CA2249558C - Method and system for optimized scheduling - Google Patents
Method and system for optimized scheduling Download PDFInfo
- Publication number
- CA2249558C CA2249558C CA002249558A CA2249558A CA2249558C CA 2249558 C CA2249558 C CA 2249558C CA 002249558 A CA002249558 A CA 002249558A CA 2249558 A CA2249558 A CA 2249558A CA 2249558 C CA2249558 C CA 2249558C
- Authority
- CA
- Canada
- Prior art keywords
- contact
- attempt
- account
- value
- 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.)
- Expired - Fee Related
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
- H04M3/5158—Centralised call answering arrangements requiring operator intervention, e.g. call or contact centers for telemarketing in combination with automated outdialling systems
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR 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
- G06Q30/00—Commerce
- G06Q30/02—Marketing; Price estimation or determination; Fundraising
- G06Q30/0283—Price estimation or determination
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04M—TELEPHONIC COMMUNICATION
- H04M15/00—Arrangements for metering, time-control or time indication ; Metering, charging or billing arrangements for voice wireline or wireless communications, e.g. VoIP
Abstract
A method of optimizing the scheduling of tasks may be used to schedule telephone contact attempts through an automated dialing system (20). The method includes receiving an account data file comprising account informatio n for a plurality of accounts to be processed over a processing period. This processing period is divided into a plurality of contact attempt periods. Th e method also includes producing for each of the plurality of accounts an acti on result probability for each contact attempt period. The method also includes producing or determining a priority value for each of the accounts, determining resource costs for each contact attempt. These quantities or values, along with the action result probability, represent parameters over which the scheduling may be optimized.
Description
- METHOD AND SYSTEM FOR
OPTIMIZED SCHEDUllNG
BACKGROUND OF THE INVEN'rION
This invention relates to the sche~ ling of resources to perform a series of tasks.
Although the invention may be applied to many sih~atior~, it lias pal.icular appiication in u~ the sch~ ing of telephone contact attempts for collections or marketing purposes. The invention includes a method for optimi7ing the schP~1l1ing of resources and a co~ uL~r system for ~ rollllillg optimized schedllling.
Where a series of tasks are to be performed with relatively limited resources, the tasks must be scheduled to accommodate the resources available. For example, bank card issuers commonly attempt to contact delinquent accounts by telephone in order to inform the account holder of the delinquency and reach an agreement on payment. However, the number of delinquent accounts may extend into the many thousands, well beyond the number of collectors or operators available in a given period. The collection calls in this situation are therefore sch~lllecl in some manner with each collector h~nt11ing one call after another over the course of the day.
Another example of resource schl~dllling arises in telem~rketing where operatorscontact potential customers directly by telephone to make a sales solicitation. Since the targeted market may include thousands of potential customers, the telemarketing operators cannot reach all potential customers at the same time, but must make calls according to some sch~ ? to reach the desired potential cllstclm~r~.
Beyond direct telephone contacts for collections or tel~ .k~L~g purposes, there are many other situations in which l~uu~-~~ must be sche~lled to p~lru~ a number of tasks relatively larger than the amount of resources available for p~;lrO,l~ g the tasks at any one time. Specific collection actions may, for example, be scheduled for different accounts acco,.ling to the probability of success and the resources each is expected to utilize. In this case, the resources may include available collector time and the monetary budget for cont~ting accounts by mail.
At least in the telephone contact arena, technological advances have effectivelyincreased resources available by allowing an ~ dlol to contact more persons in a given time period. For example, automated dialing systems have greatly increased the efficiency of telephone contact operations. Allt ~m~flo~l dialing systems include telephone dialing and switching h~Ldw~le controlled or operated by a coln~uL~l which ~ccesses a ~l~t~h~e having a series of accounts or prospects to ~e contacted. The col~ L~l which conkols the automated dialing haldw~l~ causes the dialer to dial numbers associated with dirL; ~llL
S accounts or prospects according to some order. After dialing, if the colu~uL~l detects that a person has ~vv~ L~d the call, the automated dialing system quickly switches the call to an available operator. Also, the collll.ul~l transfers pertinent information to a terminal operated by the assigned operator to enable the operator to effectively c~ ic~te with the person reached. During the course of the call or immerli~t~ly after the call is ~ r~l, the u~el~Lor records the result of the contact at the operator's terminal and this information may be added to a historical tl~t~ha~e contzlining a contact history for each account or prospect. This rapid m~cllinP di~ling through the automated dialing system greatly increased the pace of telephone corltact operations and effectively increased the resources available for making contact by making the operators more productive.
More advanced ~lltc)m~t~1 dialing systems known as "predictive dialers" were developed to further increase operator productivity. Predictive dialers are a~ltom~tlod dialers which include some mecl~ ix", for directing calls at a rate clesi~nlod to keep each operator as busy as possible with ...il,i,,lll,,~ down time and without causing a contacted person to wait on hold for the next available operators.
Although these technological advances have allowed collectors or operator to be more productive, the number of collectors or operators available for h~nfiling telephone contacts at a given time is still much smaller than the total number of calls to be made.
Calls must still be h~nrilçd or sch~ led in some order according to collector or operator availability. This ordering or srhe~ ing of calls was initially random or arbitrary and was not opL.~ ed for any purpose.
U.S. Patent No. 5,436,965 was directed to a sch~ -le optimi7:;ng system for outbound telephone call camr~ign~. This system first sorted a set of accounts for teleproces~in~ according to relative priorities and then associated with each account a probability of cont~rt;ng an individual associated with the account. The prioritized accounts with probability of contact data added were then se~alaL~d into optimized c~mr~ign~ that created a chronological schedule of contact ~LL~llpL~. The contact ~Ll~ L~
for each account were scheduled at the time each targeted individual was most l~ely to CA 02249~8 1998-09-16 WO 97/35~1~; PCT/US97/04023 answer the telephone within the constraint of dialer resources available to make the call at a particular time.
Although the system disclosed in U.S. Patent No. 5,436,965 provided some schpr~ ing optimi7~tion for telephone contact operations, the disclosed system addressed S sch~ l1ing only in terms of contact probability. However, limitin~ the schedule to highest contact probability results in a sub-opLilllulll sch~-~-llin~ in terms of other factors that may be important in the particular telephone contact campaign. For example, assume there are two accounts, account A and account B, that are to be contacted in a bank card collections situation. Account A may be subst~nti~lly delinquent and therefore have a relatively high contact priority but a relatively low contact probability that is constant over a given time period. Account B may be only slightly delinquent and therefore have a much lower priority, and may have a relatively high probability of contact for a given time of day.
Given that scenario and given limited teleprocessing or operator availability, it may be more important or optimal to attempt to contact both accounts at times of the day that do not correspond to the respective highest contact probability. After all, the object of the telephone collections operation is not simply to make the m~rimllm number of contacts, but to maximize collections.
SUMMARY OP THE INVENTION
It is an object of the invention to overcome the above-described problems and others associated with prior s~h.o~ ling systems. In particular, it is an object of the invention to provide an a~L/dlatus and method of optimi7ing the allocation of resources to a series of tasks by scheclllling the tasks to be performed over a period of time. The o~ e~l allocation or s~h~ lling according to the invention may be applied in a number of situations, including telephone m~ g or telephone collections operations as well as direct or personal contact operations.
In order to accomplish these obiects, the c ~ ;on method and system according to the invention schedules or orders tasks not only in terms of highest contact probability but also in terms of relative priority, quantity of resources available for accomplishing the tasks, and the cost of an attempt at accomplishing eac~ task. Therefore, according to the invention, srhP~lllling is optimized in light of a total cost~benefit analysis, the cost of making an attempt at a certain time and the benefit of making the attempt at that time.
~=
CA 02249~8 1998-09-16 WO 97/354~5 PCT/US97/04023 The invention facilitates opt;mi7ing the ob.1ectives of performing the series of tasks rather than optimi7ing simply one component that affects the overall objectives.
For convenience, the invention will be rli~c--~se-l below primarily in terms of an outbound telephone call system for collection purposes. However, the invention is not S lirnited to the telephone collections application. The invention may be used to optimize outbound telephone calls for telemarketing rather than collections. Also, the invention may be employed to opLi~ e tasks other than telephone contact attempts. For example, the invention may be applied to optimize personal collections or sales. Those skilled in the art will readily appreciate that the invention may be applied to optimize sch~ 1ing in subst~nti~lly any situation calling for the sch~rlllling of a series of tasks which subst~nti~lly outnumber the resources available for performing the tasks at any one time.
In the telephone collections area, the method of the invention first includes receiving a number of accounts to be contacted over a processing period. The processing period is preferably a single day, and is further divided into a series of contact attempt 1~ periods preferably corresponding to the hours in the day in which collection calls can be made. In addition to receiving the accounts, the invention includes producing or defining parameters over which the sc~ ln~ing will be optimized. One parameter is an action result probability, which includes a right party contact probability in the telephone collections application. Another parameter may be referred to as a priority value, usually defined in terms of some characteristic or combination of characteristics of an account which make contact more or less important relative to other accounts. In any case, the priority value parameter defines a relative contact priority between accounts. Additional parameters over which the s~h~odnling will be o~Li~ ed are the ~ anliLy of collector or dialer resources available over the processing period, and the cost of each contact aLLe~ L. In the telephone collections arena, the cost of a telephone contact attempt is defined in terms of the operator or collector time required for h~n-lling the call.
After selecting the number of accounts to be processed in the day and defining the parameters over which the schedule will be optimi7~fl, the method includes the step of producing a relative contact attempt value for each attempt period and each account. This relative contact attempt value is preferably derived from the solution to a linear program m~cimi7.ing the product of the priority value and the probability of right party contact at a particular time for a particular account, sllmm~i for each account and each contact CA 02249~8 1998-09-16 period. There are three families of con~tr~int~ to the linear program. The first family of constraints is that the proportion of a particular accoun~ called at a particular time is greater than or equal to zero. The second family of C~ ld~lt, iS that the summedproportions of each account called at a particular time over the contact periods is less than or e~ual to one. The third and final family of constraints is that the total time necessary for making the contact aU~ t~ in a particular time period is less than or equal to the total collector time available in that particular time period.
After producing a contact attempt value for each account and each contact attempt period, the method includes making contact ~LIe~ Ls within each attempt period in order of descending contact attempt value. In the telephone collections operation, this step of making contact attempts is preferably p~lrol,lled through an automated dialing system.
The automated dialing system sorts accounts within each attempt period according to the contact attempt value and causes the dialer to dial the various accounts according to that sorting.
Finally, the ~lefell~d form of the invention for telephone collection operationsincludes, for each account, recording a contact attempt history. The contact attempt history is used to update parameters for o~tillli~,~Lion in future processing periods.
The step of producing the action result probability includes ~stim~tinp probabilities for three potential outcomes of the contact attempt. The three potential outcomes are right party contact, wrong party contact, and no connect, the latter l~lt;sellL~lg all colltillgellcies other than right and wrong party contacts. The right party contact probability, wrong party contact probability, and no connect probability sum to one. These contact probabilities may be produced by any suitable ~t~3ti~tiC~l modeling technique and preferably a step-wise polytomous logistic regression technique using contact attempt history data from previous processing periods and using account history data.
The contact attempt or resource cost values are defined for each potential outcome of a contact attempt and also may be produced using the contact attempt history data.
However, the contact attempt cost values may be produced by any suitable method and may, in fact, be fixed values provided by the user of the optirnized scheduling method.
The costs in the bank card collection application of the invention preferably are estim~teS
of handle times for contact alLelllpts resulting in a right party contact, a wrong party contact, and a no connect.
CA 02249~8 1998-09-16 WO 97/3~;41S PCT/US97/04~23 The priority value and qll~ntific~tion of contact attempt resources are both defined by the user of the optimized sch~ ling method. The quantity of contact attempt resources available is simply the number of operator hours available over the course of the processing period. The priority value may be developed by substantially any method to S rank relative priority of contact among the various accounts to be processed.
