcome data include handle time, task outcome, conversion, compensation, task type, task mix, and the like.
  • the first pairing strategy is an optimized pairing strategy such as a behavioral pairing strategy, a performance based routing pairing strategy, a prioritized call routing pairing strategy, or an optimized first-in-first-out (FIFO) pairing strategy.
  • the plurality of outcome values are determined from task-agent pairings involving the plurality of agents according to the first pairing strategy only, or according to the first pairing strategy and/or one or more other pairing strategies.
  • a distributional representation of the outcome data is generated.
  • the distributional representation of the outcome data is generated by a control module such as the control module 312 of the task assignment system 300 shown in Figure 3.
  • An example distributional representation is shown in Figure 7 as described below.
  • the distributional representation of the outcome data provides a representation of the distribution of the outcome data over the first time period.
  • the distributional representation is then analyzed to determine characteristics of task-agent pairings over the first time period (such as the average, or representative, outcome) as well as to identify outlier agents whose outcomes may be distorted or perturbed as a result of being paired with tasks according to the first pairing strategy.
  • one or more outlier agents within the plurality of agents are identified from the distributional representation of the outcome data.
  • Each of the one or more outlier agents have outlier outcome values in relation to an average outcome value of the distributional representation of the outcome data.
  • the one or more outlier agents are identified by a control module such as the control module 312 of the task assignment system 300 shown in Figure 3.
  • the average outcome value of the distributional representation of the outcome data corresponds to a representative outcome value for task-agent pairings involving the plurality of agents over the first time period.
  • the average outcome value may be a modal average, a mean average, a median average, a weighted average, or the like.
  • Each of the outlier outcome values differ from the average outcome value by more than a first threshold amount.
  • the first threshold amount is based on a statistical function of the distributional representation of the outcome data.
  • the statistical function is a standard deviation of the distributional representation (e.g., a z-score).
  • the first threshold amount is at least two standard deviations of the distributional representation or at least four standard deviations of the distributional representation.
  • the first threshold amount is a first predetermined value. That is, the first threshold amount is static and/or does not depend on the distributional representation.
  • the task assignment system is caused to transition the one or more outlier agents from the first pairing strategy to a second pairing strategy over a second time period.
  • the transitioning of the task assignment system is caused (i.e., performed or controlled) by a control module such as the control module 312 of the task assignment system 300 shown in Figure 3.
  • the second pairing strategy is a first-in-first-out (FIFO) pairing strategy (a round robin pairing strategy or a random agent selection pairing strategy).
  • the second pairing strategy is a modified optimized pairing strategy as described above in relation to Figure 4.
  • the one or more outlier agents may be transitioned to the second pairing strategy over the second time period according to a first target proportion.
  • the first target proportion determines a proportion of pairings involving the one or more outlier agents made by the second pairing strategy at a first target time point during the second time period.
  • the first target time point may be determined based on a first transition rate.
  • each of the one or more outlier agents have the same first target proportion, first target time point, and first transition rate.
  • the first target proportion, first target time point, and/or first transition rate varies across the one or more outlier agents such that each agent can be transitioned to the second pairing strategy according to a custom transition strategy.
  • the predetermined return criterion may be satisfied when the outlier agent is determined to have an outcome value which differs from a target outcome value by less than a second threshold amount.
  • the outcome value is an average outcome value calculated from task-agent pairings involving the outlier agent over a fourth time period.
  • the target outcome value is an average outcome value calculated from task-agent pairings involving the plurality of agents over the fourth time period.
  • the plurality of agents are paired with tasks during the fourth time period according to the first pairing strategy.
  • the target outcome value is a predetermined value.
  • the method 600 proceeds to the step of maintaining 612 where the outlier agent is maintained on the second pairing strategy.
  • the outlier agent may be subsequently re-evaluated with respect to the predetermined return criterion at a later time point. If it is determined that the outlier agent does satisfy the predetermined return criterion, then the method 600 proceeds to the step of causing 614.
  • the task assignment system is caused to transition the outlier agent from being paired with tasks according to the second pairing strategy to being paired with tasks according to the first pairing strategy over a third time period.
  • the outlier agent may be transitioned to the first pairing strategy over the third time period according to a second target proportion.
