Docstoc

_12_ United States Patent 1 1 - Google Compute Engine

Document Sample
_12_ United States Patent 1 1 - Google Compute Engine Powered By Docstoc
					                                                                                                                                              US008346593B2


(12) United States Patent                                                                                     (10) Patent N0.:                         US 8,346,593 B2
       Fanelli et a].                                                                                         (45) Date of Patent:                            Jan. 1, 2013

(54)    SYSTEM, METHOD, AND SOFTWARE FOR                                                                                             OTHER PUBLICATIONS
        PREDICTION OF ATTITUDINAL AND                                                                  Punj and SteWart;Cluster Analysis in Marketing Reasearch: Review
        MESSAGE RESPONSIVENESS                                                                         and Suggestion for Application; vol. XX (May 1983), 134-48.*
                                                                                                       Sara Dolnicar; Using cluster analysis for market segmentation-typi
(75) Inventors: Marc Christian Fanelli, Kinnelon, NJ                                                   cal misconception, established methodological Weaknesses and
                        (US); Patricia Kay Gormley, Lincoln,                                           some recommendations for improvement; 2003*
                                                                                                       An ef?cient clustering algorithm for market basket data based on
                        NE (US); Kymberly Ann Kulle,                                                   small large ratios CH Yun, KT Chuang, MS Chen4Computer Soft
                        Cincinnati, OH (US); Thomas G.                                                 ware and . . . , 200liieeexplore.ieee.org.*
                        Nocerino, Bethlehem, PA (US);
                        Kaushik Sanyal, Jersey City, NJ (US)                                           * cited by examiner

(73) Assignee: Experian Marketing Solutions, Inc.,                                                     Primary Examiner * Jonathan G Sterrett
                        Schaumburg, IL (US)                                                            Assistant Examiner * Luis Santiago
                                                                                                       (74) Attorney, Agent, or Firm * Nancy R. Gamburd;
(*)     Notice:         Subject to any disclaimer, the term of this                                    Gamburd LaW Group LLC
                        patent is extended or adjusted under 35
                        U.S.C. 154(b) by 2255 days.                                                    (57)                                     ABSTRACT
                                                                                                       The present invention provides a system, method, software
(21) Appl.No.: 10/881,436                                                                              and data structure for independently predicting attitudinal
(22) Filed:             Jun. 30, 2004                                                                  and message responsiveness, using a plurality of attitudinal or
                                                                                                       other identi?cation classi?cations and a plurality of message
(65)                        Prior Publication Data                                                     content or version classi?cations, for a selected population of
                                                                                                       a plurality of entities, such as individuals or households,
        US 2006/0004622 A1                   Jan. 5, 2006                                              represented in a data repository. The plurality of predictive
                                                                                                       attitudinal (or identi?cation) classi?cations and plurality of
(51)    Int. Cl.                                                                                       predictive message content (ore version) classi?cations have
        G06Q 10/00                       (2012.01)                                                     been determined using a plurality of predictive models devel
(52)    US. Cl. .................................... ..                     705/7.29                   oped from a sample population and applied to a reference
(58)    Field of Classi?cation Search ................ .. 705/729                                      population represented in the data repository, such as attitu
        See application ?le for complete search history.                                               dinal, behavioral, or demographic models. For each predic
                                                                                                       tive attitudinal (or identi?cation) classi?cation, at least one
(56)                        References Cited                                                           predominant predictive message content or version classi?
                                                                                                       cation is independently determined. The exemplary embodi
                   U.S. PATENT DOCUMENTS                                                               ments also provide, for each predictive attitudinal classi?ca
       5,930,764 A *          7/1999     Melchione et al. ........ .. 705/729                          tion, corresponding information concerning predominant
       7,013,285 B1*          3/2006     Rebane ......... ..              .. 705/7.31                  communication media (or channel) types, predominant com
       7,080,027 B2*          7/2006 Luby et al.                          .. 705/7.31
       7,165,037 B2*          1/2007 Lazarus et al.                       .. 705/7.31
                                                                                                       munication timing, predominant communication frequency,
       7,212,979 B1*          5/2007     MatZ et a1 ..... ..                 . 705/1.1                 and predominant communication sequencing.
       7,783,534 B2*          8/2010 Armstrong et al                         .. 705/29
 2005/0091077 A1*            4/2005      Reynolds ........................ .. 705/1                                                 35 Claims, 7 Drawing Sheets



                                                                                             ?
                                                              mun: THE rwmm 0F wmlmvz ATTITUDINAL :ussmcmws BASED on
                                                              rmnve mama mu SELECTED PoPuunuu SIZES m: PENETRATIW moms
                                                                                                                                       43"



                                                          ‘             DETEPMINE CURE mnumm CLASSIFICATIONS mvms                      ‘'35
                                                                   COHPNTATIVELY [HEATER PELETMTION IHDIBES ANJ CUTPARATIVELV
                                                                         GIUTET PRWURTIONS 11: THE SELECTED POPULATION

                                                                                               1
                                                                       IJETETHITE NIEHE ATTITUDINAL CLASSIFICATIONS HAVING             “0
                                                                   EWPAHATIVELY GREATER PENETRATION IHDIEEB All] CDWAHATIVELV
                                                                         LESSER PROPTHTIONS BF THE REFERENCE POPULATIGN

                                                                                               1
                                                                     05121111115 sauna mnumm cussmcnms 11mm;                           “5
                                                                   cwmmnv LESSER Psmmwu IIUICES AND cwmnvm
                                                                       swim mwnnnms 0F m2 PEFHTENII POPULATION

                                                                                               1
                                                                  FOR EACH FBEDTETIVE ATTITUDINAL CLhSSIFICATIDN. IMTEPENDEHTLY      /450
                                                               DETERMINE THE PREDDMINANT FREDICTTVE NEW CONTENT CLASSIFICATIUNS
                                                                BASED UPUT THE INDIVIDUALS [UT BRUUPS) If TIE SELECTED POPULATION
                                                                   ASSISTED TO THE BIVEN PREDIETIVE ATTITUDINAL IILASSIFICATIIIN

                                                                                               1
                                                                     run um PHEDICTI‘JE ATTITUDINAL cussIFIcAnuu. m                   ‘55
                                                                   m: mmmrmm pnsmcuvs nessms cum cussmcmuns
US. Patent                        Jan. 1, 2013             Sheet 1 017                   US 8,346,593 B2


                            175
                            /                    FIG. 1
                                COMPUTER

              17o /
               /                                     175
                   REMOTE                             /
              NETNRRK (WEB) <:=<>          COMPUTER
                 SERVER

                                                                              COMPUTER

       180                                                                       8
                                                                               REMOTE       / 17°
                                                 -                       4> NETWORK (WEB)
                                                                               sERvER

                     l 150
                        /
                                                                                 E
                                                                              COMPUTER      / 175
                 NETNORT<
              (NERT sERvER
               /\       A




        150
         I_..___ ........................... -.
  l—                                                          _|
  i            V                f 125                           |
  =           APPLICATION
  !              sERvER                                         !
  l                                                                                            =    115
  i                   1
                    H 140                                       i                ii
                                                                            PROcEss0RTs1       !
  ,             DATABASE                                             I                              120
  I1             sERvER
               MANAGEMENT                                       |I             MEMORY          i

  L ................................. .._J                                    i/?/NOA
                                                                     L                       J
US. Patent                        Jan. 1, 2013                Sheet 2 0f 7                        US 8,346,593 B2



                                                                  FIG. 2




                         § DATA HEPOSITOHY //
                  220~
                         \ [gEAMT%HTAAPBHLIEC                                     230                            240
                                                          CENSUS DATA /                                      f
                                                             TABLE                             ATTITDDINAL
                  225                                                                     CLASSIFICATION TABLE
                         \ RE?gggL?BLE                                [215
                                                                                                MESSAGE
                         \     BEsPDNsE                       DATA TABLE                      THEME TABLE
                              DATA TABLE                                   205
                                                r--I-
                                                .
                                                                      /
                                                              NANE TABLE     —
                                                                                                             /'25U
                         \THANSACTIUNAL !                                  200          COMMUNICATION CHANNEL/
                                  DATA          :                    f                        MEDIA TABLE




           .-q|
                                                ;        INDIVIDUAL 0B _                                         255
                         .......... "I--- HOUSEHOLD ID                                                       /
                                                :                                            cDNNuNIcATIoN
                                                5                                             TININB TABLE
                                                6 fm
                                         ADDBEss                                             COMMUNICATION
                                          TABLE                                             FREQUENCY TABLE

                                                    [235                   COMMUNICATION                     \260
                                    suNNABIzED
                               FINANCIAL DATA TABLE                        SEQUENCE TABLE
                                                        230                                \2B5
                                                    /                  DTNEB INEDBNATIDN
                                 CENSUS DATA TABLE                               TABLES
  270\
        SELECTED
    POPULATION DATA
                                                                                           EV
US. Patent             Jan. 1, 2013         Sheet 3 of7                  US 8,346,593 B2




                                   FIG. 3A
                     (START; ATTITUDINAL SURVEY RESULTS)/3o0
                                        '                                           30s
                             PERFORM FACTOR ANALYSIS                            /


                                        I
              DETERMINE A PLURALITY 0F EMPIRICAL ATTITUDINAL FACTORS            /310

                                        I
                SCORE 0R EVALUATE SURVEY PARTICIPANTS. AS A SAMPLE              /315
               POPULATION. USING EACH EMPIRICAL ATTITUDINAL FACTOR

                                        L
                     SEARCH AND MATCH SURVEY PARTICIPANTS. AS                   /320
                  INDIVIDUALS. T0 CORRESPONDING DATABASE RECORDS

                                        1
  FOR MATCHING INDIVIDUALS, APPEND LINK OR MATCH THEIR CORRESPONDING DEMOGRAPHIC. fazs
           LIFESTYLE. BEHAVIORAL AND OTHER VARIABLES FROM THE DATABASE

                                        L
            USING APPENDED DATABASE VARIABLES AS INDEPENDENT VARIABLES          /330
         AND SCORES FROM THE EMPIRICAL ATTITUDINAL FACTORS AS DEPENDENT
         VARIABLES. DEVELOP A PLURALITY OF PREDICTIVE ATTITUDINAL MODELS

                                        L
  USING THE PLURALITY OF PREDICTIVE ATTITUDINAL MODELS, SCORE (OR EVALUATET EACH fags
   INDIVIDUAL (OR GROUP) REPRESENTED IN THE DATABASE. AS A REFERENCE POPULATION.
      TO DETERMINE A PLURALITY OF PREDICTIVE MESSAGE CONTENT CLASSIFICATIONS
US. Patent              Jan. 1, 2013           Sheet 4 of7                 US 8,346,593 B2




                                        FIG. 3B

                                                                                   340
                                                                                   S
   STORE ATTITUDINAL MODEL SCORING RESULTS IN THE DATABASE. AS DICHOTOMOUS VARIABLES.
      AS PROBABILITIES. OR AS ANOTHER FORM. FOR EACH INDIVIDUAL (0R GROUP) OF THE
      REFERENCE POPULATION. ACROSS EACH PREDICTIVE MESSAGE CONTENT CLASSIFICATION