The ~l~;r~lled form of the invention also includes producing a forecast for eachprocessing period. The forecast in the telephone collections example preferably comprises expected values for the number of right party contacts, number of dialing attempts, and number of objectives met over the course of a processing period. The forecast may be compared with the actual results from the contact attempt historical data collected during the forecasted proces.~ing period to determine the effectiveness of the priority value definition.
After determining the effectiveness of the priority value and other parameters used to produce contact attempt values, the forecast may be used to make collector staffing decisions concerning the number of collectors used over future processing periods. The method may be used to produce forecast data using several different qn~ntiti~os for available collector hours. The forecasted results may inflir~t~ that the collector hours may be increased or reduced in certain contact attempt periods in order to optimize collection pel rollnance.
In addition to a method of optimi7ing sche-ll7ling, the invention also includes a colll~ul~l system for sche-lnling tasks. The co~ ul~l system includes an account storage device or account storage means to store data relating to the accounts to be processed and parameters used in the oL~Limi~tion process. The system also includes a downloadconn~octi~-n or download means for enabling the account storage device to receive an account data file co~ lg inroll"ation on the desired accounts to be processed. Ahistorical data storage device or storage means is included in the system as well as historical data upload means for enabling contact attempt history data to be transferred to the contact attempt data storage device.
The coni~,ule,- system further includes a user input or parameter input means enabling a user to input some of the optimi7~tion parameters used in the process. The user input is used to enter priority values or functions for defining priority values for each CA 02249~8 1998-09-16 account, the quantity of available resources, and contact attempt costs where such costs are not produced using historical data.
A first processor or first processor means within the c~ uLel system functions to produce the contact result probabilities and also the contact attempt values for each account and each contact attempt period over the proces~ing period. The contact attempt values are stored in the account storage device along with account information in an ~ignment data file. The ~ignmPnt data file is used to produce a modified download file which is transferred to a processor which sorts the accounts into descending contact ~LLe~ L values within each contact period. This processor may comprise a second processor or proces~in~
means associated with a separate computer used to control an ~ o~ dialing system.
Associated with the second processor is a hi~torir~l data storage device or means for collecting contact attempt history data derived from the contact ~LLtlll~L~ made under the control of the second processor. The historical data upload means effectively connects the first and second processors for enabling the collected contact attempt history data to be transferred to the historical data storage device.
These and other objects, advantages, and features of the invention will be apparent from the following description of the ~lcL~ d embo-liment~, considered along with the acco,l~pally ~ .lg drawings .
BRIEF DESC3~IPTION OF THE DRAWINGS
FIGURE I is a sçhP~ tic diagram of a colll~uLel embodying the principles of the invention for schefll-ling telephone contact attempts.
FIGURE 2 is a system flow chart for the telephone contact attempt system shown in FIGURE 1.
FIGURE 3 is a flow chart showing the steps for receiving account data file download.
FIGURE 4 is a flow chart showing the steps for producing contact probabilities.
FIGURE 5 is a flow chart showing the steps for receiving the contact attempt history data upload.
FIGURE 6 is a flow chart showing the steps for ~ llllhl.~illg the contact attempt history data.
CA 02249~8 l99X-09-16 WO 97/35415 PCT/US97/04~23 DESCRIPTION OF THE P~EFERRED EMBODIMENTS
A co~ uLer system 10 according to the invention is shown schem~ffr~lly in FIGURE 1. The co~ uL~l system 10 is connected through a download connection 11 to receive data from a separate billing and collections system, shown in FIGURE~ 1 as a separate colll~uleL system 12 having a processor 1~. In the form of the invention illuskated in FIGURE 1, the system 10 includes two separate processors, a contact probability and sch~ ling processor or first processor means 16, and a separate automated dialer processor or second processor means 18. The second processor 18 is a component of an ~nt ~m~ted dialing system 20 which also includes telephone connecting and switching hardware 22 and a data storage device 24. Pre~erably, the second processor 18 comprises a separate work station or microcomputer and the al-t-m~t~rl dialer storage is simply the mass storage device associated with the work station. An example of an automated dialer system 20 suitable for use in the system 10 is the MOSAIX SERIES 5000 system produced by Digital Systems Intern~tinn~l The first processor 16 has associated with it a user input or par~m~tt.r input means 26 and also mass storage devices 28 and 30. The ~ler~lled first processor 16 comprises a work station or microcomputer and the user input 26 and storage devices 28 and 30 are simply t'ne input and storage devices associated with the work station. Although separate data storage devices 28 and 30 are shown associated with the first processor 16, those skilled in the art will readily appreciate that a single storage device may be used rather than two. Separate data storage devices 28 and 30 are shown in FI~URE 1 to re~lect the fact that essentially two types of data are a~ces~efl by tne first processor 16, historical data and optimi~tion parameters or values, as will be discussed below.
Alt'nough FIGURE 1 illllstr~tf s one preferred implem~nt~tion of the co~ uLer 2~ system 10 according to the invention, those skilled in the art will readily appreciate that the system may be implemented in a number of different hardware configurations. For example, although the system 10 is shown as having two separate processors 16 and 18, a single processor may be util~zed to perform the functions performed by the two separate processors shown in the figure. In this alternative single processor form of the invention, the process tasks described below for the first and second processors 16 and 18 are simply performed by a single processor. Therefore a single processor implementation is to be considered an equivalent to the two processor system shown in FIGURE 1. Furthermore, CA 02249~8 1998-09-16 WO 97t35415 PCT/US97/04023 the invention may be implemented in a distributed c~ uLillg environment in whichnumerous processors are networked together and functions may be performed by several different processors within the distributed system, perhaps in parallel fashion. This distributed system allell~Live is also to be considered an equivalent to the two processor system shown in FI&URE 1.
Also, although the various storage devices shown in FIGURE 1 may be associated with a particular processor, and are preferably hard drives associated with the respective work station, the storage devices may be distributed over a n~Lwolk with the first and second processors, 16 and 18 respectively, having access to the storage devices over the network. Similarly, the account master file storage 32 shown in FIGURE 1 need not nf ces~rily be associated directly with the billing and collection processor 14, but may be located in a network ~cceccihle by the billing and collection processor. FIGURE 1 simply reflects the common implementation for telephone collections operations. Telephone collections operations normally use a billing and collections system implemented on a m~hlrldllle co~ uLel~, and an automated dialer system controlled by a separate work station. In this common situation, the ~lrst processor 16 according to the invention is interposed between the maiL~l~lle billing and collections system 12 and the automated dialer system 20 as shown in FIGURE 1.
The account master files contained in the account master file storage device 32 include all account data pertinent to billing and collections. For example, the account master files may include an account ID, names on the account, telephone numbers, a billing address, account balance, last payment, days delinquent or delinquency history, and purchase history. The system 10 according to the invention is adapted to receive from the account master files an account data file co.ll~i..i..~ data on a plurality of accounts to be processed through the system 10 over a certain proce~ing period, a single day for example. This processing comprises allocating dialer and collector resources to make contact attempts for the accounts according to a schedule C~Lillli~d to meet desired objectives. Thus, the system 10 operates to i~L~I~;epL the master file download that would otherwise go to an ~uLolllaLed dialer. The first processor 16 of system 10 preferably puts the received accounts in condition to be sorted and scheduled by the second processor 18 to optimize the chances of achieving the desired user-defined objectives. The automated dialer system 20 within the system 10 then directs the contact attempts or calls to accounts CA 02249~8 1998-09-16 through the telephone cormection haldw~le 22, and the associated automated dialer storage 24 collects historical information regarding each contact attempt for use by the first processor 16.
FIGURE 2 shows an overall system flow chart for the telephone collection system S 10 shown in FIGURE 1. Dashed rect~ngle 33 in FIGURE 2 shows the functions performed by the first processor 16. The final two steps 42 and 44 shown at the bottom of FIGURE 2 are performed by the second processor 18 shown in FIGURE 1.
Referring to both FIGURES 1 and 2 at the beginning of each day, or BOD shown in FIC;~URE 2, the first processor 16 receives an upload of contact attempt history data from the previous day's operation of the automated dialer system 20. The uploaded contact attempt history data comprises raw call attempt history data which the first processor 16 uses to update a call history data database within historical data storage device 28 associated with the first processor. This call history fl~t~h~e updating step is sho~vn at reference numeral 48 in FIGURE 2. The call history database m~int~in~qA within the storage device 28 includes for each account, account identifying information and information relating to the outcome of past contact aLL~ LJL~ for the respective account.
In addition to receiving the raw end-of-day data from the automated dialer or second processor 18, the first processor 16 also summarizes the contact attempt history data at step 50 to put the information in a form more usable by the system operator. The ~um~ tion of data may include the number of contact attempts, contacts, and promises made over a prece~ing period or periods. The system operator defines the manner in which data is ~ ~t;d at step 50 through the user input 26 associated with the first processor 16. The s lmm~rized contact history data is stored in a separate summary file within the call history data storage device 28 associated with the first processor 16.
At step 52 in the process shown in FIGURE 2, the first processor 16 receives theaccount data file preferably as a flat file downloaded from the master billing and collection system 12 through the data transfer connection or download means 11 and stores the account data file in the account storage device 30. After storing the account data file, the processor 16 operates to make any desired or n~ce~c~ly data format changes. As f~icclTcc~d above, the downloaded account data file includes information on plurality of accounts to be contacted within the processing period.
CA 02249~8 1998-09-16 WO 97/3541~; PCT/US97/04023 Il After receiving the-account data file download from the account master file at step 52, the first processor 16 produces an action result probability for each account and for each contact period as will be ~ c~l~se(l in detail below with reference to FIGURE 4 and as shown generally in FIGURE 2 as step 54. The action resuIt probability for each S account and for each contact period is an important factor in optimi7:ing the srh~ llin~ of contact attempts between the various accounts. However, action result probability is just one factor considered according to the invention.
Other important factors considered in the sch~ ling optimi7~ion process of the invention include the resources available for making contact attempts at a given time and a relative contact priority value of attempting to contact a particular account. Among these factors or parameters, the resources available may simply be entered at step 56 into the storage device 30 through the user input device 26 associated with the first processor 16.
The contact priority value for each account may be based upon a single characteristic such as the dollar amount an account is delinquent or may be derived from a more complex function using a number of dirr~lellL elements that vary from account to account. An example of a more complex function for ~leterrnining priority value could be the amount ~lelinqll~nt, number of days delinquent, and number of delinquencies in the past year multiplied together and divided by the amount paid by the account over the previous year, or even the result of another modelling effort. In any event, the priority value for each account is d~lell~ ed using the function or value entered by the user at step 56 through the user input 26 associated with the first processor 16.
In addition to relative contact priorit.v values for each account to be processed and the resources available for making contact ~llelll~Ls, the system and method of the invention schedules contact ~Ll~ L~ based also upon the cost of making a contact attempt.
In the telephone collections example of the invention, the cost of making a call attempt may be defined in terms of collector time. Since the time required in m~king a contact attempt via telephone is dependent upon the result of the attempt, the resource cost is defined in terms of multiple values. l~or example, where the action result probability includes probabilities of contact divided into right party contact probability and all other potential outcomes, two values of resource costs are required, that is the time cost of making a right party contact and the time cost of all other potential outcomes. In the preferred form of the invention the probabilities are defined as right party contact , CA 02249~8 1998-09-16 probability, wrong party contact probability, and all other potential outcomes or the no connect probability. In this case the resource costs will be defined by three different values, one value represf~nting the time required for a right party contact, another value for the time required for a wrong party contact, and a third value for al~ other results.
