Invented by Marc Christian Fanelli, Patricia Kay Gormley, Kymberly Ann Kulle, Thomas G. Nocerino, Kaushik Sanyal, Experian Information Solutions LLC
The Experian Information Solutions LLC invention works as follows
The present invention is a system, a method, a software, and a data structure that can independently predict attitudinal responses and message content, based on a plurality attitudinal classifications or other identification classifications as well as a multitude of message version or content classifications for a select population of multiple entities such households or individuals represented in a database. A plurality predictive attitudinal classifications (or identification classifications) and a plurality predictive message content classifications (or more version) have been determined by using a variety of predictive models, developed from a representative sample and applied to the reference population in the data repository. These models include attitudinal behavioral or demographic models. At least one predictive message or version classification for each predictive attitudinal classification (or identification classification) is determined independently. For each predictive attitude classification, the exemplary embodiments provide corresponding information regarding predominant communication media types (or channels), predominant communication timing, dominant communication frequency, or predominant communication sequence.Background for System, method, software, and data structure to predict attitudinal responses and communication preferences, including media, channels, timing, frequency and sequences, by using an integrated database
Various database applications are being developed to try to use the vast amount of information in these databases for analytical and marketing purposes. Demographic data can be added to customer records to determine the demographic composition of an entire group of customers. Then, marketing could be targeted at people with similar demographic characteristics.
These database applications in their many forms attempt to understand distinct groups of customers and prospects and then deliver the correct message to each individual, household, unit, or other audience. In most cases, the database will segment all the households and/or individuals into distinct groups based on their lifestyles, demographics, and consumer behaviors. In other applications, following such segmentation, consumer attitudes and motivations are assumed and attributed to those individuals/households within each such segment or cluster. The number of segments used varies greatly by application.
In addition, these database marketing applications assume and assign consumer attitudes, and the preferred themes and types of marketing messages to segments, without independent empirical research or analysis. Once a population has been segmented, further analysis of that population based upon preferred messaging themes will not add any independent information and only repeat the assumptions underlying the message theme assumptions for any segment.
The data resulting from the analysis may be ambiguous, inaccurate, or not actionable. The attitudes, motivations, and behaviors that are attributed to or assigned to each segment, for example, may not reflect the actual data and may not have been based on empirical, factual research. These attitudes, motivations and behavior may or may reflect representative attitudes in a customer database.
The low response rates of direct mail marketing, for example, are a further indication of the diminished accuracy. It is necessary to use other methods and systems to motivate and target the rest of the target audience and to identify potentially new, underdeveloped audiences. “New methods and systems will be required to maximize the marketing return by not oversaturating the targeted audience with excessive, ineffective communication, but instead, to communicate with them using their preferred methods and time of communication.
As such, there is a continuing need for a system and methodology that can accurately predict attitudes, motivations, and behaviors. This could be used for marketing purposes. This method and system must be empirically based. It should use actual demographic, behavioral, or attitudinal research, as well as other information, from a sample of a population. In addition, it should have accurate modeling that can predict or extrapolate this attitudinal information or other information for a larger population. This method and system would provide information about preferred message themes, or message content, independent of any grouping or segmentation process. A method and system of this type should also be actionable. It should provide not only audience attitudes and preferred message contents, but also information on preferred communication channels or other preferred media, as well as preferred frequency or other contact and timing information.
The present invention is a method, system and software that predicts independently a plurality first, content classifications for messages and a second, attitude classifications for a select population of individuals, families, living units, or other groupings in a data repository. For example, a select population of customers, prospects, or clients in a database, or data files. The invention’s system, method, and software can also, depending on which embodiment is selected, determine preferences regarding communication channels or other media, communication timing, communication frequency, and/or sequences between types of communication.
The illustrated, example embodiments of the invention are empirically based, using real attitudinal data and other information from an actual population sample. You can also use other data or research, including transactional data or demographic data. The invention uses this empirical base to provide accurate modeling for predicting and extrapolating such attitudinal or behavioral data, demographic information, or other types of information to a large reference population. This allows for accurate predictions of attitudes, motivations, and behaviors that can be used for marketing purposes, for example. The exemplary embodiments provide additional information regarding preferred message themes or content independent of any population grouping, clustering or segmentation process. The exemplary embodiments provide actionable results by providing audience attitudes and preferred message content as well as preferred communication channels, communication media and frequency, and timing and sequencing of communication information.
The power of this invention is not to be underestimated. Prior art methods focused on finding “who”, i.e., the individuals or households that marketers should target with their communications. These prior art methods do not provide, independent of selecting ‘who’, a determination of ‘what? Communication, including preferred content or versions. These prior art methods do not provide independent information about the “when” of the communication. These methods do not provide independent information on the?when? These prior art methods do not provide independent information about the “how” of communication. Communication methods, including the preferred communication medium, like direct mail, email, broadcast media, or print media. “Finally, none of these methods in the prior art provide independent information about the frequency (how frequently) and sequence (ordering), based on preferences. For example, print media is used a certain number of time, then direct mail, and finally email.