                                          )
            PERFORM CLUSTER OR OTHER GROUPING ANALYSIS USING THE PLURALITY               /345
              OF PREDICTIVE MESSAGE CONTENT CLASSIFICATIONS TO DETERMINE
                 A PLURALITY OF PREDICTIVE ATTITUDINAL CLASSIFICATIONS

                                           )
              ASSIGN EACH INDIVIDUAL (GROUP) OF THE REFERENCE POPULATION                 faso
                TO A PREDOMINANT PREDICTIVE ATTITUDINAL CLASSIFICATION

                                           )
       STORE ASSIGNMENTS OF EACH INDIVIDUAL (GROUP) OF THE REFERENCE POPULATION.         /355
        TO A PREDOMINANT PREDICTIVE ATTITUDINAL CLASSIFICATION. IN THE DATABASE


                                                      360
US. Patent            Jan. 1, 2013          Sheet 5 of7                  US 8,346,593 B2




                                     FIG. 4A

                      (START: SELECTED POPULATION DAIA)/4°0

             SEARCH AND MATCH SELECTED POPULATION DATA TO A PLURALITY
                OF INDIVIDUALS OR GROUPS REPRESENTED IN A DATABASE

                                        T
                 APPEND. TO EACH MATCHED INDIVIDUAL 0R GROUP. THE                /410
           CORRESPONDING PREDICTIVE ATTITUDINAL CLASSIFICATION AND THE
             CORRESPONDING PREDICTIVE MESSAGE CONTENT CLASSIFICATIONS

                                        T
             DETERMINE DISTRIBUTION OF SELECTED POPULATION ACROSS THE            /415
                PLURALITY OF PREDICTIVE ATTITUDINAL CLASSIFICATIONS

                                        T
           COMPARE EACH SELECTED POPULATION DISTRIBUTION TO A REFERENCE          /42°
           DISTRIBUTION. FOR EACH PREDICTIVE ATTITUDINAL CLASSIFICATION

                                        T
    DETERMINE PENETRATION INDEX FOR EACH PREDICTIVE ATTITUDINAL CLASSIFICATION   /425
US. Patent            Jan. 1, 2013          Sheet 6 of7                   US 8,346,593 B2




                                     FIG. 4B




     EVALUATE THE PLURALITY OF PREDICTIVE ATTITUDINAL CLASSIFICATIONS BASED ON   /430
     RELATIVE REFERENCE AND SELECTED POPULATION SIZES AND PENETRATION INDICES

                                        L
                 DETERMINE CORE ATTITUDINAL CLASSIFICATIONS HAVING               f435
            COMPARATIVELY GREATER PENETRATION INDICES AND COMPARATIVELY
                   GREATER PROPORTIONS OF THE SELECTED POPULATION

                                        L
                DETERMINE NICHE ATTITUDINAL CLASSIFICATIONS HAVING               /440
           COMPARATIVELY GREATER PENETRATION INDICES AND COMPARATIVELY
                  LESSER PROPORTIONS OF THE REFERENCE POPULATION

                                        L
               DETERMINE GROWTH ATTITUDINAL CLASSIFICATIONS HAVING               /445
            COMPARATIVELY LESSER PENETRATION INDICES AND COMPARATIVELY
                 GREATER PROPORTIONS OF THE REFERENCE POPULATION

                                        L
           FOR EACH PREDICTIVE ATTITUDINAL CLASSIFICATION. INDEPENDENTLY         /450
       DETERMINE THE PREDOMINANT PREDICTIVE MESSAGE CONTENT CLASSIFICATIONS
         BASED UPON THE INDIVIDUALS TOR GROUPS) OF THE SELECTED POPULATION
            ASSIGNED TO THE GIVEN PREDICTIVE ATTITUDINAL CLASSIFICATION


                                        L
               FOR EACH PREDICTIVE ATTITUDINAL CLASSIFICATION. RANK              /455
            THE PREDOMINANT PREDICTIVE MESSAGE CONTENT CLASSIFICATIONS
US. Patent         Jan. 1, 2013         Sheet 7 of7         US 8,346,593 B2




                                  FIG‘ .   46




      FOR EACH PREDICTIVE ATTITUDINAL CLASSIFICATION. DETERMINE    f460
   PREDOMINANT COMMUNICATION CHANNEL AND! OR MEDIA CLASSIFICATIONS

                                    L
     FOR EACH PREOICTIVE ATTITUDINAL CLASSIFICATION, DETERMINE    /455
         PREDOMINANT COMMUNICATION TIMING CLASSIFICATIONS

                                    T
     FOR EACH PREDICTIVE ATTITUDINAL CLASSIFICATION. DETERMINE    /470
       PREDOMINANT FREQUENCY OF COMMUNICATION CLASSIFICATIONS

                                    T
                  FOR EACH PREDICTIVE ATTITUDINAL                 /475
               CLASSIFICATION. DETERMINE PREDOMINANT
                     SEQUENCE OF conuumcmous
                           CLASSIFICATIONS

