Customized digital content generation systems and methods

Abstract

The invention provides in some aspects a method, executed on a digital data processing system, of mass generation of customized digital content that includes continuously identifying current external events taken by or with respect to a plurality of respective prospective targets and, upon identification of such an event, generating a set of actions, each identifying a digital content piece and a digital delivery mechanism therefor. Each action is generated, according to the method, based on the current identified events for a particular prospective target and on a database of information about prior events taken by or with respect to him/her. The sets of actions are queued upon generation and continuously retrieved on a first-in-first-out basis. And, upon retrieval, an action for generation of digital content for the respective prospective target is selected for transmittal from the set based on quotas associated with that target and/or the delivery mechanism identified for it per the selected action.

Description

BACKGROUND OF THE INVENTION

The invention pertains to digital data processing and, more particularly, to customized digital content generation. The invention has application, by way of example, in generating digital content based an ever-changing context of user interactions with an enterprise or other publisher of that content.

In order to generate and distribute digital content that will have impact on a target audience of users, a large enterprise generates millions of digital content pieces per campaign and distributes them at a rate of tens of thousand per minute or more. Achieving those rates and quantities requires a digital data processor. And, though those machines are fast, the rate and quantity at which the digital content pieces in the campaign must be generated destines a sizable portion of them to staleness.

By way of example, an email campaign targeting one million potential readers is typically launched and its substantive focus defined at least one to three days before the first digital content piece is generated and transmitted. By the time the one millionth piece has been sent, another day or two may have passed. Given the number of recipients and the length of time from launch to completion, its likely that five to ten percent of the content pieces that might otherwise have been relevant to their intended recipients have lost that by the time of receipt.

More concretely, such a campaign launched by an enterprise on a Monday and completed on a Thursday promoting air conditioners to potential customers in a metropolitan region in advance of a weekend heatwave would normally be expected to have a measurable impact—e.g., a click-through or response rate—on twenty to thirty thousand of those customers. During that four-day period, one thousand of those would-be responders are likely to have already purchased an air conditioner from the enterprise. For them, any email received as part of the campaign is wasted. In fact, it could be worse, if it offers better terms than the customers had gotten for their purchases.

And, of course, it is not just prior purchasers to whom the email campaign may be for naught: a myriad of other circumstances may change during the course of the campaign that renders it wasted as to another sizable segment of the target audience. The complexity is further magnified by the multitude of channels through which a modern-day campaign is effected, from social media (in all its forms) to email to snail mail, to name a few.

To be clear, it is not the fact of the email campaign that is, itself, problematic. It is the static nature of the content, which is typically determined days in advance of its actual final assembly, generation, and distribution. It is also the content delivery mechanism, which does not take into account the many alternate channels through which the modern-day enterprise reaches its audience.

The prior art has proposed various solutions to content generation, including Herman, U.S. Pat. No. 10,453,083, and Hyman, U.S. Pat. No. 10,601,751, neither of which hits the mark. These and other prior art solutions fail to suggest mechanisms for generating digital content that is both customized to the individual recipients to whom it is sent and that takes into account the ever-changing and last-minute circumstances of those recipients, particularly, vis-à-vis their interactions with the enterprise that publishes the content.

The problem is one very particular to computers: they are capable of—and, indeed necessary for—generating and distributing large quantities of digital content quickly; however, they lack the ability to adapt it to the ever-changing circumstances of the large numbers of recipients to whom it is distributed.

In view of the foregoing, an object of the invention is to provide improved digital data processing systems and methods.

A more particular object is to provide improved such systems and methods as are suited for customized digital content generation and distribution.

A still further object is to provide such improved systems and methods for digital content generation and distribution as take into account an ever-changing context of user interactions with the publisher of that content.

SUMMARY OF THE INVENTION

The foregoing are among the objects attained by the invention, which provides in some aspects a method, executed on a digital data processing system, of mass generation of customized digital content that includes continuously identifying current external events taken by or with respect to a plurality of respective prospective targets and, upon identification of such an event, generating a set of actions, each identifying a digital content piece and a digital delivery mechanism therefor. Each action is generated, according to the method, based on the current identified events for a particular prospective target and on a database of information about prior events taken by or with respect to him/her.

The sets of actions are queued upon generation and continuously retrieved on a first-in-first-out basis. And, upon retrieval from the queue, an action for generation of digital content customized for the respective prospective target is selected from the set based on quotas associated with that target and/or the delivery mechanism identified for it per the selected action. The digital data processing method further includes invoking, continuously and on selection of such an action, a content server to transmit the designated digital content piece to the respective prospective target via the delivery mechanism identified by that action.