  • the second target proportion determines a proportion of pairings involving the outlier agent made by the first pairing strategy at a second target time point during the third time period.
  • the second target time point may be determined based on a second transition rate.
  • bias adjustment may be achieved by adjusting operation of the task assignment system to address perceived bias.
  • perceived bias is to be understood as a potential bias (skew, distortion, perturbation) which is not quantified by differences in attributes or outcomes in task-agent pairings (as described above); but is identified through a perceived difference in attributes or outcomes in task-agent pairings. That is, perceived bias may arise with regard to an attribute that is not present within the agent data (e.g., as measured by a monitoring module such as the monitoring module 310 of the task assignment system 300 in Figure 3) or may not be otherwise measurable. For example, a perceived bias may arise in relation to the proportion of tasks assigned to an agent from a specific geographic region, or the proportion of tasks assigned to an agent from a certain demographic, which are perceived by the agent to impact their performance detrimentally compared to the performance of other agents.
  • a potential perceived bias may therefore arise in relation to an agent’s performance with respect to task-agent pairings involving the agent and assigned according to a first pairing strategy (e.g., an optimized pairing strategy).
  • the perceived difference may be identified by the agent through feedback obtained during an agent’s shift or shifts, survey data, an appraisal by a manager, etc.
  • an agent may notice through discussion with her peers that her compensation is lower than those of her peers when she is paired with tasks according to a first pairing strategy (e.g., a behavioral pairing strategy or a performance based routing strategy). Reporting this perceived difference (e.g., to a manager or via a central reporting portal) therefore identifies a potential perceived bias in relation to taskagent pairings assigned to the agent by the first pairing strategy.
  • the agent may be transitioned from being assigned with tasks according to the first pairing strategy to being assigned with tasks according to a second pairing strategy (e.g., FIFO).
  • the transitioning process may be the same as the transitioning process described in relation to Figures 4-6 above.
  • the agent may then be monitored whilst being assigned tasks according to the second pairing strategy. For example, a manager may perform regular reviews with the agent to determine if the potential perceived bias has been mitigated or reduced. As a further example, the agent may be asked to report changes in potential perceived bias to a central reporting portal. If the potential perceived bias is determined to be mitigated or substantially reduced, then the agent may be transitioned back to the first pairing strategy (as described above in relation to Figures 4-6).
  • Figure 7 shows a plot 702 of a distributional representation of attribute data according to embodiments of the present disclosure.
  • a distributional representation can be used to determine characteristics of task-agent pairings involving a pairing strategy over a time period (such as the average/representative attribute, outcome, or performance) as well as to identify outlier agents whose outcomes may be distorted or perturbed as a result of being paired with tasks according to the first pairing strategy.
  • the plot 702 comprises a distribution 704 of attribute values for a plurality of agents within a task assignment system.
  • the x-axis of the plot 702 is shown in relation to a score which is a statistical function of the distribution 704 of attribute values (alternatively referred to as outcome values or performance values).
  • Figure 7 further shows a first score 706 associated with a representative attribute value, a first threshold amount 708 (or first threshold value or first threshold score), a second threshold amount 710 (or second threshold value or second threshold score), a second score 712 associated with a first agent, and a third score 714 associated with a second agent.
  • the distribution 704 is calculated from task-agent pairings involving the plurality of agents within the task assignment system over a first time period (e.g., the n agents 316A-316B of the task assignment system 300 shown in Figure 3).
  • a first time period e.g., the n agents 316A-316B of the task assignment system 300 shown in Figure 3.
  • each of the plurality of agents were paired with tasks according to an optimized pairing strategy (e.g., behavioral pairing, performance based routing, etc.) and/or one or more other pairing strategies (e.g., optimized, FIFO).
  • an attribute of the task-agent pairing was recorded (e.g., the handle time, the outcome, etc.).
  • the distribution 704 corresponds to the distribution of average attributes for the plurality of agents over the first time period. For example, the distribution of average handle time, average conversion rate, etc. for the plurality of agents.
  • the first score 706 corresponds to a representative value of the attribute (i.e., the modal
  • agents having a score associated with the right hand tail of the distribution 704 when paired with tasks according to the optimized pairing strategy are determined to be potentially negatively impacted by the optimized pairing strategy. That is, a positive score is associated with a potentially negative impact on the agent by the optimized pairing strategy.
  • the second score 712 is above the first threshold amount 708 and is thus identifiable as an outlier score.