S These cost values may be defined by the user or entered through the input device 26 at step 56, or may be averaged or otherwise derived from call attempt history data.With the accounts to be processed defined by the account data file download and the other sch~lllin~ parameters determined or calclll~t~, the method of the invention includes the step 58 of producing a contact attempt value for each account and for each contact period within the processin~ period. The contact attempt value represents a relative desirability of making a contact attempt for a particular account in a particular contact attempt period. This contact attempt value for each contact attempt period is then associated with its respective account to produce an ~si~nm~q-nt data file from which is developed a modified download for the auLollla~d dialer system 20 at step 60. The modified download may comprise a file that includes account information for a plurality of accounts, and for each account a function including contact attempt value for each contact attempt period. ~ i vely, the modified download may comprise a file c-~nt~ining account information fior a plurality of accounts and information on the optimal contact attempt period to call for each account, and preferably, the contact attempt value associated with the optimal contact attempt period for each account. Although the modified download may be form~tte~l in a number of different ways, the contact aU~ pt value associated with each account for each contact attempt period allows the accounts to be sorted or in-l~xed into descending order of contact attempt value for each contact period. This descending order of contact attempt value is preferably produced with the second processor 18 at step ~2 shown in FIGURE 2, and ~ ,ellLs the sch~ ile of contact aLLe~ ~ for each contact period. Of course where the modified download includes orlly account hlrollllalion and the op~ lulll contact attempt period to call for each account, the second processor simply groups the accounts for calls during their respective optimal contact attempt period, and the sortin~ is thus accomplished using both the first processor 16 and second processor 18.
After sorting, the second processor 18 causes the dialer hardware 22 to call theaccounts according to the descending contact attempt value schedule. For each contact 700-768 PCT P~ 4 0 ~ 3 ~ r-2~ n ~ ~T ~7 attempt or call the second processor 18 also collects contact attempt history data relating to the contact attempt. This contact attempt history data may include the time the attempt is made, the result of the attempt, the duration of the call, and payment amount promised or other in~ tors such as "hardship" for example. The contact attempt history S information is collected at step 44 in FIGURE 2 and stored in the au~ulllal~d dialer storage device 24. associated with the second processor 18. The illÇo~ alion is later uploaded or L d~r~ d to the first processor and associated storage 28 at the beginning of the following day or other processing period as shown in FIGURE 2.
The step 58 in FI~URE % of pr.oducing a contact attempt value f(3r each account --'fO . and each contact penod pr~ferably in'volves maximi7ing.over all accoun.ts and all contact-~ periods the .rel'ative accQunt . priority value, probab~lity of right party contact, and . :' . proportion of a particuiar account cal'led at a particular.ti~ne. The basic problern can'be c,~ ssed in terrns of the linear prograrn:
)' ' -max; r E' -Cjtpj,-Xjt '-- ~ ~ -- ' ' ' ' ~ '- '' - ~ ~ -~ '- -xt i t . with the co.~ .;..l*
Vi,t: xk > 0 vi: ~xit ~ 1 ~-~" ~t: ~(tm (q~) + tc (pj,) + tn (rj~)) x~t < ht where i = account 1N
t = contact period 1.. T
Xit = ~ Lion of aç~ led at time t ' Pit = right-party contact probability at tirne t for account i qit = wrong-party contact probability at time t for account i rh = no co~L ~r~ability at tirne t for accuuu~ i (ri, = 1 - q~t ~ Pit) tc = average time of contact attempt that resulted in right-party contact trn = average time of contact attempt that resulted in wrong-par~f contact tn = ~ a~,~, time of contact a~)~ that resulted in no co ht = total collector time available at time t ck = priority value for accoulll i at time t.
~ t . ~
CA 02249~8 1998-09-16 The nature of the solution to this linear program guarantees that a solution exists that is essenti~lly integer, either 0 or 1. The linear program can be solved with a general solution process, for example simplex, an interior point method, or preferably using a ~agrangian function.
S As shown at step 62 in FIGURE 2, the pl~re"~d form of the invention includes producing forecasted results. The forecasted results include a value for the expected number of contacts and a value for expected objectives met for each contact period and the total. The expected number of contacts are calculated by sllmm;ng the probabilities of right party contact for scheduled calls, that is calls that will be scheduled according to the modified download produced in step 60. The value for expected ob3ectives met is calculated by sllmming the product of the probability of right party contact and the priority value for scheduled calls.
FIGURE 3 shows the ~r~r~ d process steps included in receiving the master file download shown at step 52 in FIGURE 2. TmmP~ tely upon reading the master file download, the data is preferably lefu~ d at step 64 to put all information in a standardized form for further proces~in~. Dates are l~ple~,ellLed as a number of days since a certain date, times are represented as seconds since mitlni~ht, and numerical representations are represented as integers or real numbers. These format standardization changes simply f~cilit~tP the further manipulations required by the invention.
Once the format changes are accompIished, the preferred form of the invention includes at step 65 in FIGURE 3, merging the reform~ttPcl download data file with information from the s-lmmzlrized account activity file created at step 50 shown in FIGURE 2. This file merger places ~ lla~ d call history data in a single file with the reform~tted download data to facilitate the calculation of action result probability at step 54 in FIGURE 2.
Referring still to FIGUR~ 3, the ~l~fel,ed form of the invention also gives the user the option of segrnenting the download population at step 66. Thus, the user may segment or group accounts which it wishes to collect differently. For example, new or marginally delinquent accounts may be separated from seriously delinquent accounts which are best handled by the most experienced collectors. At step 67, the user may also generate account characteristics such as days since last payment to be added to the merged file.
These characteristics may be used later in producing the contact priority value as shown CA 02249~8 1998-09-16 1~5 at step 56 in FIC:;URE 2 or the right-party contact, wrong-party contact and no connect probability estimates shown at step 72 FIGURE 4.
FICi~URE 4 shows the ~lcrclled procedure for producing the action result probabiIity according to the invention. The first step 70 in the process is to designate contact attempt periods within the overall processing period. In collections and marketing operations, the processing period may be a single day and the contact attempt periods may be defined as the sixteen hours in the day available for making telephone contact a~
However, those skilled in the art will readily appreciate that the proce~ing period need not be any particular length of time and the contact attempt periods may be substantially any convenient subdivision of the overall processing period. Also, for applications of the invention outside the telephone collection area, periods of a larger proce~ing period may not be used at all. Rather than time being the variant and the contact aLLt;~ L periods representing permissible variation of the time variant, optimi7ation may be with respect to some other variant. For example, in an application of the invention to sales, the dirr~lcll~ products may be defined rather than contact periods. In this sale example the product lcplc~ellL~7 the variant and the different products the permissible variation.
Furthermore, in the sales example, the outcomes or action results could be defined as acc~L~llce, reJection of offer, and "other." In the telephone collections application the action result probabilities are preferably defined as right party contact probability, wrong party contact probability, and no connect or all other results probability.
After defining the contact attempt periods, the accounts to be processed must begrouped or ~si~n~l to probability models or scorecards at step 72 in FI~URE 4. Each model or scorecard colll~lises a set of probability score values. The user defines con(lition~ which are used in a~ nin~ accounts to a particular scorecard. DirrelenL
?5 sco~calds are used for subsets of accounts which behave dirr~lcllLly. That is, ~;lirrele attributes enter the scorecards or models with varying weights.
Once the accounts are ~ign~c1 to probability models, the models may be built or rebuilt at step 74 as n~cec~ry. The user defines the time intervals at which the models are rebuilt using the additional historical data collected from the last time the models were rebuilt. The rebuild interval may range from infinity, that is, never rebuild, to preferably not less than one or two months for the telephone collections application.
_ - = =
CA 02249~8 1998-09-16 The probability models or scorecards may be developed by any suitable method but are preferably developed by applying a step-wise polytomous logistic regression technique. ~ccording to the ~le~ll~d process, three distinct categories of outcome are identified, right party contacts (group Gl), wrong party contacts (group G2), and no S connect or "other" (group G3). A random sample Cont~ining all three categories is drawn from the call history ~l~t~h~se. Given a random sample X = (~1~ x2, through xn) where x~ is an account in the sample (an m-dimensional vector whose components are thec~n~ lRte characteristics), the likelihood function for a pair of m--lim~nsional vectors a and b is defined as follows:
L(a,b) = II exp(aT x i)/(l ~exp(aT x j) ~exp(bT x j)) x i~Gl II exp(b x i)/(l +exp(a x i)+exp(bT x j)) x i~G2 II 1/(1 ~exp(a x i) +exp(b x ;)) i~G3 "a" is the right party contact score, and "b" is the wrong party contact score.
The modelling process involves a series of steps in which characteristics are added to the models. At each step of the modelling process the adjusted chi-squared statistics for all of the variables for either of the two possible scores not in the model are computed and ex~minf tl If the largest of these statistics is signifir~nt at a defined level, the variable with this largest ~ t~l chi-squared statistic is entered into the model. Once a variable is entered into a model it is never removed from the model. This stepwise process of e..lP. ;I~g variables into the models is continued until none of the rem~ining variables meet the defined si ni~icallce test, in the preferred case 99%. After the models or scorecards are constructed, the intercepts are adjusted to reflect the true proportions of the three ~ aLg~)~ies, rather than those represented in the sample.
Once the probability models or scorecards are rebuilt or updated, model parameters are produced in terms of a series of characteristics associated with a weighting factor or score. The probability values are then produced at step 76 in FIGURE 4 during the master file download by m~tchin~ characteristics from the downloaded accounts to the model CA 02249~8 l998-09-l6 parameters and then converting the rF~.slllt,.nt model value to a probability between zero and one.
Although contact probabilities may be defined as a right party contact probability and a non-right party contact probability, the p~ d form of the invention defines three probabilities for three potential contact attempt results. Those preferred probabilities comprise right party contact probability, wrong party contact probability, and no connect probability, the latter representing the probability of all results other than right or wrong party contact.
FIGURE 5 shows the l.lerell~d steps of receiving or uploading the end of day contact attempt history data from the previous day's automated dialer files and updating the call attempt history database contained in the historical data storage device 28 associated with the first processor 16. The first step 80 comprises reading the dialer activity file and ch~n3~ing the file format as required to produce standardized data similar to the format changes in the master file download at step 64 in FIGURE 3. The reforn~ttin~ step produces standardized dates and times, collveliillg to local time where the dialer is in another time zone.~ The dialer activity data is then standardized at step 84 to simplify future use or manipulation of the data. For example, the dialer activity data may be grouped into categories corresponding to the possible call outcomes. At step 86 the user then identifies the data to store perrnanently and stores the identified data to the call history ~l~t~b~ce within a storage device 28. The user det~ s what data to keep and how long to use it.
Referring now to FIGURE 6, the raw data collected at the dialer processor 18 is preferably sllmm~rized to put the data in a form more useful in the overall process shown in FIGURE 2. The sllmm~ri7~tion process begins with reading dialer activity from the call history data storage at step 88 in FIGU~E 6. The data has already been refo~n~ttecl and standardized as discussed in connection with FIGURE 5. The retrieved data is then used to gel~ldLe .sllmm~ry characteristics at 90 such as the number of calls made within a certain period. The user controls how the data is sumlnarized. The surnmarized data is then stored in a database at step 92 for later use.
The above described pl~felled embo~iment~ are intl~n-1ecl to illustrate the principles of the invention, but not to limit the scope of the invention. Various other embodiments CA 02249558 l998-09-l6 WO 971354tS PCT/US97/04023 and mo~ifi~tions to these ~ r~ d embodiments may be made by those skilled in the art without departing from the scope of the following claims.
OPTIMIZED SCHEDUllNG
BACKGROUND OF THE INVEN'rION
This invention relates to the sche~ ling of resources to perform a series of tasks.
Although the invention may be applied to many sih~atior~, it lias pal.icular appiication in u~ the sch~ ing of telephone contact attempts for collections or marketing purposes. The invention includes a method for optimi7ing the schP~1l1ing of resources and a co~ uL~r system for ~ rollllillg optimized schedllling.
