Estimating number of agents for multiple chat channels
View Patent ↗Method and system of estimating a number of agents to service contacts arriving on multiple chat channels in an automatic contact distribution system. In some embodiments, estimated average times that agents and contacts spend, respectively, composing a single chat message and an estimated average time of a chat session are used to derive an inflated handle time for any selected number of chat channels, an effective number of agents is computed for the inflated handle time, and that effective number of agents is divided by the selected number of channels.
1. A method of administering an automatic contact distribution (ACD) system, the method comprising:
estimating a minimum number of actual agents required to meet a specified quality of service goal for certain contact arrival information concerning contacts arriving on a selected number of chat channels in the ACD system, chat being real-time communication between one of the agents and one of the contacts via an electronic communication network, and the selected number of chat channels being a maximum number of contacts with whom a single agent will be engaged in chat at any one time, the estimating step comprising:
a) deriving an inflated handle time of chat sessions for the selected number of chat channels, each of the chat sessions being exchanges of chat messages, each of the chat messages being a unit of communication composed by a sender who triggers a send event at an end of a composition interval, where the unit of communication is sent in real-time to a recipient upon triggering of the send event, and the inflated handle time being a projected average duration of each of the chat sessions given the selected number of chat channels;
b) computing a number of effective agents using an Erlang calculator, the number of effective agents being a minimum number of agents that would be required to meet the specified quality of service goal given the contact arrival information and the inflated handle time, if it were assumed that each of the effective agents were servicing a single chat channel and that an average duration of chat sessions of the effective agents were the inflated handle time;
c) obtaining the number of actual agents by dividing the number of effective agents by the selected number of chat channels.
2. The method in accordance with claim 1 ,
the contact arrival information comprising a predicted contact arrival rate;
the quality of service goal comprising a specified probability that an arriving contact will be blocked from receiving service;
the Erlang calculator comprising an Erlang B calculator.
3. The method in accordance with claim 1 ,
the contact arrival information comprising a maximum length of a queue, a predicted contact arrival rate, and a predicted probability that a contact will balk at an offer of service if the queue is not empty;
the quality of service goal being selected from a group consisting of: a specified probability that an arriving contact will be blocked from receiving service, a specified average amount of time that a contact will spend waiting in the queue, and a specified probability that a contact will spend less than a specified amount of time waiting in the queue;
the Erlang calculator comprising an Erlang C calculator.
4. The method in accordance with claim 1 ,
the contact arrival information comprising a maximum length of a queue, a predicted contact arrival rate, a predicted probability that a contact will balk at an offer of service if the queue is not empty, and a predicted rate at which queued contacts will renege on the offer of service and will abandon the queue;
the quality of service goal being selected from a group consisting of: a specified probability that an arriving contact will be blocked from receiving service, a specified average amount of time that a contact will spend waiting in the queue, and a specified probability that a contact will spend less than a specified amount of time waiting in the queue;
the Erlang calculator comprising an Erlang Abandon calculator.
5. The method in accordance with claim 1 , the deriving step comprising using estimates of at least two numerical parameters that characterize contact/agent interactions.
6. The method in accordance with claim 1 , the deriving step comprising using an estimate of an average duration of chat sessions of agents servicing a chosen number of chat channels.
7. The method in accordance with claim 1 ,
the actual agents being in an agent category based on at least one factor selected from a group consisting of: communication subject matter, training in the communication subject matter, experience in the communication subject matter, training in chat messaging, experience in chat messaging, training in multi-channel chat messaging, and experience in multi-channel chat messaging;
the contacts being in a contact category based on at least one factor selected from a group consisting of: the communication subject matter, training in the communication subject matter, experience in the communication subject matter, training in chat messaging, and experience in chat messaging;
the deriving step comprising using an estimate of a ratio of an instantaneous rate at which the actual agents complete and send the chat messages they compose to an instantaneous rate at which the contacts complete and send the chat messages they compose, the ratio being a characteristic of a pair consisting of the agent category and the contact category.
8. The method in accordance with claim 1 , the deriving step comprising:
using an estimate of a ratio of an average length of time that the contacts spend composing a single chat message to an average length of time that the actual agents spend composing a single chat message, and
using an estimate of an average duration of chat sessions of agents servicing a chosen number of chat channels.
9. The method in accordance with claim 1 , the deriving step comprising using a formula for a probability that all of the contacts who are engaged in chat with a particular agent at any one time will be in a composition interval for a chat message, as opposed to any of the contacts who is engaged in chat with the particular agent being idle and waiting for a reply chat message from the particular agent.
10. The method in accordance with claim 1 , the deriving step comprising:
using an estimate of an average duration of chat sessions of agents servicing a chosen number of chat channels, where “m” represents the chosen number and “H[m]” represents the average duration of chat sessions for m channels;
using an estimate of a ratio of an average length of time that the contacts spend composing a single chat message to an average length of time that the actual agents spend composing a single chat message, where “r” represents the ratio;
computing the inflated handle time with the following formula:
H[m′]=H[m ]*( m′/m )*(1− P[m ](0))/(1− P[m ′](0)),
where “m′” represents the selected number of chat channels, where “H[m′]” represents the inflated handle time, and where for x=m and for x=m′, as applicable,
P[x ](0)=( r x /x !)*(1/(1+ r/ 1!+ r 2 /2!+ . . . + r x /x !)).