  • the second score 712 is calculated from task-agent pairings assigned by the optimized pairing strategy and involving the first agent.
  • the first agent is an outlier agent and may therefore be considered to be potentially negatively impacted by the optimized pairing strategy.
  • agents having a score above the first threshold amount 708 are within a tail of the distribution 704, they may be considered outliers with respect to the rest of the population of agents.
  • the third score 714 (calculated from task-agent pairings assigned by the optimized pairing strategy and involving the second agent) is below the first threshold amount 708 and is thus not an outlier score.
  • the second agent is not an outlier agent.
  • agents associated with the left hand tail of the distribution 704 are determined to be potentially positively impacted by the optimized pairing strategy. That is, a negative score is associated with a potentially positive impact on the agent by the optimized pairing strategy. For example, agents having a score below the second threshold amount 710 may be considered potentially positively impacted by the optimized pairing strategy. Given that agents having a score below the second threshold amount 710 are within a tail of the distribution 704, they may be considered outliers with respect to the rest of the population of agents.
  • Figure 8 illustrates the above system and methods for bias mitigation according to embodiments of the present disclosure.
  • Figure 8 shows a plot of relative performance (relative outcome or relative change in attribute value) of an agent within a task assignment system over a time period -> t 7 .
  • Figure 8 further shows a plot of which pairing strategy (PSI or PS2) is being used to assign tasks to the agent over the time period.
  • PSI pairing strategy
  • the agent is paired with tasks according to a first pairing strategy (PSI) which is an optimized pairing strategy such as behavioral pairing (BP) or performance based routing (PBR).
  • PSI first pairing strategy
  • BP behavioral pairing
  • PBR performance based routing
  • the performance e.g., handle time, outcome, conversion rate, etc.
  • Figure 8 shows a plot of the relative performance of the agent over the first time period 802 which is calculated as the difference between the agent’s performance and a representative performance of other agents within the task assignment system which are paired with tasks according to PSI and/or PS2 during the first time period 802.
  • the average of the relative performance of the agent over the first time period 802 — as indicated by the first average performance value 804 calculated at time point t 2 — is determined to be above a first threshold amount a v . That is, at time point t 2 , a monitoring module (e.g., the monitoring module 310 shown in Figure 3) calculates the average performance of the agent over the first time period 802 (i.e., the first average performance value 804) and determines that the agent is an outlier agent with respect to the performance of the other agents because the first average performance value 804 is above the first threshold amount a v . This indicates that the agent is being potentially penalized by PSI with respect to the performance being measured (e.g., handle time, outcome, etc.).
  • the first threshold amount a v corresponds to 3 standard deviations away from the average relative performance according to the distribution of average performance values for all agents paired with tasks according to PSI during the first time period 802.
  • a control module e.g., the control module 312 of the task assignment system 300 shown in Figure 3
  • PS2 is a FIFO pairing strategy.
  • the agent is transitioned to the second pairing strategy over a second time period 806. At the starting time point t 2 of the second time period 806 the agent is paired with tasks according to PSI.
  • the agent is gradually transitioned to PS2 such that at the ending time point t 3 of the second time period 806 the agent is paired with tasks according to PS2. Therefore, during the second time period 806 the agent is paired with tasks according to either PSI or PS2.
  • the proportion of tasks paired using either PS 1 or PS2 during the second time period 806 is changed according to a logistic scaling factor.
  • the performance of the agent is monitored over a third time period 808.
  • the average of the relative performance of the agent over the third time period 808 is calculated as the second average performance value 810.
  • the second average performance value 810 corresponds to the average relative performance of the agent when paired with tasks according to PS2 as compared to the average performance of agents paired with tasks according to PSI during the third time period 808.
  • the second average performance value 810 is below a second threshold amount a L thereby indicating that the performance of the agent substantially corresponds to the average performance of the plurality of agents paired with tasks according to PS 1.
  • the performance of the agent has shifted toward the average performance of agents when paired with PSI (i.e., the potential bias incurred by the agent as a result of being paired according to PSI has been mitigated).
  • the second threshold amount a L corresponds to 1 standard deviation away from the average relative performance according to the distribution of average performance values for all agents paired with tasks according to PSI during the third time period 808.
  • the control module adjusts operation of the task assignment system to transition the agent from being paired with tasks according to PS2 to being paired with tasks according to PSI.
  • the agent is transitioned to PSI over a fourth time period 812.