Related aspects of the invention provide a digital data processing method, e.g., as described above, that includes ingesting incoming data streams, e.g., from any of the web, social media, call centers, transaction servers, and data warehouses, in order to identify the respective events. Those events, which occur asynchronously with respect to one another, can according to aspects of the invention, be identified substantially in real time with their respective occurrences. Other related aspects of the invention include memorializing the identified events in the aforesaid database.

Yet still other related aspects of the invention provide a digital data processing method, e.g., as described above, that includes generating identifiers of prospective targets on a round-robin basis from a list of known targets. This can be done continuously and concurrently, yet asynchronously, with respect to identifying current external events that may occur with respect to those perspective targets—though, according to some aspects of the invention, identifiers of targets for whom such events are detected are winnowed from among those generated from the list. Yet, like identifiers for prospective targets for whom external events are detected, the method includes generating sets of actions for prospective targets from identifiers generated round robin-style from the list.

Still further aspects of the invention provide digital data processing methods, e.g., as described above, comprising invoking one or more content servers based on the sets of actions to transmit 100,000 or more, 1,000,000 or more, or 10,000,000 or more, digital content pieces to different respective targets per day.

Yet other aspects of the invention provide digital data processing systems that generate custom digital content in accord with the methods above.

These and other aspects of the invention are evident in the drawings and in the text below.

BRIEF DESCRIPTION OF THE DRAWINGS

A more complete understanding of the invention may be attained by reference to the drawings, in which:

FIG. 1 depicts a digital data processing system of the type in which the invention operates;

FIG. 2 depicts a procedure executing within the system of FIG. 1 for generation of custom digital content.

DETAILED DESCRIPTION OF THE ILLUSTRATED EMBODIMENT

Architecture

FIG. 1 depicts a digital data processing system 10 according to the invention, as well as a system 10 in which the invention may be practiced. The system 10 includes a digital data device 12 for managing the generation of customized digital content (hereafter, alternatively “digital data processor 12”, “server 12”, “custom content server 12”, or the like) that is coupled to one or more further server digital data devices 13, 15, 17, and to a plurality of client digital data processors, here, represented by elements 14, 16 and 18, via a network 20 comprising a combination of one or more of the Internet, a wide area network (WAN), virtual private network (VPN), metropolitan area network (MAN), local area network (LAN), telephone networks and/or a combination of these and other networks (wired, wireless, public, private or otherwise).

The illustrated client digital data devices 14, 16, 18 are conventional tablet computers, PDAs, mobile phones or other digital data apparatus (including, for example, desktop computers, workstations, minicomputers, and laptop computers) of the type commercially available in the marketplace and suitable for operation as client devices for the consumption of digital content by their respective users (not shown) in the conventional manner known in the art as adapted in accord with the teachings hereof. The devices 14, 16, 18 each comprise central processing, memory, storage and input/output units and other constituent components (not shown) of the type conventional in the art that cooperate to execute client applications 14a-18a, such as web browsers, client apps, and so forth, all of the type conventional in the art as adapted in accord with the teachings hereof, for purposes of receiving and responding to digital content pieces of the type transmitted from web servers, email servers, social media servers, e-commerce servers or otherwise.

Although client digital data processors 14, 1618 are depicted and described in like manner here, it will be appreciated that this is for sake of generality and convenience: in other embodiments, these devices may differ in architecture and operation from that shown and described here and/or from each other, all consistent with the teachings hereof. And, although only three client devices 14, 16, 18 are shown in the drawing, in practice of the invention, there may be tens, hundreds, thousands, or millions of such devices, operated or operable by their respective users.

Like client digital data processors 14, 16, 18, server digital data processors 13, 15, 17 are digital data processing apparatus of the known in the art operating as web servers, email servers, social media servers, e-commerce servers or other digital content servers, all as adapted in accord with the teachings hereof. Though the servers 13, 15, 17 are typically implemented in a server-class computers, such as minicomputers, they may also be implemented in desktop computers, workstations, laptop computers, tablet computers, PDA or other suitable apparatus (again, as adapted in accord with the teachings hereof).

As above, although server digital data devices 13, 15, 17 are depicted and described in like manner here, it will be appreciated that this is for sake of generality and convenience: in other embodiments, these devices may differ in architecture and operation from that shown and described here and/or from each other, all consistent with the teachings hereof. And, although only three server devices 13, 15, 17 are shown in the drawing, in practice of the invention, there may a greater or lesser number of such devices, operated or operable by their respective administrators.