Where a series of tasks are to be performed with relatively limited resources, the tasks must be scheduled to accommodate the resources available. For example, bank card issuers commonly attempt to contact delinquent accounts by telephone in order to inform the account holder of the delinquency and reach an agreement on payment. However, the number of delinquent accounts may extend into the many thousands, well beyond the number of collectors or operators available in a given period. The collection calls in this situation are therefore sch~lllecl in some manner with each collector h~nt11ing one call after another over the course of the day.
Another example of resource schl~dllling arises in telem~rketing where operatorscontact potential customers directly by telephone to make a sales solicitation. Since the targeted market may include thousands of potential customers, the telemarketing operators cannot reach all potential customers at the same time, but must make calls according to some sch~ ? to reach the desired potential cllstclm~r~.
Beyond direct telephone contacts for collections or tel~ .k~L~g purposes, there are many other situations in which l~uu~-~~ must be sche~lled to p~lru~ a number of tasks relatively larger than the amount of resources available for p~;lrO,l~ g the tasks at any one time. Specific collection actions may, for example, be scheduled for different accounts acco,.ling to the probability of success and the resources each is expected to utilize. In this case, the resources may include available collector time and the monetary budget for cont~ting accounts by mail.
At least in the telephone contact arena, technological advances have effectivelyincreased resources available by allowing an ~ dlol to contact more persons in a given time period. For example, automated dialing systems have greatly increased the efficiency of telephone contact operations. Allt ~m~flo~l dialing systems include telephone dialing and switching h~Ldw~le controlled or operated by a coln~uL~l which ~ccesses a ~l~t~h~e having a series of accounts or prospects to ~e contacted. The col~ L~l which conkols the automated dialing haldw~l~ causes the dialer to dial numbers associated with dirL; ~llL
S accounts or prospects according to some order. After dialing, if the colu~uL~l detects that a person has ~vv~ L~d the call, the automated dialing system quickly switches the call to an available operator. Also, the collll.ul~l transfers pertinent information to a terminal operated by the assigned operator to enable the operator to effectively c~ ic~te with the person reached. During the course of the call or immerli~t~ly after the call is ~ r~l, the u~el~Lor records the result of the contact at the operator's terminal and this information may be added to a historical tl~t~ha~e contzlining a contact history for each account or prospect. This rapid m~cllinP di~ling through the automated dialing system greatly increased the pace of telephone corltact operations and effectively increased the resources available for making contact by making the operators more productive.
More advanced ~lltc)m~t~1 dialing systems known as "predictive dialers" were developed to further increase operator productivity. Predictive dialers are a~ltom~tlod dialers which include some mecl~ ix", for directing calls at a rate clesi~nlod to keep each operator as busy as possible with ...il,i,,lll,,~ down time and without causing a contacted person to wait on hold for the next available operators.
Although these technological advances have allowed collectors or operator to be more productive, the number of collectors or operators available for h~nfiling telephone contacts at a given time is still much smaller than the total number of calls to be made.
Calls must still be h~nrilçd or sch~ led in some order according to collector or operator availability. This ordering or srhe~ ing of calls was initially random or arbitrary and was not opL.~ ed for any purpose.
U.S. Patent No. 5,436,965 was directed to a sch~ -le optimi7:;ng system for outbound telephone call camr~ign~. This system first sorted a set of accounts for teleproces~in~ according to relative priorities and then associated with each account a probability of cont~rt;ng an individual associated with the account. The prioritized accounts with probability of contact data added were then se~alaL~d into optimized c~mr~ign~ that created a chronological schedule of contact ~LL~llpL~. The contact ~Ll~ L~
for each account were scheduled at the time each targeted individual was most l~ely to CA 02249~8 1998-09-16 WO 97/35~1~; PCT/US97/04023 answer the telephone within the constraint of dialer resources available to make the call at a particular time.
Although the system disclosed in U.S. Patent No. 5,436,965 provided some schpr~ ing optimi7~tion for telephone contact operations, the disclosed system addressed S sch~ l1ing only in terms of contact probability. However, limitin~ the schedule to highest contact probability results in a sub-opLilllulll sch~-~-llin~ in terms of other factors that may be important in the particular telephone contact campaign. For example, assume there are two accounts, account A and account B, that are to be contacted in a bank card collections situation. Account A may be subst~nti~lly delinquent and therefore have a relatively high contact priority but a relatively low contact probability that is constant over a given time period. Account B may be only slightly delinquent and therefore have a much lower priority, and may have a relatively high probability of contact for a given time of day.
Given that scenario and given limited teleprocessing or operator availability, it may be more important or optimal to attempt to contact both accounts at times of the day that do not correspond to the respective highest contact probability. After all, the object of the telephone collections operation is not simply to make the m~rimllm number of contacts, but to maximize collections.
SUMMARY OP THE INVENTION
It is an object of the invention to overcome the above-described problems and others associated with prior s~h.o~ ling systems. In particular, it is an object of the invention to provide an a~L/dlatus and method of optimi7ing the allocation of resources to a series of tasks by scheclllling the tasks to be performed over a period of time. The o~ e~l allocation or s~h~ lling according to the invention may be applied in a number of situations, including telephone m~ g or telephone collections operations as well as direct or personal contact operations.
In order to accomplish these obiects, the c ~ ;on method and system according to the invention schedules or orders tasks not only in terms of highest contact probability but also in terms of relative priority, quantity of resources available for accomplishing the tasks, and the cost of an attempt at accomplishing eac~ task. Therefore, according to the invention, srhP~lllling is optimized in light of a total cost~benefit analysis, the cost of making an attempt at a certain time and the benefit of making the attempt at that time.
~=
CA 02249~8 1998-09-16 WO 97/354~5 PCT/US97/04023 The invention facilitates opt;mi7ing the ob.1ectives of performing the series of tasks rather than optimi7ing simply one component that affects the overall objectives.
For convenience, the invention will be rli~c--~se-l below primarily in terms of an outbound telephone call system for collection purposes. However, the invention is not S lirnited to the telephone collections application. The invention may be used to optimize outbound telephone calls for telemarketing rather than collections. Also, the invention may be employed to opLi~ e tasks other than telephone contact attempts. For example, the invention may be applied to optimize personal collections or sales. Those skilled in the art will readily appreciate that the invention may be applied to optimize sch~ 1ing in subst~nti~lly any situation calling for the sch~rlllling of a series of tasks which subst~nti~lly outnumber the resources available for performing the tasks at any one time.
In the telephone collections area, the method of the invention first includes receiving a number of accounts to be contacted over a processing period. The processing period is preferably a single day, and is further divided into a series of contact attempt 1~ periods preferably corresponding to the hours in the day in which collection calls can be made. In addition to receiving the accounts, the invention includes producing or defining parameters over which the sc~ ln~ing will be optimized. One parameter is an action result probability, which includes a right party contact probability in the telephone collections application. Another parameter may be referred to as a priority value, usually defined in terms of some characteristic or combination of characteristics of an account which make contact more or less important relative to other accounts. In any case, the priority value parameter defines a relative contact priority between accounts. Additional parameters over which the s~h~odnling will be o~Li~ ed are the ~ anliLy of collector or dialer resources available over the processing period, and the cost of each contact aLLe~ L. In the telephone collections arena, the cost of a telephone contact attempt is defined in terms of the operator or collector time required for h~n-lling the call.
After selecting the number of accounts to be processed in the day and defining the parameters over which the schedule will be optimi7~fl, the method includes the step of producing a relative contact attempt value for each attempt period and each account. This relative contact attempt value is preferably derived from the solution to a linear program m~cimi7.ing the product of the priority value and the probability of right party contact at a particular time for a particular account, sllmm~i for each account and each contact CA 02249~8 1998-09-16 period. There are three families of con~tr~int~ to the linear program. The first family of constraints is that the proportion of a particular accoun~ called at a particular time is greater than or equal to zero. The second family of C~ ld~lt, iS that the summedproportions of each account called at a particular time over the contact periods is less than or e~ual to one. The third and final family of constraints is that the total time necessary for making the contact aU~ t~ in a particular time period is less than or equal to the total collector time available in that particular time period.
After producing a contact attempt value for each account and each contact attempt period, the method includes making contact ~LIe~ Ls within each attempt period in order of descending contact attempt value. In the telephone collections operation, this step of making contact attempts is preferably p~lrol,lled through an automated dialing system.
The automated dialing system sorts accounts within each attempt period according to the contact attempt value and causes the dialer to dial the various accounts according to that sorting.
Finally, the ~lefell~d form of the invention for telephone collection operationsincludes, for each account, recording a contact attempt history. The contact attempt history is used to update parameters for o~tillli~,~Lion in future processing periods.
The step of producing the action result probability includes ~stim~tinp probabilities for three potential outcomes of the contact attempt. The three potential outcomes are right party contact, wrong party contact, and no connect, the latter l~lt;sellL~lg all colltillgellcies other than right and wrong party contacts. The right party contact probability, wrong party contact probability, and no connect probability sum to one. These contact probabilities may be produced by any suitable ~t~3ti~tiC~l modeling technique and preferably a step-wise polytomous logistic regression technique using contact attempt history data from previous processing periods and using account history data.
The contact attempt or resource cost values are defined for each potential outcome of a contact attempt and also may be produced using the contact attempt history data.
However, the contact attempt cost values may be produced by any suitable method and may, in fact, be fixed values provided by the user of the optirnized scheduling method.
The costs in the bank card collection application of the invention preferably are estim~teS
of handle times for contact alLelllpts resulting in a right party contact, a wrong party contact, and a no connect.
CA 02249~8 1998-09-16 WO 97/3~;41S PCT/US97/04~23 The priority value and qll~ntific~tion of contact attempt resources are both defined by the user of the optimized sch~ ling method. The quantity of contact attempt resources available is simply the number of operator hours available over the course of the processing period. The priority value may be developed by substantially any method to S rank relative priority of contact among the various accounts to be processed.
The ~l~;r~lled form of the invention also includes producing a forecast for eachprocessing period. The forecast in the telephone collections example preferably comprises expected values for the number of right party contacts, number of dialing attempts, and number of objectives met over the course of a processing period. The forecast may be compared with the actual results from the contact attempt historical data collected during the forecasted proces.~ing period to determine the effectiveness of the priority value definition.
After determining the effectiveness of the priority value and other parameters used to produce contact attempt values, the forecast may be used to make collector staffing decisions concerning the number of collectors used over future processing periods. The method may be used to produce forecast data using several different qn~ntiti~os for available collector hours. The forecasted results may inflir~t~ that the collector hours may be increased or reduced in certain contact attempt periods in order to optimize collection pel rollnance.
In addition to a method of optimi7ing sche-ll7ling, the invention also includes a colll~ul~l system for sche-lnling tasks. The co~ ul~l system includes an account storage device or account storage means to store data relating to the accounts to be processed and parameters used in the oL~Limi~tion process. The system also includes a downloadconn~octi~-n or download means for enabling the account storage device to receive an account data file co~ lg inroll"ation on the desired accounts to be processed. Ahistorical data storage device or storage means is included in the system as well as historical data upload means for enabling contact attempt history data to be transferred to the contact attempt data storage device.
The coni~,ule,- system further includes a user input or parameter input means enabling a user to input some of the optimi7~tion parameters used in the process. The user input is used to enter priority values or functions for defining priority values for each CA 02249~8 1998-09-16 account, the quantity of available resources, and contact attempt costs where such costs are not produced using historical data.
A first processor or first processor means within the c~ uLel system functions to produce the contact result probabilities and also the contact attempt values for each account and each contact attempt period over the proces~ing period. The contact attempt values are stored in the account storage device along with account information in an ~ignment data file. The ~ignmPnt data file is used to produce a modified download file which is transferred to a processor which sorts the accounts into descending contact ~LLe~ L values within each contact period. This processor may comprise a second processor or proces~in~
means associated with a separate computer used to control an ~ o~ dialing system.