  • the agent is paired with tasks according to PS2.
  • the agent is gradually transitioned to PSI such that at the ending time point t 6 of the fourth time period 812 the agent is paired with tasks according to PSI. Therefore, during the fourth time period 812 the agent is paired with tasks according to either PSI or PS2.
  • the proportion of tasks paired using either PSI or PS2 during the fourth time period 812 is changed according to a logistic scaling factor.
  • the potential bias incurred by the agent as a result of being paired with tasks by PSI during the first time period 802 is substantially mitigated (i.e., at time point t 7 the relative performance of the agent in comparison to other agents paired with tasks according to PSI is below the first threshold amount a y ).
  • Figure 9 shows an example computing system for bias identification and handling. Specifically, Figure 9 shows a block diagram of an embodiment of a computing system according to example embodiments of the present disclosure.
  • Computing system 900 can be configured to perform any of the operations disclosed herein such as, for example, any of the operations discussed with reference to the functional units (or modules) described in relation to Figure 3.
  • Computing system includes one or more computing device(s) 902.
  • Computing device(s) 902 of computing system 900 comprise one or more processors 904 and memory 906.
  • One or more processors 904 can be any general purpose processor(s) configured to execute a set of instructions.
  • one or more processors 904 can be one or more general-purpose processors, one or more field programmable gate array (FPGA), and/or one or more application specific integrated circuits (ASIC).
  • one or more processors 904 include one processor.
  • one or more processors 904 include a plurality of processors that are operatively connected.
  • One or more processors 904 are communicatively coupled to memory 906 via address bus 908, control bus 910, and data bus 912.
  • Memory 906 can be a random access memory (RAM), a read only memory (ROM), a persistent storage device such as a hard drive, an erasable programmable read only memory (EPROM), and/or the like.
  • Computing device(s) 902 further comprise I/O interface 914 communicatively coupled to address bus 908, control bus 910, and data bus 912.
  • Memory 906 can store information that can be accessed by one or more processors 904.
  • memory 906 e.g., one or more non-transitory computer-readable storage mediums, memory devices
  • the computer-readable instructions can be software written in any suitable programming language or can be implemented in hardware. Additionally, or alternatively, the computer-readable instructions can be executed in logically and/or virtually separate threads on one or more processors 904.
  • memory 906 can store instructions (not shown) that when executed by one or more processors 904 cause one or more processors 904 to perform operations such as any of the operations and functions for which computing system 900 is configured, as described herein.
  • memory 906 can store data (not shown) that can be obtained, received, accessed, written, manipulated, created, and/or stored.
  • the data can include, for instance, the data and/or information described herein in relation to Figures 1 to 14.
  • computing device(s) 902 can obtain from and/or store data in one or more memory device(s) that are remote from the computing system 900.
  • Computing environment 900 further comprises storage unit 916, network interface 918, input controller 920, and output controller 922.
  • Storage unit 916, network interface 918, input controller 920, and output controller 922 are communicatively coupled to central control unit 902 via I/O interface 914.
  • Storage unit 916 is a computer readable medium, preferably a non-transitory computer readable medium, comprising one or more programs, the one or more programs comprising instructions which when executed by the one or more processors 904 cause computing environment 900 to perform the method steps of the present disclosure.
  • storage unit 916 is a transitory computer readable medium.
  • Storage unit 916 can be a persistent storage device such as a hard drive, a cloud storage device, or any other appropriate storage device.
  • Network interface 918 can be a Wi-Fi module, a network interface card, a Bluetooth module, and/or any other suitable wired or wireless communication device.
  • network interface 918 is configured to connect to a network such as a local area network (LAN), or a wide area network (WAN), the Internet, or an intranet.
  • LAN local area network
  • WAN wide area network
  • intranet an intranet
  • task assignment and bias identification or mitigation in accordance with the present disclosure as described above may involve the processing of input data and the generation of output data to some extent.
  • This input data processing and output data generation may be implemented in hardware or software.
  • specific electronic components may be employed in a behavioral pairing module or similar or related circuitry for implementing the functions associated with task assignment in accordance with the present disclosure as described above.
  • one or more processors operating in accordance with instructions may implement the functions associated with task assignment in accordance with the present disclosure as described above.
  • Such instructions may be stored on one or more non-transitory processor readable storage media (e.g., a magnetic disk or other storage medium), or transmitted to one or more processors via one or more signals embodied in one or more carrier waves.
  • processor readable storage media e.g., a magnetic disk or other storage medium

