Machine learning system, method, and computer program for evaluation of customer service agents

A method, comprising:

• collecting training data that indicates outcomes of prior interactions between specified customers and specified agents of a physical retail store, wherein the outcomes are collected from logs of the prior interactions made by the specified agents;
• training a machine learning model, using the training data, to infer, for different combinations of a given customer with each of a plurality of given agents of the physical retail store, an outcome expected with respect to the given agent communicating with the given customer to provide retail assistance within the physical retail store, wherein the outcome that the machine learning model is trained to infer is a monetary value selected from:
• a positive value for customer purchases,
• a negative value for customer refunds, and
• zero for no customer action;
• identifying presence of a customer at retail store;
• determining a plurality of agents of the physical retail store available to assist the customer in the physical retail store;
• for each agent of the plurality of agents available to assist the customer in the physical retail store, processing information describing the customer and information describing the agent, using the machine learning model, to determine an expected outcome of an interaction of the agent with the customer occurring within the physical retail store including a monetary value that is one of:
• a positive value for a customer purchase,
• a negative value for a customer refund, and
• zero for the customer taking no action;
• selecting one agent of the plurality of agents for assisting the customer in the physical retail store, based on the expected outcome determined for each agent of the plurality of agents;
• notifying the agent to begin interacting with the customer;
• recording the interaction of the agent with the customer, based on input to a point of sale system of the physical retail store;
• after the interaction of the agent of the physical retail store with the customer, determining an actual outcome of the interaction from the recorded interaction;
• evaluating the agent by comparing the actual outcome with the expected outcome determined for the agent;
• and
• using a result of the evaluation for assigning the agent to future customers.

A system, comprising:

• a non-transitory memory storing instructions;
• and
• one or more processors in communication with the non-transitory memory that execute the instructions to perform a method comprising:
• collecting training data that indicates outcomes of prior interactions between specified customers and specified agents of a physical retail store, wherein the outcomes are collected from logs of the prior interactions made by the specified agents;
• training a machine learning model, using the training data, to infer, for different combinations of a given customer with each of a plurality of given agents of the physical retail store, an outcome expected with respect to the given agent communicating with the given customer to provide retail assistance within the physical retail store, wherein the outcome that the machine learning model is trained to infer is a monetary value selected from:
• a positive value for customer purchases,
• a negative value for customer refunds, and
• zero for no customer action;
• identifying presence of a customer at retail store;
• determining a plurality of agents of the physical retail store available to assist the customer in the physical retail store;
• for each agent of the plurality of agents available to assist the customer in the physical retail store, processing information describing the customer and information describing the agent, using the machine learning model, to determine an expected outcome of an interaction of the agent with the customer occurring within the physical retail store including a monetary value that is one of:
• a positive value for a customer purchase,
• a negative value for a customer refund, and
• zero for the customer taking no action;
• selecting one agent of the plurality of agents for assisting the customer in the physical retail store, based on the expected outcome determined for each agent of the plurality of agents;
• notifying the agent to begin interacting with the customer;
• recording the interaction of the agent with the customer, based on input to a point of sale system of the physical retail store;
• after the interaction of the agent of the physical retail store with the customer, determining an actual outcome of the interaction from the recorded interaction;
• evaluating the agent by comparing the actual outcome with the expected outcome determined for the agent;
• and
• using a result of the evaluation for assigning the agent to future customers.