                                    T
                     OUTPUT (AND STORE) RESULTS
                                                    US 8,346,593 B2
                              1                                                                     2
     SYSTEM, METHOD, AND SOFTWARE FOR                                 marketing. Other methods and systems are required to appro
        PREDICTION OF ATTITUDINAL AND                                 priately target and motivate the remainder of the target audi
           MESSAGE RESPONSIVENESS                                     ence, and to determine potentially neW and underdeveloped
                                                                      target audiences. In addition, neW methods and systems are
               FIELD OF THE INVENTION                                 required to maximiZe marketing returns, by not overly satu
                                                                      rating the target audience With excessive and ineffective com
  The present invention relates, in general, to database man          munications, and instead to appropriately communicate With
agement systems and, more particularly, to a system, method           the target audience using the audience’s preferred methods
and software for independently predicting attitudinal and             and times of communication.
message responsiveness, and preferences for communication               As a consequence, a need remains for a predictive meth
media, channel, timing, frequency, and sequences of commu             odology and system, for accurate prediction of attitudes,
nications, using an integrated data repository.                       motivations and behaviors, Which may be utiliZed for market
                                                                      ing applications. Such a method and system should be empiri
         BACKGROUND OF THE INVENTION                                  cally-based, such as based on actual attitudinal, behavioral or
                                                                      demographic research and other information from a popula
  Business and consumer records are typically contained in            tion sample, and further should provide accurate modeling to
databases and other forms of data repositories. Typical data          predict and extrapolate such attitudinal or other information
bases may contain records such as demographic data, cus               to a larger or entire population. Such a method and system
tomer data, marketing data, name and address information,             should provide information concerning preferred message
observed and self-reported lifestyle and other behavioral        20   themes or message content independently from any popula
data, consumer data, public record information, realty and            tion grouping, segmentation or clustering process. In addi
property tax information, summarized automotive statistics,           tion, such a method and system should be actionable, provid
summarized ?nancial data, census data, and so on. Virtually           ing not only audience attitudinal information and preferred
any type of information may be contained in any such data             message content, but also preferred communication channel
base. One such highly inclusive database, containing much of     25   information or other preferred communication media, pre
the above-mentioned types of data for approximately 98% of            ferred frequency of communication or other contact, and
US. individuals and living units (households), is the Expe            communication timing information.
rian INSOURCE® database.
   Various database applications have been directed to                            SUMMARY OF THE INVENTION
attempts to utiliZe the Wide array of information contained in   30
such databases for marketing and analytical purposes. For               The present invention provides a system, method and soft
example, demographic data may be appended to customer                 Ware for independently predicting a plurality of ?rst, message
records, to identify the demographic composition of a set of          content classi?cations and a plurality of second, attitudinal
customers, followed by marketing directed toWard people               classi?cations, for a selected population of individuals,
having similar demographic characteristics.                      35   households, living units or other groupings of people repre
  These database applications, in their various forms,                sented in a data repository, such as a selected population of
attempt to understand and access distinct customer and pros           customers or prospects represented in a database or data ?les.
pect groups, and then send the right message to the right             In addition, the system, method and softWare of the invention,
individual, household, living unit or other target audience.          depending on the selected embodiment, also determine pref
Typically, all of the individuals and/ or households contained   40   erences for communication channel or other media forms,
in the corresponding database are segmented into groups               communication timing, frequency of communication, and/or
Which share distinct demographic, lifestyle, and consumer             sequences of types of communication.
behavior characteristics. In other applications, folloWing              The illustrated, exemplary embodiments of the present
such segmentation, consumer attitudes and motivations are             invention are empirically-based, using actual attitudinal
assumed and attributed to those individuals/households           45   research and other information from a population sample.
Within each such segment or cluster. The number of segments           Other types of research or data may also be utiliZed, such as
utiliZed varies Widely by application.                                transactional data, demographic data, marketing research
   In addition, in these various database marketing applica           data, or other types of survey information. With this empirical
tions, consumer attitudes and preferred marketing message             basis, the invention provides accurate modeling to predict and
themes or types are generally assumed and assigned to a          50   extrapolate such attitudinal, behavioral, demographic or
segment, Without any independent empirical research and               other information to a larger reference population, thereby
analysis. As a consequence, once a population is segmented,           providing for accurate prediction of attitudes, motivations
any further analysis of the population based on preferred             and behaviors, Which may be utiliZed for marketing applica
messaging themes does not, in fact, add any additional, inde          tions, for example. The exemplary embodiments of the inven
pendent information, and merely reiterates the underlying        55   tion further provide information concerning preferred mes
message theme assumptions of any given segment.                       sage themes or message content independently from any
  The resulting data, moreover, may have a large degree of            population grouping, segmentation or clustering process. In
uncertainty, may or may not be accurate, and may or may not           addition, the exemplary embodiments of the invention pro
be actionable. For example, the attitudes, motivations and            vide actionable results, providing not only audience attitudi
behaviors attributed or assigned to each segment may not be      60   nal information and preferred message content, but also pre
accurate and may not be based on factual, empirical research.         ferred communication channel information, communication
Such attitudes, motivations and behaviors may or may not              media, communication frequency, and communication tim
actually re?ect representative attitudes found in a particular        ing and sequencing information.
customer database.                                                      The poWer of the invention cannot be overstated. As indi
  The diminished accuracy of current marketing methods is        65   cated above, prior art methods have focused on ?nding
further underscored by comparatively loW response rates,              “Who”, namely, those individuals or households to Whom
such as 1-2% response from a target audience for direct mail          marketers should direct their communications. None of these
                                                       US 8,346,593 B2
                               3                                                                         4
prior art methods provide, independently of the selection of                   having the corresponding predictive attitudinal classi?
“Who”, determination of the “What” of the communication,                       cation of the plurality of predictive attitudinal classi?
such as preferred content or versions of marketing informa                     cations.
tion. None of these prior art methods provide independent                   In addition, depending upon the selected embodiment, for
information on the “When” of the communication, such as the              each entity (e.g., individual or household) of the plurality of
customer’s or prospect’s preferred time of day to receive                entities of the selected population, the various embodiments
communications. None of these prior art methods provide                  optionally provide for appending from the data repository a
independent information on the “hoW” of the communication,               corresponding predictive communication media (or other
such as the customer’s or prospect’s preferred medium or                 channel) classi?cation of a plurality of predictive communi
                                                                         cation media classi?cations, a corresponding predictive com
channel for communication, such as direct mail, telephone,
                                                                         munication timing classi?cation of a plurality of predictive
electronic mail (email), broadcast media, print media, and so
                                                                         communication timing classi?cations, a corresponding pre
on. Lastly, none of these prior art methods provide indepen              dictive frequency of communication classi?cation of a plu
dent information on the frequency (hoW often) and sequenc                rality of predictive communication frequency classi?cations,
ing (ordering) of the communications, based on preferences,              and a corresponding predictive sequence of communications
such as print media for a ?rst number of times, folloWed by              of a plurality of predictive communication sequence classi?
direct mail for a second number of times, folloWed by email,             cations, With these classi?cations having been determined
for example.                                                             from information stored in the data repository.
   More speci?cally, in exemplary embodiments, the present                 Typically, the plurality of predictive communication media
invention provides a method, system and softWare for inde           20   classi?cations comprises at least tWo of the folloWing com
pendently predicting both a plurality of ?rst predictive clas            munication media (equivalently referred to as communica
si?cation, referred to as message content classi?cations, and            tion channels): electronic mail, internet, direct mail, telecom
a plurality of second predictive classi?cations, referred to as          munication, broadcast media (such as radio, television, cable,
attitudinal or other behavioral classi?cations, for a selected           satellite), video media, optical media (DVD, CD), print media
population of a plurality of individuals, households, living        25   (such as neWspapers, magaZines), electronic media (such as
units or other groupings of persons, as “entities”, represented          Web sites and electronic forms of neWspapers, magaZines),
in a data repository. As used herein, any reference to “entity”          and public display media (such as signage, billboards, multi
or “entities” should be understood to mean and include any               media displays). Depending upon the selected embodiment,
individual, household, living unit, group or potential group             the plurality of communication media and channel classi?ca
                                                                    30   tions may be more or less speci?c, such as further subdividing
ing of one or more people, Whether related or unrelated,
                                                                         print and electronic media channels into neWspaper, Weekly
individually or collectively, hoWever de?ned or demarcated,
such as a household, a living unit, a geographic unit, or any
                                                                         magazines, monthly magazines, journals, business reports,
                                                                         and further into their print, internet, email or electronic ver
other grouping of individuals for Whom or Which data may be
                                                                         sions. In addition, various forms of broadcast media may have
maintained, generally at a granular or atomic level, in a data      35   any of a plurality of forms, such as cable, satellite, television
base.                                                                    and radio frequency transmission, internet, etc. Also typi
  In the exemplary embodiments, empirical attitudinal                    cally, the plurality of predictive communication timing clas
research and predictive attitudinal classi?cations are illus             si?cations comprises at least tWo of the folloWing communi
trated as examples, and should be understood to mean and                 cation timing classi?cations: morning, afternoon, evening,
include other forms of research and classi?cations, such as         40   night, Weekday, Weekend, any time (no preference), and
behavioral or demographic classi?cations formed from cor                 none. The plurality of predictive communication frequency
responding empirical research, such as corresponding behav               classi?cations typically comprises at least tWo of the folloW
ioral or demographic survey research, for example.                       ing frequency of communication classi?cations: daily,
   The various exemplary method, system and softWare                     Weekly, biWeekly, monthly, semi-monthly, bimonthly, annu
embodiments of the invention, perform the folloWing:                45   ally, semi-annually, and none. Lastly, the plurality of commu
   First, for each entity (e. g., individual or household) of the        nication sequences are highly varied and may include, for
      plurality of entities of the selected population, append           example, print communications, folloWed by electronic com
      ing from the data repository a corresponding predictive            munications.
     attitudinal classi?cation of a plurality of predictive atti           In the various embodiments, the plurality of predictive
     tudinal classi?cations, and a corresponding plurality of       50   message content classi?cations are or have been determined
     predictive message content classi?cations, With the cor             by:
     responding predictive attitudinal classi?cation and cor               ?rst, developing a plurality of empirical attitudinal factors
     responding plurality of predictive message content clas                   based on a factor analysis of an attitudinal survey of a
     si?cations having been determined using a plurality of                    sample population;
     predictive (attitudinal) models developed from a sample        55     second, using each empirical attitudinal factor of the plu
    population and applied to a reference population repre                   rality of empirical attitudinal factors, scoring each par
    sented in the data repository.                                             ticipant of the attitudinal survey to create a correspond
  Second, for each predictive attitudinal classi?cation of the                 ing plurality of empirical attitudinal factor scores;
     plurality of predictive attitudinal classi?cations, deter             third, using a plurality of selected variables from the data
    mining a penetration index of the selected population           60        repository as independent variables, and using the cor
    compared to the reference population.                                      responding plurality of empirical attitudinal factor
  Third, for each predictive attitudinal classi?cation of the                  scores as dependent variables, performing a regression
     plurality of predictive attitudinal classi?cations, inde                  analysis to create the plurality of predictive attitudinal
     pendently determining at least one predominant predic                     models;
     tive message content classi?cation from the appended           65     fourth, using each predictive attitudinal model of the plu
     plurality of predictive message content classi?cations of                rality of predictive attitudinal models, scoring the plu
     the plurality of individuals of the selected population                   rality of entities represented in the data repository, as the
                                                        US 8,346,593 B2
                                5                                                                           6