The present invention, more specifically, provides a system, software and method for independently predicting a plurality first predictive classifications referred as message content classifications and a multitude of second predictive classes referred as attitudinal classifications or other behavioral categories, for a select population of a number of individuals, households or living units, or other groupings or persons, as “entities”, represented in a database. Any reference herein to an ‘entity’ is meant in the broadest sense. As used herein, any reference to?entity? or?entities’ should be understood to mean and include any individual, household, living unit group or potential grouping of one or more people. “Entities” can be defined as any person, family, household, unit of living, group, or potential groupings of people, related or not, individually or collectively.
The exemplary embodiments illustrate empirical attitudinal studies and predictive attitudinal categories. However, they should be understood as including other types of research or classifications such as demographic or behavioral classifications derived from empirical research such as demographic or behavioral survey research.
The following is what the various exemplary system, method and software embodiments perform:
The various embodiments may also provide that, depending on the embodiment selected, for each entity of the plurality (e.g. individual or household), a predictive communication media (or another channel) classification from a plurality (of predictive communication media) classifications is appended from the data repositories. This can be followed by a predictive communication timing (or other channels) classification from a multitude (of predictive communication timing) classifications.
Typically, the plurality predictive communication media classifications includes at least two communication media (also referred to as channels of communication): electronic mail (e.g., email), internet, direct mailing, telecommunications, broadcast media such as radio, TV, cable, satellite, video media, optical (DVD,CD), print media like newspapers and magazines, electronic media including web sites, electronic versions of newspapers and magazines, and public display media like signage, billboards and multimedia displays. The plurality of media and channel classifications can be more or lesser specific depending on the embodiment selected. For example, print media channels may be further divided into weekly magazines, magazines monthly, journals, business report, and then further into their email, internet or electronic versions. The various broadcast media can also be classified in a variety of ways, including cable, satellite television, radio frequency transmissions, the internet, etc. The plurality of classifications for predictive communication time includes at least two categories of communication timing: weekdays, weekends, morning, afternoons, evenings, nights, and any other time. The majority of predictive communication frequency classes typically includes at least two communication frequency categories: daily, weekly biweekly monthly semi-monthly bimonthly annually semi-annually and none. The communication sequences can be very diverse and include print communications followed by electronic ones.
In the various embodiments the plurality or classifications of the predictive message content are or have been determined:
The invention provides for the following classifications:
The method comprises: (a) for each entity of the plurality of entities of the selected population, appending from the data repository a corresponding predictive identification classification of a plurality a predictive identification classifications wherein the plurality of predictions identify a number of entities according to a specified property; and (b) for each entity of the pluraly of individuals of the selected population in a corresponding predictive message version classification of a multitude of predicted message versions. The method includes: (a), for each of the entities in the plurality, adding a corresponding prediction identification class of a number of classifications where the group of classifications are based on a select property, (b), for each of those entities in the population, adding at least one predictive message versions classifications from a number of classifications. The selected property can be derived from one or more of the following attributes: behavioral characteristics; demographic characteristics; geographic characteristics; financial characteristics and transactional characteristics.
In a further aspect of the invention the exemplary embodiments present a datastructure for independently predicting the communication responsiveness of selected populations of a plurality entities represented in a database. This data structure can be stored electronically or in a physical medium. The data structure comprises a first section with a number of predictive classifications that designate multiple entities according to selected properties; and a secondary field with at least one predominant predictive version classification for each of these predictive classifications. These predictive version classifications are determined by a variety of predictive models derived from a sample and applied to the reference population in the data repository.
The data structure can also include: a third field with, for every predictive identification class of the 1st field, a predominant predictive communications media classification; a 4th field with, for every predictive identification class of the 1st field, a predominant predictive communications timing classification; a 5th field with, for a each of the predictive identification classes of the 1st field, a predominant predictive communicating frequency classification; a 6th field with, for a each of the predictive identification classes of the1st field, a The selected property can be derived from any of the following characteristics: Attitudinal Characteristics, Behavioral Characteristics, Demographic Characteristics, Geographical Characteristics, Financial characteristics or Transactional Qualities.
The exemplary embodiments of the invention provide a method of independently predicting the responsiveness to communication media of a select population of a number of entities in a data store, which comprises: (a), for each entity from the group of entities, adding a corresponding prediction identification classification, where the plurality designates a number of entities according a certain property, and (b), for each of these predictive identification classes, independently determining the predominant predictive communication medium classification among a variety of predictive communication classifications.
Click here to view the patent on Google Patents.