Associated with the second processor is a hi~torir~l data storage device or means for collecting contact attempt history data derived from the contact ~LLtlll~L~ made under the control of the second processor. The historical data upload means effectively connects the first and second processors for enabling the collected contact attempt history data to be transferred to the historical data storage device.
These and other objects, advantages, and features of the invention will be apparent from the following description of the ~lcL~ d embo-liment~, considered along with the acco,l~pally ~ .lg drawings .
BRIEF DESC3~IPTION OF THE DRAWINGS
FIGURE I is a sçhP~ tic diagram of a colll~uLel embodying the principles of the invention for schefll-ling telephone contact attempts.
FIGURE 2 is a system flow chart for the telephone contact attempt system shown in FIGURE 1.
FIGURE 3 is a flow chart showing the steps for receiving account data file download.
FIGURE 4 is a flow chart showing the steps for producing contact probabilities.
FIGURE 5 is a flow chart showing the steps for receiving the contact attempt history data upload.
FIGURE 6 is a flow chart showing the steps for ~ llllhl.~illg the contact attempt history data.
CA 02249~8 l99X-09-16 WO 97/35415 PCT/US97/04~23 DESCRIPTION OF THE P~EFERRED EMBODIMENTS
A co~ uLer system 10 according to the invention is shown schem~ffr~lly in FIGURE 1. The co~ uL~l system 10 is connected through a download connection 11 to receive data from a separate billing and collections system, shown in FIGURE~ 1 as a separate colll~uleL system 12 having a processor 1~. In the form of the invention illuskated in FIGURE 1, the system 10 includes two separate processors, a contact probability and sch~ ling processor or first processor means 16, and a separate automated dialer processor or second processor means 18. The second processor 18 is a component of an ~nt ~m~ted dialing system 20 which also includes telephone connecting and switching hardware 22 and a data storage device 24. Pre~erably, the second processor 18 comprises a separate work station or microcomputer and the al-t-m~t~rl dialer storage is simply the mass storage device associated with the work station. An example of an automated dialer system 20 suitable for use in the system 10 is the MOSAIX SERIES 5000 system produced by Digital Systems Intern~tinn~l The first processor 16 has associated with it a user input or par~m~tt.r input means 26 and also mass storage devices 28 and 30. The ~ler~lled first processor 16 comprises a work station or microcomputer and the user input 26 and storage devices 28 and 30 are simply t'ne input and storage devices associated with the work station. Although separate data storage devices 28 and 30 are shown associated with the first processor 16, those skilled in the art will readily appreciate that a single storage device may be used rather than two. Separate data storage devices 28 and 30 are shown in FI~URE 1 to re~lect the fact that essentially two types of data are a~ces~efl by tne first processor 16, historical data and optimi~tion parameters or values, as will be discussed below.
Alt'nough FIGURE 1 illllstr~tf s one preferred implem~nt~tion of the co~ uLer 2~ system 10 according to the invention, those skilled in the art will readily appreciate that the system may be implemented in a number of different hardware configurations. For example, although the system 10 is shown as having two separate processors 16 and 18, a single processor may be util~zed to perform the functions performed by the two separate processors shown in the figure. In this alternative single processor form of the invention, the process tasks described below for the first and second processors 16 and 18 are simply performed by a single processor. Therefore a single processor implementation is to be considered an equivalent to the two processor system shown in FIGURE 1. Furthermore, CA 02249~8 1998-09-16 WO 97t35415 PCT/US97/04023 the invention may be implemented in a distributed c~ uLillg environment in whichnumerous processors are networked together and functions may be performed by several different processors within the distributed system, perhaps in parallel fashion. This distributed system allell~Live is also to be considered an equivalent to the two processor system shown in FI&URE 1.
Also, although the various storage devices shown in FIGURE 1 may be associated with a particular processor, and are preferably hard drives associated with the respective work station, the storage devices may be distributed over a n~Lwolk with the first and second processors, 16 and 18 respectively, having access to the storage devices over the network. Similarly, the account master file storage 32 shown in FIGURE 1 need not nf ces~rily be associated directly with the billing and collection processor 14, but may be located in a network ~cceccihle by the billing and collection processor. FIGURE 1 simply reflects the common implementation for telephone collections operations. Telephone collections operations normally use a billing and collections system implemented on a m~hlrldllle co~ uLel~, and an automated dialer system controlled by a separate work station. In this common situation, the ~lrst processor 16 according to the invention is interposed between the maiL~l~lle billing and collections system 12 and the automated dialer system 20 as shown in FIGURE 1.
The account master files contained in the account master file storage device 32 include all account data pertinent to billing and collections. For example, the account master files may include an account ID, names on the account, telephone numbers, a billing address, account balance, last payment, days delinquent or delinquency history, and purchase history. The system 10 according to the invention is adapted to receive from the account master files an account data file co.ll~i..i..~ data on a plurality of accounts to be processed through the system 10 over a certain proce~ing period, a single day for example. This processing comprises allocating dialer and collector resources to make contact attempts for the accounts according to a schedule C~Lillli~d to meet desired objectives. Thus, the system 10 operates to i~L~I~;epL the master file download that would otherwise go to an ~uLolllaLed dialer. The first processor 16 of system 10 preferably puts the received accounts in condition to be sorted and scheduled by the second processor 18 to optimize the chances of achieving the desired user-defined objectives. The automated dialer system 20 within the system 10 then directs the contact attempts or calls to accounts CA 02249~8 1998-09-16 through the telephone cormection haldw~le 22, and the associated automated dialer storage 24 collects historical information regarding each contact attempt for use by the first processor 16.
FIGURE 2 shows an overall system flow chart for the telephone collection system S 10 shown in FIGURE 1. Dashed rect~ngle 33 in FIGURE 2 shows the functions performed by the first processor 16. The final two steps 42 and 44 shown at the bottom of FIGURE 2 are performed by the second processor 18 shown in FIGURE 1.
Referring to both FIGURES 1 and 2 at the beginning of each day, or BOD shown in FIC;~URE 2, the first processor 16 receives an upload of contact attempt history data from the previous day's operation of the automated dialer system 20. The uploaded contact attempt history data comprises raw call attempt history data which the first processor 16 uses to update a call history data database within historical data storage device 28 associated with the first processor. This call history fl~t~h~e updating step is sho~vn at reference numeral 48 in FIGURE 2. The call history database m~int~in~qA within the storage device 28 includes for each account, account identifying information and information relating to the outcome of past contact aLL~ LJL~ for the respective account.
In addition to receiving the raw end-of-day data from the automated dialer or second processor 18, the first processor 16 also summarizes the contact attempt history data at step 50 to put the information in a form more usable by the system operator. The ~um~ tion of data may include the number of contact attempts, contacts, and promises made over a prece~ing period or periods. The system operator defines the manner in which data is ~ ~t;d at step 50 through the user input 26 associated with the first processor 16. The s lmm~rized contact history data is stored in a separate summary file within the call history data storage device 28 associated with the first processor 16.
At step 52 in the process shown in FIGURE 2, the first processor 16 receives theaccount data file preferably as a flat file downloaded from the master billing and collection system 12 through the data transfer connection or download means 11 and stores the account data file in the account storage device 30. After storing the account data file, the processor 16 operates to make any desired or n~ce~c~ly data format changes. As f~icclTcc~d above, the downloaded account data file includes information on plurality of accounts to be contacted within the processing period.
CA 02249~8 1998-09-16 WO 97/3541~; PCT/US97/04023 Il After receiving the-account data file download from the account master file at step 52, the first processor 16 produces an action result probability for each account and for each contact period as will be ~ c~l~se(l in detail below with reference to FIGURE 4 and as shown generally in FIGURE 2 as step 54. The action resuIt probability for each S account and for each contact period is an important factor in optimi7:ing the srh~ llin~ of contact attempts between the various accounts. However, action result probability is just one factor considered according to the invention.
Other important factors considered in the sch~ ling optimi7~ion process of the invention include the resources available for making contact attempts at a given time and a relative contact priority value of attempting to contact a particular account. Among these factors or parameters, the resources available may simply be entered at step 56 into the storage device 30 through the user input device 26 associated with the first processor 16.
The contact priority value for each account may be based upon a single characteristic such as the dollar amount an account is delinquent or may be derived from a more complex function using a number of dirr~lellL elements that vary from account to account. An example of a more complex function for ~leterrnining priority value could be the amount ~lelinqll~nt, number of days delinquent, and number of delinquencies in the past year multiplied together and divided by the amount paid by the account over the previous year, or even the result of another modelling effort. In any event, the priority value for each account is d~lell~ ed using the function or value entered by the user at step 56 through the user input 26 associated with the first processor 16.
In addition to relative contact priorit.v values for each account to be processed and the resources available for making contact ~llelll~Ls, the system and method of the invention schedules contact ~Ll~ L~ based also upon the cost of making a contact attempt.
In the telephone collections example of the invention, the cost of making a call attempt may be defined in terms of collector time. Since the time required in m~king a contact attempt via telephone is dependent upon the result of the attempt, the resource cost is defined in terms of multiple values. l~or example, where the action result probability includes probabilities of contact divided into right party contact probability and all other potential outcomes, two values of resource costs are required, that is the time cost of making a right party contact and the time cost of all other potential outcomes. In the preferred form of the invention the probabilities are defined as right party contact , CA 02249~8 1998-09-16 probability, wrong party contact probability, and all other potential outcomes or the no connect probability. In this case the resource costs will be defined by three different values, one value represf~nting the time required for a right party contact, another value for the time required for a wrong party contact, and a third value for al~ other results.
S These cost values may be defined by the user or entered through the input device 26 at step 56, or may be averaged or otherwise derived from call attempt history data.With the accounts to be processed defined by the account data file download and the other sch~lllin~ parameters determined or calclll~t~, the method of the invention includes the step 58 of producing a contact attempt value for each account and for each contact period within the processin~ period. The contact attempt value represents a relative desirability of making a contact attempt for a particular account in a particular contact attempt period. This contact attempt value for each contact attempt period is then associated with its respective account to produce an ~si~nm~q-nt data file from which is developed a modified download for the auLollla~d dialer system 20 at step 60. The modified download may comprise a file that includes account information for a plurality of accounts, and for each account a function including contact attempt value for each contact attempt period. ~ i vely, the modified download may comprise a file c-~nt~ining account information fior a plurality of accounts and information on the optimal contact attempt period to call for each account, and preferably, the contact attempt value associated with the optimal contact attempt period for each account. Although the modified download may be form~tte~l in a number of different ways, the contact aU~ pt value associated with each account for each contact attempt period allows the accounts to be sorted or in-l~xed into descending order of contact attempt value for each contact period. This descending order of contact attempt value is preferably produced with the second processor 18 at step ~2 shown in FIGURE 2, and ~ ,ellLs the sch~ ile of contact aLLe~ ~ for each contact period. Of course where the modified download includes orlly account hlrollllalion and the op~ lulll contact attempt period to call for each account, the second processor simply groups the accounts for calls during their respective optimal contact attempt period, and the sortin~ is thus accomplished using both the first processor 16 and second processor 18.
After sorting, the second processor 18 causes the dialer hardware 22 to call theaccounts according to the descending contact attempt value schedule. For each contact 700-768 PCT P~ 4 0 ~ 3 ~ r-2~ n ~ ~T ~7 attempt or call the second processor 18 also collects contact attempt history data relating to the contact attempt. This contact attempt history data may include the time the attempt is made, the result of the attempt, the duration of the call, and payment amount promised or other in~ tors such as "hardship" for example. The contact attempt history S information is collected at step 44 in FIGURE 2 and stored in the au~ulllal~d dialer storage device 24. associated with the second processor 18. The illÇo~ alion is later uploaded or L d~r~ d to the first processor and associated storage 28 at the beginning of the following day or other processing period as shown in FIGURE 2.