     reference population, to create the plurality of predictive           ture comprises: a ?rst ?eld having a plurality of predictive
     message content classi?cations; and                                   identi?cation classi?cations, Wherein the plurality of predic
  ?fth, independently determining the plurality of predictive              tive identi?cation classi?cations designate a plurality of enti
     attitudinal classi?cations by a cluster analysis of the               ties according to a selected property; and a second ?eld hav
     plurality of predictive message content classi?cations of             ing, for each predictive identi?cation classi?cation of the ?rst
     each entity of the plurality of entities represented in the           ?eld, at least one predominant predictive message version
     data repository.                                                      classi?cation of a plurality of predictive message version
As indicated above, in lieu of or in addition to the attitudinal           classi?cations, the plurality of predictive identi?cation clas
research and predictive attitudinal classi?cations, other types            si?cations and the plurality of predictive message version
of research and corresponding classi?cations may also be                   classi?cations having been determined from a plurality of
formed, such as behavioral, demographic, and transactional.                predictive models developed from a sample population and
  The invention also provides for determining core, niche
                                                                           applied to a reference population represented in the data
and groWth attitudinal classi?cations, as folloWs:
  determining one or more core attitudinal classi?cations by               repository.
     selecting, from the plurality of predictive attitudinal                 The data structure may also include a third ?eld having, for
     classi?cations, at least one predictive attitudinal classi            each predictive identi?cation classi?cation of the ?rst ?eld, at
     ?cation having a comparatively greater (e.g., average or              least one predominant predictive communication media clas
     above average) penetration index and having a compara                 si?cation of a plurality of predictive communication media
     tively greater proportion of a selected population;                   classi?cations; a fourth ?eld having, for each predictive iden
  determining one or more niche attitudinal classi?cations 20              ti?cation classi?cation of the ?rst ?eld, at least one predomi
     by selecting, from the plurality of predictive attitudinal            nant predictive communication timing classi?cation of a plu
     classi?cations, at least one predictive attitudinal classi            rality of predictive communication timing classi?cations; a
     ?cation having a comparatively greater penetration                    ?fth ?eld having, for each predictive identi?cation classi?ca
     index and having a comparatively lesser proportion of                 tion of the ?rst ?eld, at least one predominant predictive
     the reference population; and                                    25   communication frequency classi?cation of a plurality of pre
  determining one or more groWth attitudinal classi?cations                dictive communication frequency classi?cations; a sixth ?eld
     by selecting, from the plurality of predictive attitudinal            having, for each predictive identi?cation classi?cation of the
     classi?cations, at least one predictive attitudinal classi            ?rst ?eld, at least one predominant predictive communication
     ?cation having a comparatively lesser (e.g., beloW aver               sequencing classi?cation of a plurality of predictive commu
    age) penetration index and having a comparatively                 30   nication sequencing classi?cations; and a seventh ?eld hav
    greater proportion of the reference population.                        ing a penetration index for each predictive identi?cation clas
  In yet another aspect of the invention, the exemplary                    si?cation of the plurality of predictive identi?cation
embodiments provide a method and system for independently                  classi?cations. As indicated above, the selected property is
predicting communication responsiveness of a selected popu                 derived from at least one of the folloWing: attitudinal charac
lation of a plurality of entities represented in a data repository.   35   teri stics, behavioral characteristics, demographic character
The method comprises: (a) for each entity of the plurality of              istics, geographic characteristics, ?nancial characteristics, or
entities of the selected population, appending from the data               transactional characteristics.
repository a corresponding predictive identi?cation classi?                  In yet another aspect of the invention, the exemplary
cation of a plurality of predictive identi?cation classi?ca                embodiments provide a method for independently predicting
tions, Wherein the plurality of predictive identi?cation clas         40   communication media responsiveness of a selected popula
si?cations designate a plurality of entities according to a                tion of a plurality of entities represented in a data repository,
selected property; (b) for each entity of the plurality of entities        comprising: (a) for each entity of the plurality of entities of
of the selected population in a corresponding predictive iden              the selected population, appending from the data repository a
ti?cation classi?cation, appending at least one corresponding              corresponding predictive identi?cation classi?cation of a plu
predictive message version classi?cation of a plurality of            45   rality of predictive identi?cation classi?cations, Wherein the
predictive message version classi?cations, the plurality of                plurality of predictive identi?cation classi?cations designate
predictive identi?cation classi?cations and the plurality of               a plurality of entities according to a selected property; and (b)
predictive message version classi?cations having been deter                for each predictive identi?cation classi?cation of the plurality
mined from a plurality of predictive models developed from a               of predictive identi?cation classi?cations, independently
sample population and applied to a reference population rep           50   determining at least one predominant predictive communica
resented in the data repository; and (c) for each predictive               tion media classi?cation of a plurality of predictive commu
identi?cation classi?cation of the plurality of predictive iden            nication media classi?cations.
ti?cation classi?cations, independently determining at least                 In other embodiments, instead of step (b) above, the exem
one predominant predictive message version classi?cation                   plary method provides for each predictive identi?cation clas
from the corresponding, appended predictive message ver               55   si?cation of the plurality of predictive identi?cation classi?
sion classi?cations of the plurality of entities of the selected           cations, independently determining at least one predominant
population of the predictive identi?cation classi?cation. The              predictive communication timing classi?cation of a plurality
selected property is derived from at least one of the folloWing:           of predictive communication timing classi?cations, or for
attitudinal characteristics, behavioral characteristics, demo              independently determining at least one predominant predic
graphic characteristics, geographic characteristics, ?nancial         60   tive communication frequency classi?cation of a plurality of
characteristics, or transactional characteristics.                         predictive communication frequency classi?cations.
  In yet another aspect of the invention, the exemplary                      These and additional embodiments are discussed in greater
embodiments provide a data structure for independently pre                 detail beloW. Numerous other advantages and features of the
dicting communication responsiveness of a selected popula                  present invention Will become readily apparent from the fol
tion of a plurality of entities represented in a data repository.     65   loWing detailed description of the invention and the embodi
Such a data structure may be stored in a database, transmitted             ments thereof, from the claims and from the accompanying
electronically, or stored in a tangible medium. The data struc             draWings.
                                                      US 8,346,593 B2
                               7                                                                       8
      BRIEF DESCRIPTION OF THE DRAWINGS                                 database (or data repository) 100A. The memory 120 may be
                                                                        external, such as an external magnetic disk, tape, or optical
  The objects, features and advantages of the present inven             drive. The second system 150, such as an open or netWork
tion Will be more readily appreciated upon reference to the             system, comprises a data repository (or database) 100B (also
following disclosure When considered in conjunction With the            embodied in a form of memory, discussed beloW), a database
accompanying draWings and examples Which form a portion                 management server 140, and/or an application server 125. A
of the speci?cation, in Which:                                          “data repository”, “database”, and “data Warehouse”, as used
   Figure (or “FIG.”) 1 is a block diagram illustrating ?rst and        herein, are considered interchangeable, and may be rela
second exemplary system embodiments in accordance With                  tional, object-oriented, object-relational, or use ?les or ?at
the present invention.                                                  ?les, or any combinations of the above. Both database 100A
   Figure (or “FIG.”) 2 is a block diagram illustrating an              and 100B are instantiations of a database 100, discussed in
exemplary integrated data repository in accordance With the             greater detail beloW.
present invention.                                                         In the exemplary embodiments of system 150, the database
   Figure (or “FIG.”) 3 (divided into FIGS. 3A and 3B and               management server 140 and the application server 125 may
collectively referred to as FIG. 3), is a How diagram illustrat         be implemented together, such as implemented Within the
ing an exemplary method for determination of predictive                 application server 125. Either or both of the database man
attitudinal classi?cations and predictive message content               agement server 140 and the application server 125 are con
classi?cations using a data repository in accordance With the           nected or coupled (or couplable) to the data repository (data
present invention.                                                      base) 100B, for full duplex communication, such as for
   Figure (or “FIG.”) 4 (divided into FIGS. 4A, 4B and 4C and      20   database queries, database ?le or record transfers, database
collectively referred to as FIG. 4), is a How diagram illustrat         updates, and other forms of database communication. In the
ing an exemplary method of independently predicting a plu               second system embodiment 150, the database management
rality of attitudinal classi?cations, a plurality of message            server 140 and/or the application server 125 perform the
content classi?cations, and other predictive information, of a          methodology of the invention utiliZing a correspondingly
selected population using a data repository in accordance          25   programmed or con?gured processor as discussed beloW (not
With the present invention.                                             separately illustrated), such as a processor 115 illustrated for
                                                                        system 110, in conjunction With a database 100 (such as
   DETAILED DESCRIPTION OF THE INVENTION                                database 100B).
                                                                          Typically, the databases 100A and 100B are ODBC-com
   While the present invention is susceptible of embodiment        30   pliant (Open Database Connectivity), although this is not
in many different forms, there are shoWn in the draWings and            required for the present invention. The ?rst system 110 and
Will be described herein in detail speci?c examples and                 second system 150 may also be coupled to or may be part of
embodiments thereof, With the understanding that the present            a local area netWork (“LAN”) 130 or, not separately illus
disclosure is to be considered as an exempli?cation of the              trated, a Wide area netWork (“WAN”), such as for full duplex
principles of the invention and is not intended to limit the       35   communication With a plurality of computers (or other termi
invention to the speci?c examples and embodiments illus                 nals) 135, also for database queries, database ?le or record
trated.                                                                 transfers, database updates, and other forms of database com
   As indicated above, the present invention provides a sys             munication. The LAN 130 communication capability pro
tem, method and softWare for independently predicting a                 vides for the ?rst system 110 and second system 150 to be
plurality of attitudinal classi?cations and a plurality of mes     40   accessible for local access to the databases 100A and 100B,
sage content classi?cations, for a selected population of indi          such as for large ?le transfers or other batch processing,
viduals, households or other living units (“entities”) repre            discussed in greater detail beloW. In addition, the ?rst system
sented in a data repository, such as a database. The                    110 may also be directly accessible (185), such as for loading
embodiments of the present invention provide a predictive               of records (e.g., magnetic tape records or other media) for
methodology, system and softWare, for accurate prediction of            batch processing.
attitudes, motivations and behaviors, Which may be utiliZed               The ?rst system 110 and second system 150 may also be
for marketing, research, assessment, and other applications.            included Within or coupled to a larger data communication
The embodiments of the invention are empirically-based                  netWork 180, through netWork (or Web) server 160, for full
upon actual attitudinal research and other information from a           duplex communication With remote devices, such as a remote
population sample, and provide accurate modeling to predict        50   Internet or other netWork server 170 and remote computer (or
and extrapolate such attitudinal information to a larger refer          other terminal) 175. Such remote communication capability
ence population. The embodiments of the invention further               provides for the ?rst system 110 and second system 150 to be
provide information concerning preferred message themes or              accessible for on-line functionality, discussed in greater detail
message content independently from any population group                 beloW, such as for Web-based access, using any of the prior art
ing, segmentation or clustering process. In addition, the          55   protocols, such as hypertext transfer protocol (HTTP) or
embodiments of the invention provide actionable results, pro            other Internet Protocol (“IP”) forms of communication for
viding not only audience attitudinal information and pre                data, voice or multimedia.
ferred message content, but also preferred communication                  The data repository (or database) 100, illustrated as data
and media channel information, communication frequency,                 bases 100A and 100B, may be embodied in any number of
and communication timing and sequence information.                 60   forms, including Within any data storage medium, memory
   FIG. 1 is a block diagram illustrating ?rst exemplary sys            device or other storage device, such as a magnetic hard drive,
tem embodiment 110 and second exemplary system embodi                   an optical drive, a magnetic disk or tape drive, other machine
ment 150 in accordance With the present invention. As illus             readable storage or memory media such as a ?oppy disk, a
trated in FIG. 1, the ?rst exemplary system embodiment 110              CDROM, a CD-RW, DVD or other optical memory, a
is a computer system embodiment (e.g., a mainframe com             65   memory integrated circuit (“IC”), or memory portion of an
puter), comprising an input and output (I/O) interface 105,             integrated circuit (such as the resident memory Within a pro
one or more processors 115, and a memory 120 storing a                  cessor IC), including Without limitation RAM, FLASH,
                                                      US 8,346,593 B2
                               9                                                                       10
DRAM, SRAM, MRAM, FeRAM, ROM, EPROM or                                  ered integrated in that a plurality of different tables or types of