Landscapes

  • Business, Economics & Management (AREA)
  • Human Resources & Organizations (AREA)
  • Engineering & Computer Science (AREA)
  • Strategic Management (AREA)
  • Marketing (AREA)
  • Economics (AREA)
  • Entrepreneurship & Innovation (AREA)
  • Operations Research (AREA)
  • Development Economics (AREA)
  • Game Theory and Decision Science (AREA)
  • Educational Administration (AREA)
  • Quality & Reliability (AREA)
  • Tourism & Hospitality (AREA)
  • Physics & Mathematics (AREA)
  • General Business, Economics & Management (AREA)
  • General Physics & Mathematics (AREA)
  • Theoretical Computer Science (AREA)
  • Signal Processing (AREA)
  • Telephonic Communication Services (AREA)

Abstract

L'invention concerne un procédé d'atténuation de biais dans un système d'attribution de tâches, mis en œuvre par au moins un processeur configuré pour fonctionner dans le système d'attribution de tâches, comprenant l'identification, à l'intérieur du système d'attribution de tâches, d'une pluralité d'agents qui sont appariés avec des tâches selon une première stratégie d'appariement, la détermination d'une valeur d'attribut représentative d'un attribut d'appariements entre tâches et agents impliquant la pluralité d'agents sur un premier intervalle de temps, l'identification d'un agent aberrant dans la pluralité d'agents sur la base de la valeur d'attribut représentative, l'agent aberrant ayant une valeur d'attribut aberrante indiquant un biais lié à la première stratégie d'appariement, et l'ajustement du fonctionnement du système d'attribution de tâche de telle sorte que l'agent aberrant passe d'un appariement avec des tâches selon la première stratégie d'appariement à un appariement avec des tâches selon une seconde stratégie d'appariement.
PCT/IB2025/052216 2024-03-14 2025-02-28 Capteur de compensation d'agent Pending WO2025191389A1 (fr)

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
US202463565504P 2024-03-14 2024-03-14
US63/565,504 2024-03-14

Publications (1)

Publication Number Publication Date
WO2025191389A1 true WO2025191389A1 (fr) 2025-09-18

Family

ID=95364764

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/IB2025/052216 Pending WO2025191389A1 (fr) 2024-03-14 2025-02-28 Capteur de compensation d'agent

Country Status (1)

Country Link
WO (1) WO2025191389A1 (fr)

Citations (7)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US9300802B1 (en) 2008-01-28 2016-03-29 Satmap International Holdings Limited Techniques for behavioral pairing in a contact center system
US9712676B1 (en) 2008-01-28 2017-07-18 Afiniti Europe Technologies Limited Techniques for benchmarking pairing strategies in a contact center system
US9781269B2 (en) 2008-01-28 2017-10-03 Afiniti Europe Technologies Limited Techniques for hybrid behavioral pairing in a contact center system
US9787841B2 (en) 2008-01-28 2017-10-10 Afiniti Europe Technologies Limited Techniques for hybrid behavioral pairing in a contact center system
US9930180B1 (en) 2017-04-28 2018-03-27 Afiniti, Ltd. Techniques for behavioral pairing in a contact center system
US10142473B1 (en) * 2016-06-08 2018-11-27 Afiniti Europe Technologies Limited Techniques for benchmarking performance in a contact center system
US20230031855A1 (en) * 2019-09-24 2023-02-02 Intradiem, Inc. Adaptive Rule Trigger Thresholds For Managing Contact Center Interaction Time

Patent Citations (7)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US9300802B1 (en) 2008-01-28 2016-03-29 Satmap International Holdings Limited Techniques for behavioral pairing in a contact center system
US9712676B1 (en) 2008-01-28 2017-07-18 Afiniti Europe Technologies Limited Techniques for benchmarking pairing strategies in a contact center system
US9781269B2 (en) 2008-01-28 2017-10-03 Afiniti Europe Technologies Limited Techniques for hybrid behavioral pairing in a contact center system
US9787841B2 (en) 2008-01-28 2017-10-10 Afiniti Europe Technologies Limited Techniques for hybrid behavioral pairing in a contact center system
US10142473B1 (en) * 2016-06-08 2018-11-27 Afiniti Europe Technologies Limited Techniques for benchmarking performance in a contact center system
US9930180B1 (en) 2017-04-28 2018-03-27 Afiniti, Ltd. Techniques for behavioral pairing in a contact center system
US20230031855A1 (en) * 2019-09-24 2023-02-02 Intradiem, Inc. Adaptive Rule Trigger Thresholds For Managing Contact Center Interaction Time

Similar Documents

Publication Publication Date Title
US11647119B2 (en) Techniques for behavioral pairing in a contact center system
US12075003B2 (en) Techniques for decisioning behavioral pairing in a task assignment system
JP2023126606A5 (fr)
JPWO2021055341A5 (fr)
WO2025191389A1 (fr) Capteur de compensation d'agent
EP4627498A1 (fr) Techniques de détection et de gestion de biais potentiel dans un système d'attribution de tâches
HK40076601A (en) Techniques for decisioning behavioral pairing in a task assignment system
HK40041789A (en) Techniques for decisioning behavioral pairing in a task assignment system
HK40073635B (zh) 用於联系中心系统中的行为配对的技术
HK40073639B (zh) 用於联系中心系统中的行为配对的技术
HK40079352B (zh) 用於联系中心系统中的行为配对的技术

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 25718040

Country of ref document: EP

Kind code of ref document: A1