The step 58 in FI~URE % of pr.oducing a contact attempt value f(3r each account --'fO . and each contact penod pr~ferably in'volves maximi7ing.over all accoun.ts and all contact-~ periods the .rel'ative accQunt . priority value, probab~lity of right party contact, and . :' . proportion of a particuiar account cal'led at a particular.ti~ne. The basic problern can'be c,~ ssed in terrns of the linear prograrn:
)' ' -max; r E' -Cjtpj,-Xjt '-- ~ ~ -- ' ' ' ' ~ '- '' - ~ ~ -~ '- -xt i t . with the co.~ .;..l*
Vi,t: xk > 0 vi: ~xit ~ 1 ~-~" ~t: ~(tm (q~) + tc (pj,) + tn (rj~)) x~t < ht where i = account 1N
t = contact period 1.. T
Xit = ~ Lion of aç~ led at time t ' Pit = right-party contact probability at tirne t for account i qit = wrong-party contact probability at time t for account i rh = no co~L ~r~ability at tirne t for accuuu~ i (ri, = 1 - q~t ~ Pit) tc = average time of contact attempt that resulted in right-party contact trn = average time of contact attempt that resulted in wrong-par~f contact tn = ~ a~,~, time of contact a~)~ that resulted in no co ht = total collector time available at time t ck = priority value for accoulll i at time t.
~ t . ~
CA 02249~8 1998-09-16 The nature of the solution to this linear program guarantees that a solution exists that is essenti~lly integer, either 0 or 1. The linear program can be solved with a general solution process, for example simplex, an interior point method, or preferably using a ~agrangian function.
S As shown at step 62 in FIGURE 2, the pl~re"~d form of the invention includes producing forecasted results. The forecasted results include a value for the expected number of contacts and a value for expected objectives met for each contact period and the total. The expected number of contacts are calculated by sllmm;ng the probabilities of right party contact for scheduled calls, that is calls that will be scheduled according to the modified download produced in step 60. The value for expected ob3ectives met is calculated by sllmming the product of the probability of right party contact and the priority value for scheduled calls.
FIGURE 3 shows the ~r~r~ d process steps included in receiving the master file download shown at step 52 in FIGURE 2. TmmP~ tely upon reading the master file download, the data is preferably lefu~ d at step 64 to put all information in a standardized form for further proces~in~. Dates are l~ple~,ellLed as a number of days since a certain date, times are represented as seconds since mitlni~ht, and numerical representations are represented as integers or real numbers. These format standardization changes simply f~cilit~tP the further manipulations required by the invention.
Once the format changes are accompIished, the preferred form of the invention includes at step 65 in FIGURE 3, merging the reform~ttPcl download data file with information from the s-lmmzlrized account activity file created at step 50 shown in FIGURE 2. This file merger places ~ lla~ d call history data in a single file with the reform~tted download data to facilitate the calculation of action result probability at step 54 in FIGURE 2.
Referring still to FIGUR~ 3, the ~l~fel,ed form of the invention also gives the user the option of segrnenting the download population at step 66. Thus, the user may segment or group accounts which it wishes to collect differently. For example, new or marginally delinquent accounts may be separated from seriously delinquent accounts which are best handled by the most experienced collectors. At step 67, the user may also generate account characteristics such as days since last payment to be added to the merged file.
These characteristics may be used later in producing the contact priority value as shown CA 02249~8 1998-09-16 1~5 at step 56 in FIC:;URE 2 or the right-party contact, wrong-party contact and no connect probability estimates shown at step 72 FIGURE 4.
FICi~URE 4 shows the ~lcrclled procedure for producing the action result probabiIity according to the invention. The first step 70 in the process is to designate contact attempt periods within the overall processing period. In collections and marketing operations, the processing period may be a single day and the contact attempt periods may be defined as the sixteen hours in the day available for making telephone contact a~
However, those skilled in the art will readily appreciate that the proce~ing period need not be any particular length of time and the contact attempt periods may be substantially any convenient subdivision of the overall processing period. Also, for applications of the invention outside the telephone collection area, periods of a larger proce~ing period may not be used at all. Rather than time being the variant and the contact aLLt;~ L periods representing permissible variation of the time variant, optimi7ation may be with respect to some other variant. For example, in an application of the invention to sales, the dirr~lcll~ products may be defined rather than contact periods. In this sale example the product lcplc~ellL~7 the variant and the different products the permissible variation.
Furthermore, in the sales example, the outcomes or action results could be defined as acc~L~llce, reJection of offer, and "other." In the telephone collections application the action result probabilities are preferably defined as right party contact probability, wrong party contact probability, and no connect or all other results probability.
After defining the contact attempt periods, the accounts to be processed must begrouped or ~si~n~l to probability models or scorecards at step 72 in FI~URE 4. Each model or scorecard colll~lises a set of probability score values. The user defines con(lition~ which are used in a~ nin~ accounts to a particular scorecard. DirrelenL
?5 sco~calds are used for subsets of accounts which behave dirr~lcllLly. That is, ~;lirrele attributes enter the scorecards or models with varying weights.
Once the accounts are ~ign~c1 to probability models, the models may be built or rebuilt at step 74 as n~cec~ry. The user defines the time intervals at which the models are rebuilt using the additional historical data collected from the last time the models were rebuilt. The rebuild interval may range from infinity, that is, never rebuild, to preferably not less than one or two months for the telephone collections application.
_ - = =
CA 02249~8 1998-09-16 The probability models or scorecards may be developed by any suitable method but are preferably developed by applying a step-wise polytomous logistic regression technique. ~ccording to the ~le~ll~d process, three distinct categories of outcome are identified, right party contacts (group Gl), wrong party contacts (group G2), and no S connect or "other" (group G3). A random sample Cont~ining all three categories is drawn from the call history ~l~t~h~se. Given a random sample X = (~1~ x2, through xn) where x~ is an account in the sample (an m-dimensional vector whose components are thec~n~ lRte characteristics), the likelihood function for a pair of m--lim~nsional vectors a and b is defined as follows:
L(a,b) = II exp(aT x i)/(l ~exp(aT x j) ~exp(bT x j)) x i~Gl II exp(b x i)/(l +exp(a x i)+exp(bT x j)) x i~G2 II 1/(1 ~exp(a x i) +exp(b x ;)) i~G3 "a" is the right party contact score, and "b" is the wrong party contact score.
The modelling process involves a series of steps in which characteristics are added to the models. At each step of the modelling process the adjusted chi-squared statistics for all of the variables for either of the two possible scores not in the model are computed and ex~minf tl If the largest of these statistics is signifir~nt at a defined level, the variable with this largest ~ t~l chi-squared statistic is entered into the model. Once a variable is entered into a model it is never removed from the model. This stepwise process of e..lP. ;I~g variables into the models is continued until none of the rem~ining variables meet the defined si ni~icallce test, in the preferred case 99%. After the models or scorecards are constructed, the intercepts are adjusted to reflect the true proportions of the three ~ aLg~)~ies, rather than those represented in the sample.
Once the probability models or scorecards are rebuilt or updated, model parameters are produced in terms of a series of characteristics associated with a weighting factor or score. The probability values are then produced at step 76 in FIGURE 4 during the master file download by m~tchin~ characteristics from the downloaded accounts to the model CA 02249~8 l998-09-l6 parameters and then converting the rF~.slllt,.nt model value to a probability between zero and one.
Although contact probabilities may be defined as a right party contact probability and a non-right party contact probability, the p~ d form of the invention defines three probabilities for three potential contact attempt results. Those preferred probabilities comprise right party contact probability, wrong party contact probability, and no connect probability, the latter representing the probability of all results other than right or wrong party contact.
FIGURE 5 shows the l.lerell~d steps of receiving or uploading the end of day contact attempt history data from the previous day's automated dialer files and updating the call attempt history database contained in the historical data storage device 28 associated with the first processor 16. The first step 80 comprises reading the dialer activity file and ch~n3~ing the file format as required to produce standardized data similar to the format changes in the master file download at step 64 in FIGURE 3. The reforn~ttin~ step produces standardized dates and times, collveliillg to local time where the dialer is in another time zone.~ The dialer activity data is then standardized at step 84 to simplify future use or manipulation of the data. For example, the dialer activity data may be grouped into categories corresponding to the possible call outcomes. At step 86 the user then identifies the data to store perrnanently and stores the identified data to the call history ~l~t~b~ce within a storage device 28. The user det~ s what data to keep and how long to use it.
Referring now to FIGURE 6, the raw data collected at the dialer processor 18 is preferably sllmm~rized to put the data in a form more useful in the overall process shown in FIGURE 2. The sllmm~ri7~tion process begins with reading dialer activity from the call history data storage at step 88 in FIGU~E 6. The data has already been refo~n~ttecl and standardized as discussed in connection with FIGURE 5. The retrieved data is then used to gel~ldLe .sllmm~ry characteristics at 90 such as the number of calls made within a certain period. The user controls how the data is sumlnarized. The surnmarized data is then stored in a database at step 92 for later use.
The above described pl~felled embo~iment~ are intl~n-1ecl to illustrate the principles of the invention, but not to limit the scope of the invention. Various other embodiments CA 02249558 l998-09-l6 WO 971354tS PCT/US97/04023 and mo~ifi~tions to these ~ r~ d embodiments may be made by those skilled in the art without departing from the scope of the following claims.
Claims (20)
1. A method of optimizing the scheduling of telephone contact attempts through an automated dialing system, the method comprising the steps of:
(a) receiving an account data file comprising account information for a plurality of accounts to be processed over a processing period, the processing period being divided into a plurality of contact attempt periods;
(b) for each of the plurality of accounts, producing an action result probability for each contact period;
(c) producing a priority value for each of the plurality of accounts;
(d) determining a quantity of contact resources available over the processing period;
(e) determining resource costs for each contact attempt;
(f) for each account, producing a contact attempt value for each contact period, each contact attempt value comprising a numerical value representing a relative desirability of attempting to contact the respective account during the respective contact period considering the priority value fox each account, the action result probability for each account, the quantity of contact resources available over the processing period, and the resource costs for each contact attempt;
(g) within each contact attempt period, making contact attempts for the plurality of accounts in order of descending contact attempt value; and (h) for each account, collecting contact attempt history data.
(a) receiving an account data file comprising account information for a plurality of accounts to be processed over a processing period, the processing period being divided into a plurality of contact attempt periods;
(b) for each of the plurality of accounts, producing an action result probability for each contact period;
(c) producing a priority value for each of the plurality of accounts;
(d) determining a quantity of contact resources available over the processing period;
(e) determining resource costs for each contact attempt;
(f) for each account, producing a contact attempt value for each contact period, each contact attempt value comprising a numerical value representing a relative desirability of attempting to contact the respective account during the respective contact period considering the priority value fox each account, the action result probability for each account, the quantity of contact resources available over the processing period, and the resource costs for each contact attempt;
(g) within each contact attempt period, making contact attempts for the plurality of accounts in order of descending contact attempt value; and (h) for each account, collecting contact attempt history data.
2. ~The method of claim 1 further including the step of:
(a) producing for each contact attempt period an expected right party contact value representing a number of expected right party contacts for the respective contact attempt period.
(a) producing for each contact attempt period an expected right party contact value representing a number of expected right party contacts for the respective contact attempt period.
3. ~The method of claim 1 further including the step of:
(a) producing for each contact attempt period an expected right party contact value, an expected contact attempt quantity, and an expected objective met value.
(a) producing for each contact attempt period an expected right party contact value, an expected contact attempt quantity, and an expected objective met value.
4. ~The method of claim 1 wherein the step of producing an action result probability comprises:
(a) for each contact period, creating a set of probability score values for at least one user defined population using a polytomous logistic regression method based upon the account history for a representative sample for the plurality of accounts; and (b) for each set of probability score values, applying the respective set of probability score values to each account matching the user defined population for the respective probability score values.