11. The method in accordance with claim 1 , further comprising:
repeating the estimating step for each one of a plurality of numbers of chat channels, each time substituting the respective one of the plurality of numbers of chat channels for the selected number of chat channels;
comparing the estimated minimum number of actual agents for each one of the plurality of numbers of chat channels to determine a change in the minimum number of actual agents required to meet the specified quality of service goal for each change in the number of chat channels.
12. The method in accordance with claim 11 , further comprising:
comparing the inflated handle time derived in the deriving step for each one of the plurality of numbers of chat channels to determine the change in the inflated handle time for each change in the number of chat channels;
for each change in the number of chat channels, comparing the corresponding change in the minimum number of actual agents to the corresponding change in the inflated handle time to determine a change in the inflated handle time for each change in the minimum number of actual agents.
13. The method in accordance with claim 1 , further comprising:
repeating the estimating step for each one of a plurality of quality of service goals for the selected number of chat channels, each time substituting the respective one of the plurality of quality of service goals for the specified quality of service goal;
identifying the quality of service goal for which the estimate of the required number of actual agents corresponds with an assigned number of actual agents to estimate a quality of service goal that can be sustained by the assigned number of actual agents.
14. The method in accordance with claim 13 , further comprising repeating the steps of claim 13 for each of a plurality of numbers of chat channels, each time substituting the respective one of the plurality of numbers of chat channels for the selected number of chat channels, to determine quality of service goals that can be sustained by the assigned number of actual agents for each of the plurality of numbers of chat channels.
15. A work management system to estimate a minimum number of actual agents required to meet a specified quality of service goal for certain contact arrival information concerning contacts arriving on a selected number of chat channels in an automatic contact distribution (ACD) system, chat being real-time communication between one of the agents and one of the contacts via an electronic communication network, and the selected number of chat channels being a maximum number of contacts with whom a single agent will be engaged in chat at any one time, the system comprising:
an inflated handle time module to derive an inflated handle time of chat sessions for the selected number of chat channels, each of the chat sessions being exchanges of chat messages, each of the chat messages being a unit of communication composed by a sender who triggers a send event at an end of a composition interval, where the unit of communication is sent in real-time to a recipient upon triggering of the send event, and the inflated handle time being a projected average duration of each of the chat sessions given the selected number of chat channels;
an Erlang calculator to compute a number of effective agents, the number of effective agents being a minimum number of agents that would be required to meet the specified quality of service goal given the contact arrival information and the inflated handle time, if it were assumed that each of the effective agents were servicing a single chat channel and that an average duration of chat sessions of the effective agents were the inflated handle time;
a quotient module to obtain the number of actual agents by dividing the number of effective agents by the selected number of chat channels.
16. The system in accordance with claim 15 ,
the contact arrival information comprising a predicted contact arrival rate;
the quality of service goal comprising a specified probability that an arriving contact will be blocked from receiving service;
the Erlang calculator comprising an Erlang B calculator.
17. The method in accordance with claim 15 ,
the contact arrival information comprising a maximum length of a queue, a predicted contact arrival rate, and a predicted probability that a contact will balk at an offer of service if the queue is not empty;
the quality of service goal being selected from a group consisting of: a specified probability that an arriving contact will be blocked from receiving service, a specified average amount of time that a contact will spend waiting in the queue, and a specified probability that a contact will spend less than a specified amount of time waiting in the queue;
the Erlang calculator comprising an Erlang C calculator.
18. The system in accordance with claim 15 ,
the contact arrival information comprising a maximum length of a queue, a predicted contact arrival rate, a predicted probability that a contact will balk at an offer of service if the queue is not empty, and a predicted rate at which queued contacts will renege on the offer of service and will abandon the queue;
the quality of service goal being selected from a group consisting of: a specified probability that an arriving contact will be blocked from receiving service, a specified average amount of time that a contact will spend waiting in the queue, and a specified probability that a contact will spend less than a specified amount of time waiting in the queue;
the Erlang calculator comprising an Erlang Abandon calculator.
19. The system in accordance with claim 15 , the inflated handle time module comprising at least two receivers to receive, respectively, estimates of at least two numerical parameters that characterize contact/agent interactions.
20. The system in accordance with claim 15 , the inflated handle time module comprising a receiver to receive an estimate of an average duration of chat sessions of agents servicing a chosen number of chat channels.
21. The system in accordance with claim 15 ,
the actual agents being in an agent category based on at least one factor selected from a group consisting of: communication subject matter, training in the communication subject matter, experience in the communication subject matter, training in chat messaging, experience in chat messaging, training in multi-channel chat messaging, and experience in multi-channel chat messaging;
the contacts being in a contact category based on at least one factor selected from a group consisting of: the communication subject matter, training in the communication subject matter, experience in the communication subject matter, training in chat messaging, and experience in chat messaging;
the inflated handle time module comprising a receiver to receive an estimate of a ratio of an instantaneous rate at which the actual agents complete and send the chat messages they compose to an instantaneous rate at which the contacts complete and send the chat messages they compose, the ratio being a characteristic of a pair consisting of the agent category and the contact category.