EZPROM, or any other type of memory, storage medium, or                 tables, objects or relations are included Within the database
data storage apparatus or circuit, Which is known or Which              100, such as including an attitudinal classi?cation table 240
becomes known, depending upon the selected embodiment.                  With the other illustrated tables discussed beloW. While gen
  In the ?rst system 110, the I/O interface may be imple                erally not included Within the database 100 (as potentially
mented as knoWn or may become knoWn in the art. The ?rst                private client data), optionally one or more copies of a
system 110 and second system 150 further include one or                 selected population data (?le, table or database) 270, such as
more processors, such as processor 115 illustrated for ?rst             a client customer databases, client customer ?at ?les, or client
system 110. As the term processor is used herein, these imple           master databases, may also be utiliZed. (Use of any type of
mentations may include use of a single integrated circuit               data repository, Whether an integrated database, a non-inte
(“IC”), or may include use of a plurality of integrated circuits        grated database, or any otherWise distributed or non-distrib
or other components connected, arranged or grouped                      uted database structures or schemas, are Within the scope of
together, such as microprocessors, digital signal processors            the present invention. While referred to as tables, it should be
(“DSPs”), custom ICs, application speci?c integrated circuits           understood that the tables illustrated in the database 100 of
(“ASICs”), ?eld programmable gate arrays (“FPGAs”), adap                FIG. 2 are to be construed broadly, to mean and include
tive computing ICs, associated memory (such as RAM and                  relations, objects, object relations, multidimensional rela
ROM), and other ICs and components. As a consequence, as                tions, cubes, ?at ?les, or other similar or equivalent database
used herein, the term processor should be understood to                 constructs.) In addition, While a plurality of relations (or
equivalently mean and include a single IC, or arrangement of            connections) betWeen and among the various tables are illus
custom ICs, ASICs, processors, microprocessors, controllers,       20   trated in FIG. 2, it should be understood that in any selected
FPGAs, adaptive computing ICs, or some other grouping of                embodiment, a greater or feWer number of relations, connec
integrated circuits Which perform the functions discussed               tions, cross-references, keys, or indices may be utiliZed, all
beloW, With associated memory, such as microprocessor                   Within the scope of the present invention.
memory or additional RAM, DRAM, SRAM, MRAM, ROM,                           The database 100 generally includes, for example, a name
EPROM or EZPROM. A processor (such as processor 115),              25   table 205, an address table 210, a lifestyle and behavioral data
With its associated memory, may be adapted or con?gured                 table 215, and a demographic data table 220, and depending
(via programming, FPGA interconnection, or hard-Wiring) to              upon the selected embodiment, may also include a public
perform the methodology of the invention, as discussed above            record table 225, a census data table 230, a summariZed
and as further discussed beloW. For example, the methodol               ?nancial data table 235, and other information tables 265. In
ogy may be programmed and stored, in a processor With its          30   various embodiments, the name table 205 and address table
associated memory (and/or memory 120) and other equiva                  210 may be combined as a single table. In the exemplary
lent components, as a set of program instructions (or equiva            embodiments, the database 100 further includes an attitudinal
lent con?guration or other program) for subsequent execution            (or behavioral) classi?cation table 240, a message theme
When the processor is operative (i.e., poWered on and func              table 245, a communication media (or channel) table 250, a
tioning). Equivalently, When the ?rst system 110 and second        35   communication timing table 255, a communication fre
system 150 may implemented in Whole or part as FPGAs,                   quency table 260, a communication sequence table 285, a
custom ICs and/or ASICs, the FPGAs, custom ICs or ASICs                 response data table 275, and a transactional data table 280, as
also may be designed, con?gured and/ or hard-Wired to imple             discussed in greater detail beloW. While illustrated as separate
ment the methodology of the invention. For example, the ?rst            tables or relations, it should be understood that the informa
system 110 and second system 150 may implemented as an             40   tion contained in such tables may be contained or distributed
arrangement of microprocessors, DSPs and/orASICs, collec                betWeen or among any number of tables or relations, depend
tively referred to as a “processor”, Which are respectively             ing upon any applicable or selected schema or other database
programmed, designed, adapted or con?gured to implement                 100 structure, in any number of equivalent Ways, any and all
the methodology of the invention, in conjunction With a data            of Which being Within the scope of the present invention. The
base 100.                                                          45   data repository 100 is generally included Within a ?rst system
  The application server 125, database management server                110 and/or second system 150, and respectively accessed
140, and the system 110 may be implemented using any form               through the I/O 105 and processor 115, or an application
of server, computer or other computational device as knoWn              server 125 or database management server 140, discussed
or may become knoWn in the art, such as a server or other               above.
computing device having a processor, microprocessor, con           50     The name table 205 contains all individual, consumer,
troller, digital signal processor (“DSP”), adaptive computing           household, living unit, group or other entity names, in various
circuit, or other integrated circuit programmed or con?gured            forms, variations, abbreviations, and so on, and is also uti
to perform the methodology of the present invention, such as            liZed for searching and matching processes, as discussed
a processor 115, as discussed in greater detail beloW.                  beloW. The address table 210 contains all addresses of indi
  FIG. 2 is a block diagram illustrating an exemplary data         55   viduals, households, living units, groups or other entities
repository (or database) 100 in accordance With the present             Which Will be utiliZed for searching and matching processes,
invention. As mentioned above, “data repository” as used                as discussed beloW. In the exemplary embodiment, the lif
herein, is considered interchangeable With “database”, and              estyle and behavioral data table 215 contains lifestyle and
may be relational, object-oriented, or object-relational, or            behavioral information for individuals, households, living
utiliZe any other database structure, in accordance With a         60   units, or other groups or consumers, as “entities”, such as
selected embodiment. The database 100 may be integrated,                purchase behavior, activity data, and any self-reported data.
namely, that the information resides Within a singular, co              The demographic data table 220 contains demographic infor
located or otherWise centraliZed database structure or                  mation and possibly geodemographic information for con
schema, or may be a distributed database, With information              sumers and other individuals, households, living units, or
distributed betWeen and among a plurality of databases, some       65   other entities, such as age, gender, race, religion, household
of Which may be remotely located from the other databases.              composition, income levels, career choice, etc. The public
From another point of vieW, the database 100 may be consid              record table 225 contains information available in public
                                                        US 8,346,593 B2
                               11                                                                       12
records, such as vehicle ownership records, driving records,              table 200 may be utilized in the searching and matching
property ownership records, public proceeding records,                    processes discussed beloW, and for other database applica
secured transaction records, and so on. The census data table             tions, such as updating.
230 contains census information, typically available through                 Figure (or “FIG.”) 3 is a How diagram illustrating an exem
a government agency. The summarized ?nancial data table                   plary method for determination of ?rst predictive (attitudinal)
235, When included in database 100, typically includes sum                classi?cations and second predictive (message content) clas
maries of ?nancial information generally for individuals or               si?cations using a data repository in accordance With the
households in a given geographic region (e.g., by postal                  present invention. As indicated above, in accordance With the
code), and could possibly also include bank account informa               exemplary embodiments, such ?rst predictive classi?cations
tion, investment information, securities information, and                 are attitudinal and derived using empirical attitudinal
credit or other private information, When available and to the            research; in other embodiments, the ?rst predictive classi?
extent alloWable under any applicable regulations or laWs.                cations may be behavioral, demographic, or any combination
   The transactional data table 280 typically contains infor              thereof, and derived from corresponding empirical research,
mation concerning purchase history or other transaction his               such as behavioral or demographic survey research. The sec
tory of the various entities. The response data table 275 con             ond predictive classi?cations are determined independently,
tains information typically related to transactional data, such           and in the exemplary embodiments, provide message content
as for purchases made in response to a particular communi                 or message theme classi?cations, Which are subsequently
cations, such as in response to a catalogue, direct mail, or a            utilized to determine the “What” of a marketing or educational
magazine advertisement. The transactional data table 280 and              communication, for example.
the response data table 275, for example, may be based upon          20     The method begins, start step 300, With results of one or
data from particular clients or groups of clients. The informa            more attitudinal (or behavioral) surveys of individuals, as
tion contained in the lifestyle and behavioral data table 215,            consumers or as members of a living unit (household). As
the demographic data table 220, the public record table 225,              mentioned above, as used herein, “entity” or “entities” should
the census data table 230, and the summarized ?nancial data               be understood to mean and include any individual person,
table 235, other information tables 265, and any other avail         25   household, living unit, group or potential grouping of one or
able tables, depending upon the selected embodiment, are                  more people, Whether related or unrelated, individually or
utilized in the creation of independent variables for the pre             collectively, such as a single individual, a household, a living
dictive attitudinal (or behavioral) modeling discussed beloW.             unit, a geographic unit, or any other grouping of individuals
The selected population data 270, Which may be in any of                  for Which data may be maintained, generally at a granular or
various forms such as a table, a ?le, a ?at ?le, a relation, a       30   atomic level in a database. For example, various databases
database, or another data schema, such as a copy of a selected            may be maintained in Which information is stored and avail
customer database, contains names or names and addresses of               able at an individual level, for individual persons and, in many
individuals, households, living units, other groups or entities,          cases, also maintained at a less granular level of living units
such as customers or any other selected or designated popu                (households), While in other cases, also Within the scope of
lation, and is utilized to provide predictive attitudinal market          the present invention, information may be stored and avail
ing information for the selected population, as discussed in              able in a database only at a household level, With the predicted
greater detail beloW. Also optionally included Within database            attitudinal and messaging classi?cations then pertaining to
100 are other information tables 265, such as for other demo              corresponding households (as a larger grouping of one or
graphic information, credit information, and fraud informa                more individuals). In other cases, an individual residing at a
tion, When available or authorized.                                  40   ?rst location may also be considered to be part of a living unit
   The attitudinal classi?cation table 240, the message theme             at a second location, such as a student residing in a college
table 245, the communication channel table 250, the commu                 dormitory being considered part of a family household resid
nication timing table 255, the communication frequency table              ing at a different location. All such variations are Within the
260, and the communication sequence table 285, are gener                  scope of the present invention and, for ease of reference, any
ally created, populated or segregated based upon the predic               references to an entity or entities means and includes any
tive attitudinal (or behavioral) modeling discussed beloW                 individual, person, or grouping or collection of persons, hoW
With reference to FIG. 3.                                                 ever such a grouping may be de?ned or demarcated.
   Depending upon the selected database 100 embodiment, a                   In the exemplary embodiments, an empirical modeling
table or index (relation or look-up table) 200 of identi?ers or           process is performed in Which selected questions from a
identi?cations (“IDs”) of a plurality of individuals, consum         50   survey are utilized for obtaining information pertinent to con
ers, households, living units, other groups or entities, may be           sumer purchasing, behaviors, and attitudes (Which also may
included Within the database 1 00. The identi?ers are typically           be considered to include various behavioral and demographic
persistent, With every entity assigned at least one ID. In the            components). Survey questions, data and results are available
exemplary embodiments, the ID table 200 also provides rela                from a Wide variety of vendors, publications, and other
tions, links or cross-references to a plurality of other relations        sources. Survey questions may also be determined based
or tables, such as, for example, the name table 205, the                  upon the goals of the modeling and classi?cation processes.
address table 210, the lifestyle and behavioral data table 215,           The survey may be conducted and results obtained in any of
the demographic data table 220, the public record table 225,              various forms, such as via telephone, Written survey, email
the census data table 230, the summarized ?nancial data table             survey, mail survey, internet survey, personal intervieW, etc.
235, the attitudinal classi?cation table 240, the message                 The selected questions from the survey are then subjected to
theme table 245, the communication channel table 250, the                 a factor analysis, step 305, to statistically determine Which
communication timing table 255, the communication fre                     questions are highly related to or among other survey ques
quency table 260, the communication sequence table 285,                   tions, to What degree, and to isolate the signi?cant questions
and the other information tables 265. The ID table 200 may                and determine corresponding factors, step 510, With a
also provide relations, links or cross-references to selected             selected group of highly related survey questions forming a
population data (?le, table or database) 270, depending upon              selected, corresponding factor. This factor analysis may also
the selected embodiment and the form of the data. The ID                  be an iterative process in selected embodiments. The resulting
                                                       US 8,346,593 B2
                              13                                                                       14
plurality of empirical attitudinal factors identify attitudes,           records Within the database 100, step 320, generally utilizing