(a) for each contact period, creating a set of probability score values for at least one user defined population using a polytomous logistic regression method based upon the account history for a representative sample for the plurality of accounts; and (b) for each set of probability score values, applying the respective set of probability score values to each account matching the user defined population for the respective probability score values.
5. ~The method of claim 4 wherein the step of creating a set of probability score values for at least one user defined population is performed at user determined intervals.
6. The method of claim 1 wherein the step of producing an action result probability comprises producing for each account and each contact attempt period a right party contact probability, a wrong party contact probability, and a no connect probability.
7. ~The method of claim 6 wherein the resource costs for each contact attempt include a cost of right party contact, a cost of wrong party contact, and a cost of no connect.
8. ~The method of claim 7 wherein the step of producing a contact attempt value for each account and each contact attempt period comprises:
(a) ~solving the linear program max .SIGMA. .SIGMA. c it p it x it x it i t with constraints x it >= 0 .SIGMA.x it >= 1 t .SIGMA.(tm (q it) + tc (p it) + tn (r it)) x it >= h t i to produce a solved value x it for each account and each contact period;
where i ~= account 1...N
t ~= contact period 1...T
x it~= proportion of account i called at time t p it~= right-party contact probability at time t for account i q it~= wrong-party contact probability at time t for account i r it~= no connect probability at time t for account i (r it = 1 - q it - p it) tc ~= average time of contact attempt that resulted in right party contact tm ~= average time of contact attempt that resulted in wrong party contact tn ~= average time of contact attempt that resulted in no connect h t ~= total collector time available at time t c it = priority value for account i at time t; and (b) associating the solved value x it with its respective account and contact attempt period.
(a) ~solving the linear program max .SIGMA. .SIGMA. c it p it x it x it i t with constraints x it >= 0 .SIGMA.x it >= 1 t .SIGMA.(tm (q it) + tc (p it) + tn (r it)) x it >= h t i to produce a solved value x it for each account and each contact period;
where i ~= account 1...N
t ~= contact period 1...T
x it~= proportion of account i called at time t p it~= right-party contact probability at time t for account i q it~= wrong-party contact probability at time t for account i r it~= no connect probability at time t for account i (r it = 1 - q it - p it) tc ~= average time of contact attempt that resulted in right party contact tm ~= average time of contact attempt that resulted in wrong party contact tn ~= average time of contact attempt that resulted in no connect h t ~= total collector time available at time t c it = priority value for account i at time t; and (b) associating the solved value x it with its respective account and contact attempt period.
9. ~The method of claim 1 further comprising the step of sorting the plurality of accounts into descending order of contact attempt value with the automated dialer system.
10. A computer system for optimizing the scheduling of telephone contact attempts over a processing period which is divided into a plurality of contact attempt periods, the computer system comprising:
(a)~ account storage means for storing:
an account data file comprising account information for a plurality of accounts to be processed over the processing period, a priority value for each account, a quantity of contact resources available over the processing period, resource cost values associated with each contact attempt, a plurality of sets of probability score values, and as assignment data file comprising account information for the plurality of accounts and for each account a contact attempt value for each contact period;
(b) master file download means for enabling the account storage means to receive the account data file;
(c) historical data storage means for storing contact attempt history data for each of the plurality of accounts;
(d) parameter input means responsive to operator commands for enabling an operator to input priority data, and the quantity of contact resources available over the processing period;
(e) first processing means coupled to the account storage means and historical data storage means for producing for each account an action result probability for each contact period, each action result probability being produced using one of the sets of probability score values; and producing the contact attempt value for each account and each contact period, the contact attempt value comprising a numerical value representing a relative desirability of attempting to contact the respective account during the respective contact attempt period considering the priority value for each account, the action result probability for each account, the quantity of contact resources available over the processing period, and the resource cost for each contact attempt;
(f) modified download output means for transferring to an automated dialer system a modified download data file comprising account information for each account and contact attempt value information for each account and each contact attempt period, the contact attempt value information enabling the plurality of accounts to be sorted in descending order of contact attempt value for each contact attempt period; and (g) second processor means associated with the automated dialer system for scheduling contact attempts for each of the plurality of accounts, the contact attempts being scheduled within each contact attempt period in order of descending contact attempt value, and collecting contact attempt history data to be transferred to the historical data storage means through historical data upload means.
(a)~ account storage means for storing:
an account data file comprising account information for a plurality of accounts to be processed over the processing period, a priority value for each account, a quantity of contact resources available over the processing period, resource cost values associated with each contact attempt, a plurality of sets of probability score values, and as assignment data file comprising account information for the plurality of accounts and for each account a contact attempt value for each contact period;
(b) master file download means for enabling the account storage means to receive the account data file;
(c) historical data storage means for storing contact attempt history data for each of the plurality of accounts;
(d) parameter input means responsive to operator commands for enabling an operator to input priority data, and the quantity of contact resources available over the processing period;
(e) first processing means coupled to the account storage means and historical data storage means for producing for each account an action result probability for each contact period, each action result probability being produced using one of the sets of probability score values; and producing the contact attempt value for each account and each contact period, the contact attempt value comprising a numerical value representing a relative desirability of attempting to contact the respective account during the respective contact attempt period considering the priority value for each account, the action result probability for each account, the quantity of contact resources available over the processing period, and the resource cost for each contact attempt;
(f) modified download output means for transferring to an automated dialer system a modified download data file comprising account information for each account and contact attempt value information for each account and each contact attempt period, the contact attempt value information enabling the plurality of accounts to be sorted in descending order of contact attempt value for each contact attempt period; and (g) second processor means associated with the automated dialer system for scheduling contact attempts for each of the plurality of accounts, the contact attempts being scheduled within each contact attempt period in order of descending contact attempt value, and collecting contact attempt history data to be transferred to the historical data storage means through historical data upload means.
11. The computer system of claim 10 wherein the resource cost values include a cost of right party contact, a cost of wrong party contact, and a cost of no connect.
12. The computer system of claim 11 wherein:
(a) The action result probability comprises a right party contact probability, a wrong party contact probability, and a no connect probability.
(a) The action result probability comprises a right party contact probability, a wrong party contact probability, and a no connect probability.
13. The computer system of claim 12 wherein the first processor means produces each contact attempt value by:
(a) solving the linear program max ~.SIGMA. .SIGMA. C it P it X it X it i t with constraints X it >= 0 .SIGMA. x it~<= 1 t .SIGMA.(tm (q it) + tc (P it) + tn (r it)) x it <= h t i to produce a solved value x it for each account and each contact period;
where i = account 1...N
t = contact period 1...T
x it~= proportion of account i called at time t P k = right-party contact probability at time t for account i q k ~= wrong-party contact probability at time t for account i r k ~= non connect probability at time t for account i (r k = 1 - q k - P k) tc~= average time of contact attempt that resulted in right-party contact tm~= average time of contact attempt that resulted in wrong-party contact tn~= average time of contact attempt that resulted in no connect h t~= total collector time available at time t c k~= priority value for account i at time t; and (b) associating the solved value x k with its respective account and contact attempt period.
(a) solving the linear program max ~.SIGMA. .SIGMA. C it P it X it X it i t with constraints X it >= 0 .SIGMA. x it~<= 1 t .SIGMA.(tm (q it) + tc (P it) + tn (r it)) x it <= h t i to produce a solved value x it for each account and each contact period;
where i = account 1...N
t = contact period 1...T
x it~= proportion of account i called at time t P k = right-party contact probability at time t for account i q k ~= wrong-party contact probability at time t for account i r k ~= non connect probability at time t for account i (r k = 1 - q k - P k) tc~= average time of contact attempt that resulted in right-party contact tm~= average time of contact attempt that resulted in wrong-party contact tn~= average time of contact attempt that resulted in no connect h t~= total collector time available at time t c k~= priority value for account i at time t; and (b) associating the solved value x k with its respective account and contact attempt period.
14. The computer system of claim 10 wherein:
(a) the first processor means is also for producing forecast data including an expected right party contact value, an expected contact attempt quantity, and an expected objective met value; and (b) the account storage means is also for storing the expected right party contact value, the expected contact attempt quantity, and the expected objective met value.
(a) the first processor means is also for producing forecast data including an expected right party contact value, an expected contact attempt quantity, and an expected objective met value; and (b) the account storage means is also for storing the expected right party contact value, the expected contact attempt quantity, and the expected objective met value.
15. The computer system of claim 10 wherein the first processor means is also for producing the resource cost value using the contact attempt history data.
16. The computer system of claim 10 wherein:
(a) the parameter input means is also for enabling the operator to input contact attempt history summarization parameters;
(b) the first processor means is also for producing summarized contact attempt history data from the contact attempt history data; and (c) the historical data storage means is also for storing the contact attempt history summarization parameters and summarized contact attempt history data.
(a) the parameter input means is also for enabling the operator to input contact attempt history summarization parameters;
(b) the first processor means is also for producing summarized contact attempt history data from the contact attempt history data; and (c) the historical data storage means is also for storing the contact attempt history summarization parameters and summarized contact attempt history data.
17. The computer system of claim 10 wherein the first processor means is also for:
(a) producing for each contact attempt period one of the sets of probability score values for each of a plurality of user-defined populations, the sets of probability score values being produced using a polytomous logistic regression method based upon the contact attempt history data and account history data; and (b) producing the action result probability for each account and contact attempt period using the probability score values for the user-defined populations containing the respective account.
(a) producing for each contact attempt period one of the sets of probability score values for each of a plurality of user-defined populations, the sets of probability score values being produced using a polytomous logistic regression method based upon the contact attempt history data and account history data; and (b) producing the action result probability for each account and contact attempt period using the probability score values for the user-defined populations containing the respective account.
18. The computer system of claim 17 wherein the first processor means is also for rebuilding each set of probability score values at predetermined time intervals.
19. A method of optimizing the allocation of resources to perform a plurality of task attempts in light of a variant which comprises a plurality of permissible variations, the method comprising the steps of:
(a) computing an action result probability for each task attempt and each permissible variation of the variant;
(b) producing a priority value for each task attempt;
(c) determining a quantity of resources available within a time period for accomplishing the plurality of task attempts;
(d) determining resource costs associated with each task attempt;
(e) for each task attempt, producing a task attempt value, each task attempt value comprising a numerical value representing a relative desirability of performing the task attempt for the respective permissible variation considering the priority value for each task attempt, the probability of positive result for each task attempt and each permissible variation the quantity of resources available for making the task attempts, and the resource costs associated with each task attempt; and (f) performing the task attempts within each permissible variation in order of descending task attempt value.
(a) computing an action result probability for each task attempt and each permissible variation of the variant;
(b) producing a priority value for each task attempt;
(c) determining a quantity of resources available within a time period for accomplishing the plurality of task attempts;
(d) determining resource costs associated with each task attempt;
(e) for each task attempt, producing a task attempt value, each task attempt value comprising a numerical value representing a relative desirability of performing the task attempt for the respective permissible variation considering the priority value for each task attempt, the probability of positive result for each task attempt and each permissible variation the quantity of resources available for making the task attempts, and the resource costs associated with each task attempt; and (f) performing the task attempts within each permissible variation in order of descending task attempt value.
20. The method of claim 19 wherein the step of producing a task attempt value for each task attempt and each permissible variation of the variant comprises:
(a) solving the linear program max .SIGMA. .SIGMA. c ij p ij x ij x ij i j with constraints x ij >= 0 .SIGMA. x ij <= 1 j to produce a solved value x ij for each task attempt and each permissible variation of the variant;
where i = task attempt 1... N
j = permissible variation 1...J
k = resource 1...K
x ij = proportion of task attempt i performed with permissible variant j p ij = probability of success when performing task attempt i with permissible variant j a ~ = amount of resource k used when performing task attempt i with permissible variant j c ij = priority value of performing task attempt i with permissible variant j b k = available amount of resource k (b) associating the solved value x ij with its respective task attempt and permissible variation of the variant.
