Method and system for soft skills-based call routing in contact centers
An electronic communication method and system are disclosed. Exemplary methods include rating agent soft skills using an artificial intelligence (AI) module that continuously evaluates these skills, based on artifacts available from previous interactions with customers. The artifacts can be the voice recordings, chat transcripts, as well as Key Performance Indicators (KPIs) used for reporting. Once agents are rated, targeted soft skill-based routing is implemented for high priority calls or calls that are detected by a sentiment analyzer as requiring special attention. For training purposes, the system can be configured to route calls to agents with lower soft skills ratings during off hours. Completed calls may be used as further feedback to the AI module and the soft skill ratings acquired by the AI module may be added to an existing “hard skills” dataset for contact center call routing, to assist with continuous learning of soft skills as well as agent and supervisor training.
1. A system for soft skills-based call routing, the system comprising:
an interactive voice response module for receiving an incoming call;
a sentiment analyzer for detecting prosodic data in the incoming call indicative of a caller's emotional state;
an automatic call distributor module for routing the incoming call to an agent based on the prosodic data;
a database for storing artifacts indicative of agent soft skills based on all calls in which the agent participates; and
an artificial intelligence module that continuously updates a soft skills score of the agent based on all calls in which the agent participates, and after each call in which the agent participates the artificial intelligence module automatically communicates an updated soft skills score of the agent to the automatic call distributor module for routing a future incoming call, wherein the artificial intelligence module further includes (a) a soft skills scorer for receiving the artifacts and in response computes agent soft skills scores, and (b) an unsupervised learner module configured to receive the artifacts and in response adding and feeding one or more new parameters to the soft skills scorer.
2. The system of claim 1 , wherein the artifacts include one or more of voice recordings, chat transcripts, and key performance indicators.
3. The system of claim 1 , further comprising an overflow server for routing emergency incoming calls to the highest priority agent in cases of emergency by providing agents who are specially trained to handle situations that are deemed critical or for situations requiring specialized soft skills.
4. The system of claim 1 , wherein the agent soft skill scores are indicative of one or more of patience, empathy and timeliness.
5. The system of claim 1 , wherein the soft skills score of the agent are based at least in part on interaction evaluation data that includes customer call ratings.
6. The system of claim 1 , wherein the artificial intelligence module detects how said prosodic data changes from the start of the incoming call to the end of the incoming call to assist in generating a parameterized agent soft skill score for empathy.
7. The system of claim 1 , wherein the artificial intelligence module detects a duration of the incoming call to generate a parameterized agent soft skill score for timeliness.
8. The system of claim 1 , wherein the artifacts comprise one or more of: the agent's feedback of his/her performance and supervisor evaluation.
9. The system of claim 1 , wherein the one or more new parameters are one or more soft skills.
10. A method for soft skills-based call routing, the method comprising:
receiving an incoming call;
detecting prosodic data in the incoming call, wherein the prosodic data is indicative of a caller's emotional state;
routing the incoming call to an agent using (a) a regular routing, or (b) bypassing the regular routing and using a targeted routing to an agent with an appropriate soft skills score for the prosodic data of the caller detected by the sentiment analyzer;
storing artifacts indicative of agent soft skills based on all of the agent's completed calls; and
continuously classifying and weighting the artifacts from each of the agent's completed calls to automatically generate an updated soft skills score of the agent after each of the agent's completed calls and using the updated soft skills score of the agent to route a future call to the agent, wherein the artifacts are received by a soft skills scorer and by an unsupervised learner module and based on the received artifacts (a) the updated soft skills score is generated by the soft skills scorer, and (b) one or more new parameters are added to the soft skills scorer by the unsupervised learner module.
11. The method of claim 10 , wherein soft skill score is indicative of soft skills including one or more of patience, empathy and timeliness.
12. The method of claim 10 that further comprises the step of collecting at least one of interaction evaluation data and incidents of agent interruptions to generate a parameterized agent soft skill score for patience.
13. The method of claim 12 , wherein the interaction evaluation data includes customer call ratings.
14. The method of claim 10 that further comprises the step of detecting how said prosodic data changes from the start of the incoming call to the end of the incoming call to generate a parameterized agent soft skill score for empathy.
15. The method of claim 10 that further comprises the step of routing, during off hours, the incoming call to an agent with low soft skills.
16. The method of claim 10 that further comprises the step of routing emergency incoming calls to the highest priority agent in cases of emergency by providing agents who are specially trained to handle situations that are deemed critical or for situations requiring specialized soft skills.
17. The method of claim 10 , wherein a future call is routed to the agent, and artifact input from the future call is routed by the automatic call distributor to an artificial intelligence module, wherein the artificial intelligence module determines how much impact the updated soft skills score of the agent had on routing the future call.
18. The method of claim 10 , wherein the artifacts comprise one or more of: the agent's feedback of his/her performance and supervisor evaluation.
19. The method of claim 10 , wherein the one or more new parameters are one or more soft skills.
20. A non-transient computer readable medium comprising program instructions for causing a computer to perform the method of: processing one or more of digits, voice and text of an incoming call; detecting prosodic data in the incoming call, wherein the prosodic data is indicative of a caller's emotional state; routing the incoming call to an agent using (a) regular routing based on the one or more digits, voice and text, or (b) bypassing the regular routing and using targeted routing to an agent with a high soft skills rating appropriate for the prosodic data detected by a sentiment analyzer; storing in a database artifacts of all calls in which the agent participates, wherein the artifacts are indicative of agent hard skills and soft skills; using an artificial intelligence module to continuously update a soft skills score of the agent based on all calls in which the agent participates, and after each call in which the agent participates automatically communicating an updated soft skills score of the agent to an automatic call distributor module; and the automatic call distributor module routing a future call to the agent based on the updated soft skills score of the agent, and routing artifact input from the future call to the artificial intelligence module, wherein the artificial intelligence module determines how much impact the updated soft skill score of the agent had on routing the future call.