behaviors and motivations of the survey participants, and                the name table 205 and address table 210. For all entities such
identify the signi?cant corresponding survey questions. The              as individuals, households, living units, or other groups hav
plurality of empirical attitudinal factors Which are selected            ing matching records (i.e., matching individuals/households),
depend upon the selected purposes of the classi?cations dis              demographic, lifestyle, behavioral, and other variables (from
cussed beloW, such as marketing analyses, for example, and               the database 100) are appended or linked to each (matching)
are also dependent upon the selected purposes of the survey              survey participant, step 325, such as variables from the lif
(as are the resulting attitudinal and message content classi?            estyle and behavioral data table 215, the demographic data
cations, discussed beloW), and the cultural, sociological,               table 220, the public record table 225, the census data table
demographic, and other characteristics of the sample popu                230, the summarized ?nancial data table 235, the transac
lation of the survey.                                                    tional data table 280, and the response data table 275, for
  In an exemplary embodiment, a plurality of empirical atti              example. In the exemplary embodiment, the Experian
tudinal factors Were developed as a result of the factor analy           INSOURCE® database, as previously described, Was uti
sis and empirical attitudinal model determinations of steps              lized as the database 100. Using the appended database vari
305 and 310. Exemplary factors for marketing purposes                    ables as independent variables, and using the empirical atti
include, for example, consumer brand loyalty, impulse buy                tudinal factor scores as dependent variables, a predictive
ing behavior, incentive driven behavior, and so on. Innumer              attitudinal model (or, equivalently, an attitudinally predictive
able other factors and corresponding questions Will be appar             model) is developed for each empirical attitudinal factor, step
ent to those of skill in the art, for other behavioral or                330, thereby generating a corresponding plurality of predic
demographic modeling, for example. Corresponding exem               20   tive attitudinal models, one for each empirical attitudinal
plary statements are illustrated beloW, to Which a sample                factor. In the exemplary embodiment, a logistic regression
population Was asked to agree or disagree, on a varying scale,           analysis is performed, to identify the database variables (as
from highly agree, someWhat agree, neutral, someWhat dis                 independent variables) Which are statistically signi?cant pre
agree, and highly disagree, resulting in an equivalent question          dictors of the attitude of the corresponding empirical attitu
format. The sample questions are exemplary, for purposes of         25   dinal factor. Other statistical methods, such as multiple linear
illustration only, and any resulting set of attitudinal, behav           regression analysis, other forms of regression analysis, and
ioral or demographic models (discussed beloW) Will be                    other forms of modeling and statistical analysis, are also
empirically determined based upon the purpose of the                     considered equivalent and Within the scope of the present
research survey, the selected survey questions and results, the          invention. Selection of given database variables is also a
survey population and culture, folloWed by factor analysis.         30   function of the availability of such variables Within the data
Additional survey questions may also be utilized to expand               base 100, namely, any given database may or may not include
the attitudinal factors utilized. Exemplary statements, utilized         variables available in other databases. Not separately illus
in a representative survey of the present invention, include,            trated in step 330, the plurality of attitudinally predictive
for example:                                                             models may also be validated, such as by using a “holdout”
  If a product is made by a company I trust, I’ll buy it even if    35   (or separate) sample from the survey results.
     it is slightly more expensive.                                        Using the plurality of predictive attitudinal models of step
  I am Willing to pay more for a product that is environmen              330, all or most database entities (or members), namely, all or
     tally safe.                                                         most individuals, consumers, households, living units, or
  I like to shop around before making a purchase.                        other groups contained in the database (i.e., contained in the
  I’m alWays one of the ?rst of my friends to try neW products      40   database by having representative information in the database
     or services.                                                        100), are scored (or otherWise evaluated) across each of the
  I prefer products that offer the latest in neW technology.             plurality of predictive attitudinal models, step 335. Having
  If I really Want something I Will buy it on credit rather than         been scored/evaluated, these entities (e.g., individuals or
     Wait.                                                               households) then form a reference population, utilized for
  I’d rather receive a sample of a product than a price-off         45   comparative purposes discussed beloW. The results from each
     coupon.                                                             such entity (individual or group) being scored or evaluated
  Following development of the plurality of empirical atti               based on each of the predictive attitudinal models, in step 335,
tudinal factors of steps 305 and 310, all of the survey partici          then form or represent (or otherWise generate or determine) a
pants, as a sample population, are scored across each of the             corresponding plurality of predictive message content classi
empirical attitudinal factors, step 315. For example, various       50   ?cations, also referred to as predictive message theme clas
survey participants may have indicated various levels of                 si?cations, for each such entity (individual or group) repre
agreement (or disagreement) With the survey questions of a               sented in the database 100. More speci?cally, using the
particular attitudinal factor, and as such, Would be scored high         plurality of predictive attitudinal models, each entity (indi
(or loW) for that particular attitudinal factor. The scoring             vidual, household, living unit, or other group) represented in
process may also be computed in probabilistic terms, as a           55   the database 100 is predicted to have a corresponding prob
probability of exhibiting a particular attitude. As a conse              ability of belonging in or to a classi?cation associated With a
quence, in the exemplary embodiment, each survey partici                 particular attitude/behavior/ demographic of interest, in
pant Was scored across the plurality of empirical attitudinal            Which entities (individuals, households, living units, or other
factors. Also for example, various survey participants may               groups) exhibiting that attitude/behavior/demographic are
have similar scores across a plurality of empirical attitudinal     60   generally responsive, receptive, attentive, conducive to or
factors, such as high scores for the same ?rst factor, and loW           motivated by messages or other communications having a
scores for the same second factor.                                       particular content or theme. Such content or themes, for
   Following scoring of the survey participants across each of           example, to Which individuals or groups may be receptive,
the empirical (attitudinal) factors in step 315, records in the          may be matched, correlated or derived from various informa
database 100 pertaining to the survey participants, as indi         65   tion sources, such as information contained in the lifestyle
viduals (or entities such as living units), are searched, and the        and behavioral data table 215, the response data table 275,
survey participants are matched With their corresponding                 and the transactional data table 280. As a consequence, the
                                                         US 8,346,593 B2
                                15                                                                        16
results from the evaluations using the predictive attitude/                 cations Were developed as a result of the analysis of step 335.
behavior/ demographic models provide or form the corre                      Several examples of predictive message content classi?ca
sponding predictive message content classi?cations for each                 tions are illustrated beloW, With corresponding, exemplary
entity (individual, household, living unit, or other group) of              message content guidelines. It Will be understood in the art
the reference population.                                                   that these predictive message content classi?cations are
  These classi?cations are referred to as predictive message                exemplary and for purposes of illustration and not limitation,
content classi?cations (or message theme classi?cations)                    and that any resulting set of predictive message content clas
because, as discussed beloW, they are utiliZed to predict the               si?cations Will be empirically determined based upon the
message content or message themes to Which individuals,                     selected survey purposes; the selected survey questions and
households, living units, or other groups Within that classi?               results; the survey or sample population, demographics,
cation are likely to be receptive or responsive. As indicated               socioeconomics and culture; the plurality of predictive atti
above, the actual result of the scoring may be a probability of             tudinal models; and the extrapolated database population.
exhibiting the attitude in question, or may be a number or                  Exemplary Predictive Message Content (or Theme) Classi?
percentile Which can be equivalently translated into such a                 cations:
probability. In the exemplary embodiment, the probability                     First Exemplary Predictive Message Content (or Theme)
scores Were further classi?ed into nine tiers, Which Were then              Classi?cation: Exemplary message content guidelines
further utiliZed to create dichotomous variables in order to                include reWarding and complimenting for being the ?rst to
classify the entity (individual or group) as either exhibiting              take advantage of neW products and services, highlighting
the attitude of interest or not exhibiting the attitude of interest.        neW or cutting edge products or offers, and demonstrating the
For each such entity (individual or group), the results of step        20   prestige of the product/ service offered.
335, namely, the scores for each of the predictive attitudinal                Second Exemplary Predictive Message Content (or
models and/or each of the resulting predictive message con                  Theme) Classi?cation: Exemplary message content guide
tent classi?cations, are stored in the database, step 340, such             lines include communicating the strength and quality of a
as in message theme table or relation 245.                                  brand, the importance of relationships and customer service,
  For example, using the scores or evaluations from the plu            25   emphasizing the quality of a product, emphasiZing the num
rality of predictive attitudinal models for individual (or                  ber of years in business, and integrity and quality aWards.
group) “A”, records may be stored indicating that “A” has a                    Third Exemplary Predictive Message Content (or Theme)
high probability level of belonging to predictive message                   Classi?cation: Exemplary message content guidelines
content classi?cations “X”, “Y”, and “Z”, and a loW prob                    include a family focus, bonuses, presenting hoW a product/
ability level of belonging to each of the remaining predictive         30   offer is better than a competitive product/ service, price com
message content classi?cations. Alternatively and equiva                    parison, and value features.
lently, records for individual (or group) “A” may be stored                   Fourth Exemplary Predictive Message Content (or Theme)
indicating that “A” has certain scores from the evaluations                 Classi?cation: Exemplary message content guidelines
under each of the predictive attitudinal models, and belongs to             include appealing to altruism, activism, and appreciation for
predictive message content classi?cations “X”, “Y”, and “Z”            35   our ecology, the use of natural ingredients, and emphasis on
(With dichotomous variables of “l”), and does not belong to                 quality With details.
each of the remaining predictive message content classi?ca                    Fifth Exemplary Predictive Message Content (or Theme)
tions (With dichotomous variables of “0”). As a consequence,                Classi?cation: Exemplary message content guidelines
depending upon the selected embodiment, all or most entities                include use of celebrity endorsements and testimonials to
represented in the database 100 have associated scores for             40   emphasiZe image and style, and use of incentive gifts.
each of the plurality of predictive attitudinal models and,                   Sixth Exemplary Predictive Message Content (or Theme)
correspondingly, a membership (or no membership), or a                      Classi?cation: Exemplary message content guidelines
degree or probability of membership, in each of the corre                   include demonstrating a fair value using a straightforWard,
sponding plurality of predictive message content (or theme)                 logical approach, a masculine emphasis, and use of peer/user
classi?cations.                                                        45   comparisons and testimonials.
  It should be noted that in the exemplary embodiments, the                    Referring again to FIG. 3, folloWing scoring of all or most
empirical attitudinal factors and predictive message content                entities (individuals or groups) in the database 100 using the
classi?cations have a one-to-one correspondence, and may be                 plurality of predictive attitudinal models, concomitant
very similar. In other embodiments, there may be more or                    assignment of membership (or non-membership) of the enti
feWer predictive message content classi?cations compared to            50   ties (individuals or groups) to the corresponding plurality of
empirical attitudinal factors. As indicated above, the empiri               predictive message content classi?cations in step 335, and
cal attitudinal factors are based on the factor analysis of the             storing the resulting information in the database 100 of step
survey questions from the sample population and are utiliZed                340, the exemplary method of the invention performs a clus
to develop the predictive attitude/behavior/demographic                     ter or grouping analysis of all such database members (enti
models incorporating the database variables. The predictive            55   ties) using either (or both) the corresponding scores from
attitude/behavior/demographic models derived from the                       each of the plurality of predictive attitudinal models or the
empirical analysis of the sample population are then extended               resulting assigned membership(s) (or probability of member
into the database population, as the reference population. The              ship(s)) in the predictive message content classi?cations, step
results from this predictive modeling are then matched or                   345. This cluster or grouping analysis of step 345 not only
correlated With other database information to create the cor           60   utiliZes the plurality of predictive message content classi?ca
responding predictive message content classi?cations. In                    tions (and/or predictive attitudinal models), but also utiliZes
addition, given the appended database variables, various                    combinations of the various predictive message content clas
demographic, lifestyle and behavioral characteristics may                   si?cations (or, equivalently, scores from the corresponding
also be included in or as part of the descriptions of the pre               predictive attitudinal models). Any form of cluster or group
dictive message content (or theme) classi?cations.                     65   ing analysis may be utiliZed, as knoWn or may become knoWn
   In the exemplary embodiments, a plurality of representa                  in the ?eld. The result of this cluster or grouping analysis is a
tive, predictive message content (or message theme) classi?                 plurality of predictive attitudinal classi?cations. For example,
                                                       US 8,346,593 B2
                              17                                                                       18