22. The system in accordance with claim 15 , the inflated handle time module comprising:
a first receiver to receive an estimate of an average duration of chat sessions of agents servicing a chosen number of chat channels;
a second receiver to receive an estimate of a ratio of an average length of time that the contacts spend composing a single chat message to an average length of time that the actual agents spend composing a single chat message.
23. The system in accordance with claim 15 , the inflated handle time module comprising a calculator to calculate a probability that all of the contacts who are engaged in chat with a particular agent at any one time will be in a composition interval for a chat message, as opposed to any of the contacts who is engaged in chat with the particular agent being idle and waiting for a reply chat message from the particular agent.
24. The system in accordance with claim 15 , the inflated handle time module comprising:
a first receiver to receive an estimate of an average duration of chat sessions of agents servicing a chosen number of chat channels, where “m” represents the chosen number and “H[m]” represents the average duration of chat sessions for m channels;
a second receiver to receive an estimate of a ratio of an average length of time that the contacts spend composing a single chat message to an average length of time that the actual agents spend composing a single chat message, where “r” represents the ratio;
an Erlang B calculator to calculate the inflated handle time with the following formula:
H[m′]=H[m ]*( m′/m )*(1− P[m ](0))/(1− P[m ′](0)),
where “m′” represents the selected number of chat channels, where “H[m′]” represents the inflated handle time, and where for x=m and for x=m′, as applicable,
P[x ](0)=( r x /x !)*(1/(1+ r/ 1!+ r 2 /2!+ . . . + r x /x !)).
25. A work management system to estimate a minimum number of actual agents required to meet a specified quality of service goal for certain contact arrival information concerning contacts arriving on a selected number of chat channels in an automatic contact distribution (ACD) system, chat being real-time communication between one of the agents and one of the contacts via an electronic communication network, and the selected number of chat channels being a maximum number of contacts with whom a single agent will be engaged in chat at any one time, the system comprising:
means for deriving an inflated handle time of chat sessions for the selected number of chat channels, each of the chat sessions being exchanges of chat messages, each of the chat messages being a unit of communication composed by a sender who triggers a send event at an end of a composition interval, where the unit of communication is sent in real-time to a recipient upon triggering of the send event, and the inflated handle time being a projected average duration of each of the chat sessions given the selected number of chat channels;
means for computing a number of effective agents, the number of effective agents being a minimum number of agents that would be required to meet the specified quality of service goal given the contact arrival information and the inflated handle time, if it were assumed that each of the effective agents were servicing a single chat channel and that an average duration of chat sessions of the effective agents were the inflated handle time;
means for obtaining the number of actual agents by dividing the number of effective agents by the selected number of chat channels.
26. The system in accordance with claim 25 ,
the contact arrival information comprising a predicted contact arrival rate;
the quality of service goal comprising a specified probability that an arriving contact will be blocked from receiving service.
27. The method in accordance with claim 25 ,
the contact arrival information comprising a maximum length of a queue, a predicted contact arrival rate, and a predicted probability that a contact will balk at an offer of service if the queue is not empty;
the quality of service goal being selected from a group consisting of: a specified probability that an arriving contact will be blocked from receiving service, a specified average amount of time that a contact will spend waiting in the queue, and a specified probability that a contact will spend less than a specified amount of time waiting in the queue.
28. The system in accordance with claim 25 ,
the contact arrival information comprising a maximum length of a queue, a predicted contact arrival rate, a predicted probability that a contact will balk at an offer of service if the queue is not empty, and a predicted rate at which queued contacts will renege on the offer of service and will abandon the queue;
the quality of service goal being selected from a group consisting of: a specified probability that an arriving contact will be blocked from receiving service, a specified average amount of time that a contact will spend waiting in the queue, and a specified probability that a contact will spend less than a specified amount of time waiting in the queue.
29. The system in accordance with claim 25 ,
the actual agents being in an agent category based on at least one factor selected from a group consisting of: communication subject matter, training in the communication subject matter, experience in the communication subject matter, training in chat messaging, experience in chat messaging, training in multi-channel chat messaging, and experience in multi-channel chat messaging;
the contacts being in a contact category based on at least one factor selected from a group consisting of: the communication subject matter, training in the communication subject matter, experience in the communication subject matter, training in chat messaging, and experience in chat messaging;
the inflated handle time deriving means comprising a means for receiving an estimate of a ratio of an instantaneous rate at which the actual agents complete and send the chat messages they compose to an instantaneous rate at which the contacts complete and send the chat messages they compose, the ratio being a characteristic of a pair consisting of the agent category and the contact category.
30. The system in accordance with claim 25 , the inflated handle time deriving means comprising a means for calculating a probability that all of the contacts who are engaged in chat with a particular agent at any one time will be in a composition interval for a chat message, as opposed to any of the contacts who is engaged in chat with the particular agent being idle and waiting for a reply chat message from the particular agent.