(a) solving the linear program max .SIGMA. .SIGMA. c ij p ij x ij x ij i j with constraints x ij >= 0 .SIGMA. x ij <= 1 j to produce a solved value x ij for each task attempt and each permissible variation of the variant;
where i = task attempt 1... N
j = permissible variation 1...J
k = resource 1...K
x ij = proportion of task attempt i performed with permissible variant j p ij = probability of success when performing task attempt i with permissible variant j a ~ = amount of resource k used when performing task attempt i with permissible variant j c ij = priority value of performing task attempt i with permissible variant j b k = available amount of resource k (b) associating the solved value x ij with its respective task attempt and permissible variation of the variant.
Applications Claiming Priority (3)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| US08/620,601 | 1996-03-22 | ||
| US08/620,601 US5802161A (en) | 1996-03-22 | 1996-03-22 | Method and system for optimized scheduling |
| PCT/US1997/004023 WO1997035415A1 (en) | 1996-03-22 | 1997-03-14 | Method and system for optimized scheduling |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CA2249558A1 CA2249558A1 (en) | 1997-09-25 |
| CA2249558C true CA2249558C (en) | 2002-02-26 |
Family
ID=24486577
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CA002249558A Expired - Fee Related CA2249558C (en) | 1996-03-22 | 1997-03-14 | Method and system for optimized scheduling |
Country Status (5)
| Country | Link |
|---|---|
| US (1) | US5802161A (en) |
| EP (1) | EP0951773A4 (en) |
| AU (1) | AU2079097A (en) |
| CA (1) | CA2249558C (en) |
| WO (1) | WO1997035415A1 (en) |
Families Citing this family (43)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US6473816B1 (en) * | 1997-12-04 | 2002-10-29 | Canon Kabushiki Kaisha | Apparatus and method for determining bus use right |
| GB2347583B (en) * | 1997-12-19 | 2003-03-12 | Blake Rice | Automated right-party contact telephone system |
| US7318046B1 (en) | 1998-03-05 | 2008-01-08 | American Management Systems, Inc. | Collector's account payment promise option advisory apparatus and method |
| US6154769A (en) * | 1998-03-27 | 2000-11-28 | Hewlett-Packard Company | Scheduling server requests to decrease response time and increase server throughput |
| US6879674B2 (en) * | 1998-04-09 | 2005-04-12 | Cim Ltd. | System and method for providing an automatic telephone call back to a telephone line being used to access a computer network |
| US6792082B1 (en) * | 1998-09-11 | 2004-09-14 | Comverse Ltd. | Voice mail system with personal assistant provisioning |
| US7302051B1 (en) | 1998-09-28 | 2007-11-27 | Aspect Software, Inc. | System and method for providing an automatic telephone call back from information provided at a data terminal |
| US7006994B1 (en) | 1999-07-16 | 2006-02-28 | American Management Systems, Inc. | Automated receivables management system |
| US6678371B1 (en) | 2000-02-15 | 2004-01-13 | Avaya Technolocy Corp. | Direct customer control of a network router |
| US6631270B1 (en) * | 2000-04-05 | 2003-10-07 | Telefonaktiebolaget Lm Ericsson (Publ) | Method and apparatus for call completion in congested cells |
| US7103173B2 (en) | 2001-07-09 | 2006-09-05 | Austin Logistics Incorporated | System and method for preemptive goals based routing of contact records |
| US7142662B2 (en) * | 2000-07-11 | 2006-11-28 | Austin Logistics Incorporated | Method and system for distributing outbound telephone calls |
| WO2002063832A1 (en) * | 2001-02-05 | 2002-08-15 | Personity, Inc. | A system and method for filtering unavailable devices in a presence and availability management system |
| US7386850B2 (en) * | 2001-06-01 | 2008-06-10 | Avaya Technology Corp. | Arrangement for scheduling tasks based on probability of availability of resources at a future point in time |
| US7715546B2 (en) * | 2001-07-09 | 2010-05-11 | Austin Logistics Incorporated | System and method for updating contact records |
| US7054434B2 (en) * | 2001-07-09 | 2006-05-30 | Austin Logistics Incorporated | System and method for common account based routing of contact records |
| US7249047B2 (en) * | 2002-01-10 | 2007-07-24 | Navitaire, Inc. | Employee transfer and leave optimization processor |
| US7076041B2 (en) * | 2002-02-21 | 2006-07-11 | International Business Machines | Third party regulation of calls through a particular line based on a call context |
| US7200215B2 (en) * | 2002-02-21 | 2007-04-03 | International Business Machines Corporation | Time based regulation of access to callees |
| US7095835B2 (en) * | 2002-02-26 | 2006-08-22 | International Business Machines Corporation | Time based regulation of use of a telephone line |
| US7206388B2 (en) * | 2002-03-18 | 2007-04-17 | Openwave Systems Inc. | System and method for providing voice-activated presence information |
| US10248917B1 (en) | 2004-10-14 | 2019-04-02 | Capital One Services, Llc | System and method for developing and utilizing a contactability profile |
| US7346851B2 (en) * | 2004-11-03 | 2008-03-18 | Lenovo Singapore Pte. Ltd. | System and method for improved scroll mouse operation |
| CN101258500B (en) * | 2005-08-25 | 2016-05-11 | 新日铁系统集成株式会社 | Controlling equipment, dispatching method and recording medium |
| US8126134B1 (en) * | 2006-03-30 | 2012-02-28 | Verint Americas, Inc. | Systems and methods for scheduling of outbound agents |
| WO2008022313A2 (en) * | 2006-08-18 | 2008-02-21 | Raytheon Company | Method and system for determining task scheduling probability |
| US8655736B2 (en) * | 2007-06-12 | 2014-02-18 | Infosys Limited | Buyer-side consolidation of compatible purchase orders |
| US20090144256A1 (en) * | 2007-11-29 | 2009-06-04 | Moises Cases | Workflow control in a resource hierarchy |
| US8315609B2 (en) | 2009-02-19 | 2012-11-20 | Research In Motion Limited | Personal call center |
| ATE532322T1 (en) * | 2009-02-19 | 2011-11-15 | Research In Motion Ltd | PERSONAL CALL CENTER |
| US8812482B1 (en) | 2009-10-16 | 2014-08-19 | Vikas Kapoor | Apparatuses, methods and systems for a data translator |
| US20130060587A1 (en) * | 2011-09-02 | 2013-03-07 | International Business Machines Corporation | Determining best time to reach customers in a multi-channel world ensuring right party contact and increasing interaction likelihood |
| US11012563B1 (en) | 2012-03-06 | 2021-05-18 | Connectandsell, Inc. | Calling contacts using a wireless handheld computing device in combination with a communication link establishment and management system |
| US10432788B2 (en) | 2012-03-06 | 2019-10-01 | Connectandsell, Inc. | Coaching in an automated communication link establishment and management system |
| US9986076B1 (en) | 2012-03-06 | 2018-05-29 | Connectandsell, Inc. | Closed loop calling process in an automated communication link establishment and management system |
| US9876886B1 (en) * | 2012-03-06 | 2018-01-23 | Connectandsell, Inc. | System and method for automatic update of calls with portable device |
| US11743382B2 (en) | 2012-03-06 | 2023-08-29 | Connectandsell, Inc. | Coaching in an automated communication link establishment and management system |
| US8654964B1 (en) | 2012-12-05 | 2014-02-18 | Noble Systems Corporation | Agent-centric processing of prioritized outbound contact lists |
| US8738076B1 (en) | 2013-04-19 | 2014-05-27 | Noble Systems Corporation | Providing compliance enforcement for manually dialed wireless numbers in a contact center |
| US9036811B1 (en) | 2013-04-19 | 2015-05-19 | Noble Systems Corporation | Dialing a telephone number subject to an autodialer prohibition in a contact center |
| US9319525B1 (en) | 2014-06-23 | 2016-04-19 | Noble Systems Corporation | Best time to call parties having multiple contacts |
| US11102622B1 (en) | 2020-01-20 | 2021-08-24 | Noble Systems Corporation | Best time to send limited-content text messages to parties |
| US11622041B2 (en) * | 2020-12-23 | 2023-04-04 | Acqueon Technologies Inc | System and method for enhanced call progress analysis utilizing dual-prong analysis |
Family Cites Families (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US5594790A (en) * | 1993-01-14 | 1997-01-14 | Davox Corporation | Method for selecting and controlling the automatic dialing of a call record campaign |
| US5436965A (en) * | 1993-11-16 | 1995-07-25 | Automated Systems And Programming, Inc. | Method and system for optimization of telephone contact campaigns |
| US5592543A (en) * | 1994-06-01 | 1997-01-07 | Davox Corporation | Method and system for allocating agent resources to a telephone call campaign |
-
1996
- 1996-03-22 US US08/620,601 patent/US5802161A/en not_active Expired - Lifetime
-
1997
- 1997-03-14 WO PCT/US1997/004023 patent/WO1997035415A1/en not_active Application Discontinuation
- 1997-03-14 EP EP97909039A patent/EP0951773A4/en not_active Withdrawn
- 1997-03-14 CA CA002249558A patent/CA2249558C/en not_active Expired - Fee Related
- 1997-03-14 AU AU20790/97A patent/AU2079097A/en not_active Abandoned
Also Published As
| Publication number | Publication date |
|---|---|
| EP0951773A1 (en) | 1999-10-27 |
| AU2079097A (en) | 1997-10-10 |
| US5802161A (en) | 1998-09-01 |
| WO1997035415A1 (en) | 1997-09-25 |
| EP0951773A4 (en) | 2001-08-01 |
| CA2249558A1 (en) | 1997-09-25 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CA2249558C (en) | Method and system for optimized scheduling | |
| US5889799A (en) | Method and system for optimization of telephone contact campaigns | |
| US6098052A (en) | Credit card collection strategy model | |
| US20040128236A1 (en) | Methods and apparatus for evaluating and using profitability of a credit card account | |
| MXPA05013563A (en) | Call processing system. | |
| US7881959B2 (en) | On demand selection of marketing offers in response to inbound communications | |
| US8051030B2 (en) | System and method for automated population splitting to assist task management through analytics | |
| US20020128960A1 (en) | Systems and methods for managing accounts | |
| CN111882431A (en) | Intelligent message pushing method based on NLP deep learning | |
| CN109409862A (en) | Period service charging method, system, device and storage medium | |
| CN107977855B (en) | Method and device for managing user information | |
| US20030055706A1 (en) | System and method for determining staffing needs for functions in an office | |
| CN1395711A (en) | Personal account management device and method for financial transaction | |
| Gulati et al. | Call center scheduling technology evaluation using simulation | |
| KR100601885B1 (en) | Call recommender system for outbound call center | |
| CN111341038B (en) | Off-border tax refund self-service trusteeship refund system for overseas passenger shopping | |
| US20080059386A1 (en) | System and method for contact device dynamic downloads | |
| Takahashi et al. | Towards early detections of the bad debt customers among the mail order industry | |
| CN112085587A (en) | Remaining principal distribution method and system based on electrocatalysis | |
| JP2007094998A (en) | System for distributing telephony work and its program | |
| CN110490563A (en) | Data processing method, device, computer equipment and storage medium | |
| JP7370416B1 (en) | Call recipient determination system, call recipient determination method, and program | |
| Paradi et al. | Simulation based alternative strategies in risk and service guarantee tradeoffs in the international funds transfer system at a large Canadian Bank | |
| CN115082143A (en) | User screening method, device, equipment and storage medium | |
| CN114051077A (en) | Outbound data distribution method and system |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| EEER | Examination request | ||
| MKLA | Lapsed |