those entities that belong or are assigned to the same tWo               in the database 100, namely: (1) either or both the associated
predictive message content classi?cations, or equivalently               scores (results) for each of the plurality of predictive attitu
those entities that scored high in the same tWo corresponding            dinal models and/or, correspondingly and equivalently, a
predictive attitudinal models, may be clustered or grouped               membership (or no membership) or a degree or probability of
together into a ?rst predictive attitudinal classi?cation. Also          membership in the corresponding plurality of predictive mes
for example, those entities Who belong in one predictive                 sage content (or theme) classi?cations (message theme table
message content classi?cations and Who do not belong in                  245); and (2) an assignment into a predominant, predictive
another predictive message content classi?cation may be                  attitudinal classi?cation (attitudinal classi?cation table 240).
clustered or grouped together into a second predictive attitu            Following step 355, the method of determination of predic
dinal classi?cation. In addition, clusters may exist for those           tive attitudinal classi?cations and predictive message content
Who belong in only one predictive message content classi?                classi?cations using a data repository, in accordance With the
cation.                                                                  present invention, may end, return step 360.
  For example, for the plurality of predictive attitudinal clas            In the exemplary embodiment, a plurality of predictive
si?cations described beloW: a ?rst exemplary cluster exhib               attitudinal classi?cations Were developed as a result of the
ited both a “trend following” attitude and an “impulsive”                analysis of step 345, and representative examples are illus
attitude, but not an “incentive driven” attitude; and a second           trated immediately beloW. For each such exemplary predic
exemplary cluster exhibited an “environmentally conscious”               tive attitudinal classi?cation, corresponding marketing strat
attitude, a “brand loyal” attitude, and a “buy American” atti            egies, lifestyle and interests, demographics, behaviors and
tude, but not a “price conscious” attitude.                              attitudes, and socioeconomic indicators are illustrated, gen
   As a result of the cluster (segmentation or grouping) analy      20   erally derived from corresponding database variables and
sis of step 345, a plurality of predictive attitudinal classi?ca         other information available in a database 100, as Well as
tions are developed, generally having a greater number of                syndicated survey research. It Will be understood in the art
classi?cations than the plurality of predictive message con              that these predictive attitudinal classi?cations and their mar
tent classi?cations, and providing higher granularity or dis             keting names are exemplary and for purposes of illustration
crimination among the various attitudes/behaviors/demo              25   and not limitation, and that any resulting set of predictive
graphics exhibited among the database reference population.              attitudinal classi?cations Will be empirically determined
In the exemplary embodiments, using membership or non                    based upon the selected survey questions and results; the
membership in the plurality of predictive message content                survey population, demographics, socioeconomics and cul
classi?cations (based on probability scores from the predic              ture; the plurality of predictive attitudinal models; the
tive attitudinal models), clusters Were identi?ed Where the         30   extrapolated database population; and the selected cluster
mean value of the dichotomous variable Was 0.70 or higher,               analysis.
indicating a segment that had at least one strong loading.               Exemplary Predictive Attitudinal (or Behavioral) Classi?ca
   FolloWing the cluster analysis, in step 350, each entity              tions:
(individual or group) represented in the database is assigned              First Exemplary Predictive Attitudinal (or Behavioral)
to a predominant predictive attitudinal (behavioral or demo         35   Classi?cation: Individuals and households in this ?rst predic
graphic) classi?cation, of the plurality of predictive attitudi          tive attitudinal classi?cation stay true to themselves and the
nal (behavioral or demographic) classi?cations, based upon               brands that they prefer. They are selective With their pur
his, her or its highest probability of exhibiting the attitude(s)        chases, and look for Well-established products and services
(behaviors or demographics) of interest of the corresponding             that have demonstrated quality and value. Individuals and
classi?cation. This assignment may be determined equiva             40   households in this ?rst predictive attitudinal classi?cation are
lently by the entity’s scores from the predictive attitudinal            responsive to brand extensions and use coupons on the prod
models and/or the correspondingly determined memberships                 ucts that they already have an af?nity toWard. From the data
in one or more predictive message content classi?cations. For            base 100, their lifestyle and interests include enjoyment of
example, individuals or groups predicted to exhibit only a               reading and visits to bookstores; television vieWing and pre
single attitude of interest Would be assigned to that corre         45   ferring informative programming and movie classics; invest
sponding predictive attitudinal classi?cation (or cluster),              ing Wisely and often; maintaining an exercise and ?tness
While those exhibiting more than one attitude of interest                regimen; and participation in activities such as golf, tennis,
Would be assigned to a corresponding predictive attitudinal              ?shing, and occasional gambling. Also from the database
classi?cation, as a cluster of those particular of attitudes. In         1 00, their demographics include being established mid-lifers;
the exemplary embodiment, those entities (using dichoto             50   married, divorced or single; any children are groWn and have
mous variables or “all or none” scores for the predictive                left home; they typically oWn their oWn homes, and have
attitudinal models) not assigned at described above are then             established residences, usually in larger, a?luent cities. The
re-clustered to identify an optimal segment or cluster, Which            behaviors and attitudes of the individuals and households in
may not meet stricter scoring requirements, but nonetheless              this ?rst predictive attitudinal classi?cation include being
indicate a predominant, predictive attitudinal classi?cation.       55   ardent catalog shoppers; having a preference for outdoor
Also in the exemplary embodiments, an entity is assigned to              lifestyle companies; shopping at upscale retail stores; prefer
one and only one predictive attitudinal classi?cation; in other          ring “the real thing” to generic products; visiting the grocery
embodiments, multiple predictive attitudinal/behavioral/de               store frequently With a likelihood of using coupons to save on
mographic classi?cations may be assigned.                                preferred brands; and enjoyment of domestic and overseas
  As is the case With the scores from the plurality of predic       60   travel. Their socioeconomic indicators include a high
tive attitudinal models and the corresponding assignments to             income; an above average home value; established credit
the plurality of predictive message content classi?cations,              experience With a Well-maintained, stable credit history; an
such assignments of predictive attitudinal classi?cations are            undergraduate degree and some graduate studies; occupa
also stored in the database 100, step 355, such as in attitudinal        tions including ?nance, accounting, engineering and real
classi?cation table or relation 240. As a consequence, all or 65         estate; and they drive luxury vehicles.
most entities represented in the database 100, in accordance                Second Exemplary Predictive Attitudinal (or Behavioral)
With the present invention, have a plurality of records stored           Classi?cation: Individuals and households in this second pre
                                                       US 8,346,593 B2
                               19                                                                     20
dictive attitudinal classi?cation represent a highly af?uent,            they typically drive used, domestic vehicles, and models
successful and stable consumer market, containing estab                  include small to mid-siZe cars and small- and full-siZe pickup
lished old-Wealth and the nouveau riche. Their investments               trucks.
and dividends are as impressive as their incomes. They aspire              Fourth Exemplary Predictive Attitudinal (or Behavioral)
to oWn and use the ?nest quality brands and services, and they           Classi?cation: Individuals and households in this fourth pre
are Willing to pay the extra dollar for the privilege of living          dictive attitudinal classi?cation are conservative, content
this lifestyle. They enj oy traveling quite extensively, so incen        With the status quo and not easily sWayed. They focus on
tives that provided added bene?t in this area are preferable.            “hearth and home” for comfort and entertainment, avidly
The active lifestyles they lead drive them to utiliZe all modes          donate to the causes they support, and enjoy timeless activi
of convenient communication. The lifestyle and interests of              ties such as leisure sports, musical performances, gardening,
individuals and households in this second predictive attitudi            and reading. As consumers, they are motivated to spend
nal classi?cation include a love to travel domestically and              money on their families, homes and hobbies but are careful to
overseas, preferring cruises and tours; shopping at mid-level            spend it Well, making them highly responsive to coupons and
to upscale stores; diverse sports interests and may be avid              discount offers. Their lifestyle and interests include family
golfers; socially involved as club members, theatre and con              oriented, domestic activities such as home improvement
cert- goers, and With environmental causes. Their demo graph             projects, gardening, cooking and entertaining. They are typi
ics include a Wide age range, from young to mature adults;               cally passionate donors that support causes such as religious,
largest concentration is established and mid-life adults, Who            political and health issues. They are also devoted book and
are typically married; their children range in age from grade            magaZine lovers and sports enthusiasts. The demographics of
school to high school; they typically oWn their oWn homes,          20   the individuals and households in this fourth predictive atti
having Well-established residences, usually in comfortable               tudinal classi?cation are that they are mainly seniors and
and prosperous neighborhoods, in major and mid-siZe cities,              retirees, typically married, Whose children have left home
and in urban city settings. The behaviors and attitudes of the           (empty nesters). They typically oWn their oWn homes, usually
individuals and households in this second predictive attitudi            multi-dWelling units rather than single-family homes, and
nal classi?cation include “Working to live” rather than “living     25   prefer to live in rural toWns and small city communities. The
to Work”; they are active, a?luent, have an in?uential lifestyle         behaviors and attitudes of the individuals and households in
and are ?nancially astute; they make time for family and                 this fourth predictive attitudinal classi?cation include a
individual interests, and Want the “good life” for their family;         relaxed living attitude, With a healthy standard of living,
they are technology- and internet-savvy, With frequent Web               deriving signi?cant pleasure from daily activities With family
accessing. The socioeconomic indicators include a high              30   and friends. They make the mo st of their spending and utiliZe
income; an above average home value; extensive, established              coupons. They like to keep up on interests in music, trivia and
and good credit experience; they have an undergraduate                   collectibles. Socioeconomic indicators for this classi?cation
degree With some graduate studies; their occupations include             include a loW income, With an average to beloW average home
?nance, engineering, healthcare, counseling, computer/tech               value; stable, consistent and capable credit experience; they
nology and marketing; they are more likely to lease vehicles        35   are typically high school graduates With some college; and
than to buy; drive neW and used import cars and light trucks,            primarily are retired. They typically drive domestic used
and are draWn to near-luxury, luxury, specialty and SUV                  vehicles that include mid-range cars and pick-up trucks
models.                                                                     The plurality of predictive attitudinal classi?cations and
  Third Exemplary Predictive Attitudinal (or Behavioral)                 plurality of predictive message content classi?cations, With
Classi?cation: Individuals and households in this third pre         40   additional information available in a database 100 as dis
dictive attitudinal classi?cation are dedicated sports fans that         cussed beloW, become extraordinarily poWerful tools When
enjoy a Wide variety of outdoor pursuitsifrom do-it-yourself             applied to a selected population, such as a group of individu
home improvement projects to scuba diving, and they enjoy                als represented in a customer database, a prospect database, a
their lifestyle. Their independence may make it a challenge to           client database, a membership database, an association data
establish relationships With these customers, and they prefer       45   base, and so on. In the exemplary embodiment, the additional
product samples to coupons to provide immediate proof of the             information available in the database 100 includes, for all or
product’s quality and immediate savings. Their lifestyle and             mo st of the represented (or matched) individuals, households,
interests include being outdoor enthusiasts; they have no pref           living units or other entities: their preferred methods of com
erence for brand name goods over generic brands; they are                munication and/or communication media (communication
dedicated sports fans, they enjoy Working on mechanics,             50   media table 250), their preferred times (time of day) of com
home improvement, boating, motorcycles, scuba diving and                 munications (communication timing table 255), their pre
video games. The demographics of individuals and house                   ferred frequencies of communication (communication fre
holds in this third predictive attitudinal classi?cation include         quency table 260), and their preferred sequences of
mainly being young adults, Who are single or divorced, With              communication (communication sequence table 285). This
loW indications of children present in the household; they          55   additional information may be determined in a Wide variety of
typically rent instead of oWn residences, and live in apart              Ways, including self-reported preferences and behaviors,
ments rather than single-family homes, With a Wide variety of            third-party reported preferences and behaviors (such as trans
residential settings, often transient or in rural toWns. Individu        actions, purchases, and activities), observed preferences and
als and households in this third predictive attitudinal classi           behaviors, and inferred preferences and behaviors based on
?cation have behaviors and attitudes such as liking things to       60   modeled data.
be simple and straightforWard, “rough and rugged”, and self                 Figure (or “FIG.”) 4 is a How diagram illustrating an exem
determining. Socioeconomic indicators of individuals and                 plary method of independently predicting a plurality of atti
households in this third predictive attitudinal classi?cation            tudinal classi?cations and a plurality of message content clas
include a beloW average income; a slightly beloW average                 si?cations of a selected population in accordance With the
home value; a neWer credit experience With average exten            65   present invention. In addition, depending upon the selected
sion; varied education levels; typically employed in service             embodiment, predicted communication and/or media chan
and consumer-oriented industries and/or may be students;                 nels, predicted communication timing, predicted communi
                                                       US 8,346,593 B2
                              21                                                                       22
cation frequency, and predicted communication sequencing,                predictive attitudinal classi?cation of the plurality of predic
