Method for predicting agent requirements in a force management system

• US 5,185,780 A
• Filed: 10/12/1990
• Issued: 02/09/1993
• Est. Priority Date: 10/12/1990
• Status: Expired due to Term

• Current Assignee: IEX Corporation (Nice Ltd)
• Original Assignee: IEX Corporation (Nice Ltd)
• Inventors: Leggett, Ernest W.
• Primary Examiner(s): SCHREYER, STAFFORD
• Application Number:
  US07/596,689
• Time in Patent Office: 851 Days
• Field of Search: 379/34, 379/133, 379/113, 379/134, 379/140
• US Class Current: 379/265.08
• CPC Class Codes: G06Q 10/06 Resources, workflows, human or project management; Enterprise or organisation planning; Enterprise or organisation modelling
• CPC Class Codes: H04M 3/36 Statistical metering, e.g. recording occasions when traffic exceeds capacity of trunks
• CPC Class Codes: H04M 3/523 with call distribution or queueing

Abstract

A method for predicting a number of agents required to provide a given service level in a force management system is described. The force management system has the capability to generate call handling performance data including average call arrival rate and average handling time per call. From this data, the method calculates an offered load "a" equal to the average arrival rate of calls for the system times the average handling time per call. Using the calculated offered load, the method calculates two predictor values and then uses these values in successive Erlang C calculations to locate the desired number of agents required to provide a given service level.

Claims

1. A method for managing personnel in a telephone call center in which there is a varying workload by time of day and by day of week to be staffed with a variable number of agents, comprising the steps of:

(a) generating call handling performance data for the telephone call center, the call handling performance data including average call arrival rate and average handling time per call;

(b) generating a forecast of a call load expected to occur at the telephone call center during a predetermined period in the future;

(c) defining a desired service level for the telephone call center during the predetermined period, the desired service level equal to a percentage of calls answered within a predetermined time "t";

(d) predicting a number of agents required to provide the desired service level during the predetermined period by;

(1) calculating an offered load "a" equal to the average arrival rate of calls for the call center times the average handling time per call;

(2) calculating PWtn using an Erlang C function C(n,a), for n=a+1, where PWtn is a percentage of calls that wait more than "t" seconds if there are "n" agents;

(3) if (100-PWtn) is not greater than or equal to the desired service level, repeating step (2) for PWtn1, where PWtn1 is a percentage of calls that wait more than "t" seconds if there are "n+1" agents;

(4) if (100-PWtn) is greater than or equal to the desired service level, define n as the predicted number of agents;

(5) if (100-PWtn1) is not greater than or equal to the desired level for n=n+1, calculate a drop percentage (PWtn-PWtn1)/PWtn;

(6) set an estimated percentage waiting time ePWt equal to PWtn1;

(7) calculate ePWt=ePWt-(ePWt*drop percentage) to determine how many increments of the drop percentage must be made until (100-ePWt) is greater than or equal to the desired service level;

(8) set a predictor value p equal to the determined number of increments plus (n+1);

(9) calculating Erlang C(n,a) for n=p-1, p and p+1;

(10) repeating steps (3)-(4) for successive values of n=p+1, p and p+1 until a value of n is located such that (100-PWtn) is greater than or equal to the desired service level; and

(e) allocating personnel in the telephone call center according to the predicted number of agents.

2. A method for managing personnel in a work center in which there is a varying workload by time of day and by day of week to be staffed with a variable number of workers, comprising the steps of:

(a) generating work request performance data for the work center, the work request performance data including average work request arrival rate and average handling time per work request;

(b) generating a forecast of a work request load expected to occur at the work request center during a predetermined period in the future;

(c) defining a desired service level for the work center during the predetermined period, the desired service level equal to a percentage of work requests answered within a predetermined time "t";

(d) predicting a number of agents required to provide the desired service level during the predetermined period by;

(1) calculating an offered load "a" equal to the average arrival rate of work requests for the work center times the average handling time per work request;

(2) calculating PWtn using an Erlang C function C(n,a), for n=a+1, where PWtn is a percentage of work requests that wait more than "t" seconds if there are "n" workers;

(3) if (100-PWtn) is not greater than or equal to the desired service level, repeating step (2) for PWtn1, where PWtn1 is a percentage of work requests that wait more than "t" seconds if there are "n+1" workers;

(4) if (100-PWtn) is greater than or equal to the desired service level, define n as the predicted number of workers;

(5) if (100-PWtn1) is not greater than or equal to the desired level for n=n+1, calculate a drop percentage (PWtn-PWtn1)/PWtn;

(6) set an estimated percentage waiting time ePWt equal to PWtn1;

(7) calculate ePWt=ePWt-(ePWt*drop percentage) to determine how many increments of the drop percentage must be made until (100-ePWt) is greater than or equal to the desired service level;

(8) set a predictor value p equal to the determined number of increments plus (n+1);

(9) calculating Erlang C(n,a) for n=p-1, p and p+1;

(10) repeating steps (3)-(4) for successive values of n=p+1, p and p+1 until a value of n is located such that (100-PWtn) is greater than or equal to the desired service level; and

(e) allocating personnel in the work center according to the predicted number of workers.

3. A method for managing personnel in a telephone call center in which there is a varying workload by time of day and by day of week to be staffed with a variable number of agents, comprising the steps of:

(a) generating a forecast of a call load expected to occur at the telephone call center during a predetermined period in the future;

(b) defining a desired service level for the telephone call center during the predetermined period, the desired service level equal to a percentage of calls answered within a predetermined time;

(c) predicting a number of agents required to provide the desired service level during the predetermined period by a direct calculation; and

(d) allocating personnel in the telephone call center according to the predicted number of agents.