may also be provided as part of the method illustrated With              tive attitudinal classi?cations, a “penetration” or comparative
reference to FIG. 4.                                                     index (or rate) is determined, step 425, With a comparatively
  Referring to FIG. 4, the method begins, start step 400, With           greater or higher penetration index indicative of a higher
data about or concerning a selected population, such as name             proportional concentration of entities of the selected popula
and/ or name and address information from a customer data                tion Within a given predictive attitudinal classi?cation com
base or ?le, a customer prospect ?le or list, or any other               pared to the reference distribution, and With a comparatively
identifying data or information of or for a group of individu            loWer or lesser penetration index indicative of a loWer pro
als, households, living units, groups or other entities, for any         portional concentration of entities of the selected population
selected purpose. The database 100 is then searched and the
                                                                         Within a given predictive attitudinal classi?cation compared
selected population data is matched With the records of the
                                                                         to the reference distribution. For example, a 15% distribution
database 100, step 405, such as matched With the records of
the INSOURCE® database. For all records Where a match is
                                                                         of the selected population for the predictive attitudinal clas
found in step 405, the method appends, references or links, to           si?cation of “Q”, When compared to an 8% distribution for the
each (matched) entity of the selected population, their corre            reference population for this same “Q” predictive attitudinal
sponding (i.e., predominant) predictive attitudinal classi?ca            classi?cation, indicates a comparatively higher (or above
tion, and their corresponding predictive message content                 average) penetration index or rate (a ratio of 1.875) of the
classi?cation(s), step 410. As discussed above, for each such            selected population in this classi?cation. Similarly, an 11%
entity, their corresponding predictive attitudinal classi?cation         distribution of the selected population for the predictive atti
is generally their predominant attitudinal classi?cation of the     20   tudinal classi?cation of “P”, When compared to a 16% distri
plurality of predictive attitudinal classi?cations, and their            bution for the reference population for this same “P” predic
corresponding predictive message content classi?cations are              tive attitudinal classi?cation, indicates a comparatively loWer
generally their memberships or probabilities of membership               (or beloW average) penetration index or rate (a ratio of
in each of the plurality of predictive message content classi            0.6875) of the selected population in this classi?cation.
?cations. In the exemplary embodiments, optionally as part of       25      In addition, for each predictive attitudinal classi?cation,
step 410, the method also appends, references or links the               the reference distribution may be normaliZed to a particular
entity’s associated information concerning predicted com                 value, such as 100 or 1.0, e.g., a reference distribution of 11%
munication and media channels, predicted communication                   in a ?rst predictive attitudinal classi?cation may be normal
timing, predicted communication frequency, and predicted                 iZed to 100 and a reference distribution of 7% in a second
communication sequence.                                             30   predictive attitudinal classi?cation may also be normaliZed to
  There are a Wide variety of alternatives or defaults for               100. Also for example, for the selected population, and for a
non-matching entities of step 405, including variations                  given predictive attitudinal classi?cation, a penetration index
depending upon degrees or levels of matching. Exemplary                  of 150 or 1.5 may be utiliZed to indicate that the selected
alternatives include, for non-matching individuals or groups,            population has proportionally (or percentage-Wise) 50% (or
appending and utiliZing the average, most common or mode            35   1.5 times) more individuals (households, living units or other
classi?cations for a particular geographic region, such as a             groups) in that given predictive attitudinal classi?cation com
postal code area. Another alternative includes excluding                 pared to the larger reference population, such as a national or
those non-matching individuals or groups from the remainder              regional population. As illustrated above With the various
of the method and, equivalently, the selected population may             percentage distributions for the “Q” and “R” predictive atti
be considered to be comprised of the matching entities from         40   tudinal classi?cations, these comparisons are performed on a
step 405. Those of skill in the art Will recogniZe that the              proportional or percentage basis, rather than a comparison of
matching step 405 and the appending step 410 may be per                  pure or gross numbers, as the selected population generally
formed in a plurality of Ways, including use of conditional              concerns a considerably smaller total number of individuals
loops or iterations, With each iteration corresponding to the            (or groups) compared to the reference population represented
matching and appending for a given entity, and With iterations      45   in the database 100.
continuing until all entities have been matched (or found to                As a result of step 425, penetration indices or rates are
not match) and corresponding data appended.                              determined for each predictive attitudinal classi?cation of the
  The method then determines the distribution of the selected            plurality of predictive attitudinal classi?cations, comparing
population across or Within each of the predictive attitudinal           the proportion or distribution of the selected population in
classi?cations, to form a corresponding plurality of selected       50   that classi?cation to the proportion or distribution of the
population distributions, step 415. Each selected population             reference population in that classi?cation. The plurality of
distribution is compared to a reference distribution for each of         predictive attitudinal classi?cations are then evaluated by
the predictive attitudinal classi?cations, step 420. Typically, a        their penetration indices and, depending upon the selected
reference or baseline distribution is or may be the distribution,        embodiment, are also evaluated based upon the relative (or
across or Within each of the predictive attitudinal classi?ca       55   proportional) reference and selected population siZes Within
tions, of the larger, often national or regional population              each predictive attitudinal classi?cation, step 430. Using the
represented in the database 100, referred to above as the                penetration indices and relative or comparative reference and
reference population. For example, for a selected population,            selected population siZes of each predictive attitudinal clas
such as the purchasers of a particular automobile brand, When            si?cation, three additional levels of attitudinal classi?cations
compared to a larger regional or national population on a           60   are determined, namely, core attitudinal classi?cations, niche
proportional or percentage basis, that selected population               attitudinal classi?cations, and groWth attitudinal classi?ca
may be comparatively or relatively over-represented in cer               tions (steps 435, 440, 445). While core determinations are
tain predictive attitudinal classi?cations, and that selected            usually determined ?rst (to avoid potential confusion With
population may be comparatively or relatively under-repre                niche determinations, as based upon proportions of the
sented in other predictive attitudinal classi?cations.              65   selected population in addition to penetration indices), the
  Based on these comparisons of the distribution of the                  other determinations may be performed in any order. In other
selected population With a reference distribution, for each              variations, depending upon the selected evaluation algorithm,
                                                       US 8,346,593 B2
                              23                                                                       24
other determination orders for core, niche and growth attitu             determines one or more predictive message content classi?
dinal classi?cations may be available.                                   cations, based on the predictive message content classi?ca
  More speci?cally, in step 435, one or more core attitudinal            tions of the actual entities (individuals or groups) of the
classi?cations are determined by selecting, from the plurality           selected population assigned to that selected predictive atti
of predictive attitudinal classi?cations, at least one predictive        tudinal classi?cation. For each predictive attitudinal classi?
attitudinal classi?cation having a comparatively greater (e.g.,          cation, the plurality of predictive message content classi?ca
average or above) penetration index and having a compara                 tions may also be ranked, such as by comparative or relative
tively greater proportion of the selected population. These              penetration, proportion or distribution of a given predictive
core attitudinal classi?cations represent predictive attitudinal         message content classi?cation for that predictive attitudinal
classi?cations having the largest percentage of the selected             classi?cation, step 455.
population, such as customers, and corresponding, signi?cant                This independent determination of predictive message
market share. With respect to a selected population of cus               content classi?cations based upon the actual, selected popu
tomers of a particular brand, the core attitudinal classi?ca             lation (step 450 and optional ranking step 455) within each
tions represent signi?cant brand appeal to population seg                predictive attitudinal classi?cation, may be used to produce
ments exhibiting corresponding behavioral characteristics.               (or effectively results in) an information matrix or data struc
  In step 440, one or more niche attitudinal classi?cations are          ture, consisting of the plurality of predictive attitudinal clas
determined by selecting, from the plurality of predictive atti           si?cations (e. g., as rows) and the plurality of predictive mes
tudinal classi?cations, at least one predictive attitudinal clas         sage content classi?cations (e.g., as columns), both of which
si?cation having a comparatively greater (e. g., average or              may be further ranked or ordered according to relative distri
above) penetration index and having a comparatively lesser          20   bution, penetration and/or population siZe. As a result, not
proportion of the reference population. These niche attitudi             only may a selected population be predictively classi?ed or
nal classi?cations represent predictive attitudinal classi?ca            segmented attitudinally and behaviorally, using the plurality
tions having a high penetration rate (and corresponding mar              of predictive attitudinal classi?cations, they may also be inde
ket share), but a relatively small percentage of the reference           pendently and predictively classi?ed based on content or
population, such as a small percentage of a prospect popula         25   theme receptivity, using the plurality of predictive message
tion.                                                                    content classi?cations. Communication channel, media, tim
   In step 445, one or more growth attitudinal classi?cations            ing, frequency, and sequencing information may also be
are determined by selecting, from the plurality of predictive            included in such a matrix, e.g., as columns, and is discussed in
attitudinal classi?cations, at least one predictive attitudinal          greater detail below, as the various ?elds of a data structure of
classi?cation having a comparatively lesser (e.g., below aver       30   the present invention.
age) penetration index and having a comparatively greater                   In the exemplary embodiments, with the availability of
proportion of the reference population. These growth attitu              channel, media, timing, frequency, and sequencing informa
dinal classi?cations represent predictive attitudinal classi?            tion in the database 100, the method continues with step 460,
cations having some penetration success, and with the com                in which the predominant communication channel and/or
paratively large percentages of the reference population, such      35   media preferences are determined for each predictive attitu
as prospective customers, indicate signi?cant opportunities to           dinal classi?cation of the plurality of predictive attitudinal
increase penetration and add new customers from an other                 classi?cations, based upon the preferred communication
wise underrepresented group.                                             channels and/or preferred media types of the entities (indi
   To this point in the method of the present invention, con             viduals or groups) of the selected population assigned to the
siderable attitudinal and behavioral information has been pro       40   given predictive attitudinal classi?cation, such as email, inter
vided, which may be utiliZed for a wide variety of purposes.             net, direct mail, telecommunication, radio (broadcast, cable
Based on empirical modeling, actual attitudes and behaviors              and satellite), television (network (broadcast), cable or satel
of segments of a selected population may be predicted, using             lite), video (or DVD) media, print media, electronic media,
the plurality of predictive attitudinal classi?cations. Depend           visual or other public display media, and depending upon the
ing upon selected purposes of the embodiment, additional            45   selected embodiment, the plurality of communication and
information may be provided, such as the actual attitudes and            media channel classi?cations may be more or less speci?c,
behaviors of individuals or groups in the predictive attitudinal         such as further subdividing print and electronic media chan
classi?cations, including the core, niche and growth classi?             nels into newspaper, weekly magaZines, monthly magaZines,
cations.                                                                 joumals, business reports, and further into their print, inter
  Additional information is also independently provided in          50   net, email or electronic versions. For example, predominant
accordance with the present invention. While a selected popu             communication channels for a ?rst predictive attitudinal clas
lation has been predictively classi?ed as exhibiting certain             si?cation may be, in preferred order, direct mail followed by
attitudes and behaviors, as “who” segments (such as who                  radio followed by email, while predominant communication
among the population are signi?cant customers or prospects),             channels for a second predictive attitudinal classi?cation may
an additional, independent and more ?ne-grained level of            55   be, also in preferred order, television followed by telecom
information is also provided, based upon the plurality of                munication followed by direct mail.
predictive message content classi?cations, providing inde                  In step 465, the predominant timing (time of day) prefer
pendent “what” segments (such as what content will be most               ences for communications are determined for each predictive
effective). More speci?cally, the actual members of the                  attitudinal classi?cation of the plurality of predictive attitu
selected population, although assigned to a predictive attitu       60   dinal classi?cations, also based upon the communication tim
dinal classi?cation as a predominant classi?cation, may also             ing preferences of the entities (individuals or groups) of the
exhibit other or different attitudes and behaviors, represented          selected population assigned to the given predictive attitudi
by a probability or membership in one or more predictive                 nal classi?cation. For example, predominant timing prefer
message content classi?cations, in addition to those of the              ences for a ?rst predictive attitudinal classi?cation may be, in
predominant predictive attitudinal classi?cation. As conse          65   preferred order, weekends followed by evening, while pre
quence, in step 450, for each of the plurality of predictive             dominant timing preferences for a second predictive attitudi
attitudinal classi?cations, the method also independently                nal classi?cation may be, also in preferred order, mornings

				
DOCUMENT INFO
Shared By:
Categories:
Tags:
Stats:
views:3
posted:11/25/2013
language:Unknown
pages:28