Custom content server 12 is a digital data processing apparatus of the type commercially available in the marketplace suitable for operation in accord with the teachings hereof to create and drive custom content to client devices 14, 16, 18 via servers 13, 15, 17. Though the server 12 is typically implemented in a server-class computer, such as a minicomputer, it may also be implemented in a desktop computer, workstation, laptop computer, tablet computer, PDA or other suitable apparatus (again, as adapted in accord with the teachings hereof).

Like the other digital data processors 13-18 shown in the drawing, custom content server 12 of the illustrated embodiment comprises central processing, memory, storage and input/output units and other constituent components (not shown) of the type conventional in the art, here, in the case of server 12, configured in accord with the teachings hereof to form rules engine 22 and rules base 24, transaction database 26, language database 28 and context registry 30, one or more of which may be absent in various embodiments of the invention.

Although only a single server digital data processor 12 is depicted and described here, it will be appreciated that other embodiments may have greater or fewer numbers of these devices. Those other servers may differ in architecture and operation from that shown and described here and/or from each other, all consistent with the teachings hereof. Still further, although server 12 of the illustrated embodiment is depicted as being remotely disposed from the client digital data devices 14, 16, 18, and server digital data devices, 13, 15, 17, in other embodiments, server 12 may be co-housed and/or co-extensive with one or more of those other devices.

Rules base 24 comprises a conventional rules base of the type known in the art (albeit configured in accord with the teachings hereof) for storing digitally encoded rules 25 and other application-related information in tables, database records, database objects, and so forth. Such stored rules 25 are likewise formatted and stored in the conventional manner known in the art (albeit configured in accord with the teachings hereof).

Here, rules base 24 is configured and contains rules for use in business process management applications, though in other embodiments it may be configured and used for other applications. Such a rules base can be of the type described in the US patents and patent applications assigned to the assignee hereof, e.g., U.S. Pat. Nos. 5,826,250, 6,976,144, 7,665,063, 8,335,704, 7,711,919, 7,640,222, 8,250,525, US 20080208785, US 20090132232, U.S. Pat. Nos. 8,843,435, 8,479,157, 8,468,492, 8,880,487, 8,924,335, 9,189,361, US 20130231970, U.S. Pat. Nos. 9,678,719, 8,959,480, 9,270,743, US 20150127736, U.S. Pat. No. 9,658,735, US 20160098298, US 20170090685, US 20170255341, all by way of non-limiting example, the teachings of which are incorporated by reference herein; though, a rules base that is architected and/or operated differently may be used as well or in addition.

Although shown in the illustrated embodiment as part of server 12, rules base 24 (or other repository) may reside elsewhere, e.g., remote to server 12. Moreover, some embodiments may utilize multiple rules bases, e.g., an enterprise-wide rules base 24 on the server 12 and domain-specific rules bases on one or more of client devices 14, 16, 18, all by way of example. To the extent that multiple rules bases are provided in any given embodiment, they may be of like architecture and operation as one another; though, they may be disparate in these regards, as well. Utilization of multiple rules bases may be accomplished in the manner described in one or more of the above-cited incorporated-by-reference U.S. patents and patent applications, e.g., U.S. Pat. No. 8,924,335, entitled “Systems and Methods for Distributed Rules Processing,” the teachings of which are incorporated heron by reference.

Transactional database 26 comprises a conventional database of the type known in the art (albeit configured in accord with the teachings hereof) for storing corporate, personal, governmental or other data that may be any of generated, stored, retrieved and otherwise processed (hereinafter, collectively referred to as “processed”) by rules in rules base 24 and/or rules stored/executed elsewhere. The data may be financial data, customer records, personal data, run-time data related to an application, or other type of data and it may be stored in tables, database records, database objects, and so forth, all by way of non-limiting example.

As above, some embodiments may utilize multiple transactional databases, e.g., an enterprise-wide database 26 on the server 12 and branch-office specific databases on the client devices 14, 16, 18, all by way of example. To the extent that multiple transactional databases are provided in any given embodiment, they may be of like architecture and operation as one another; though, may they be disparate in these regards, as well. Utilization of multiple transactional databases may be accomplished in the manner described in incorporated-by-reference U.S. Pat. No. 8,924,335.

Language base 28 encodes information regarding the syntax of the language (or languages) in which user interfaces generated by server 12 are presented on devices 14-18 and, more particularly, in the illustrated embodiment, by applications 14a-18a. That syntax can include one or more of grammar, spelling, usage, punctuation, and/or style. The language base 28 may comprise a language database of the variety commercially available in the marketplace—e.g., in the manner of spelling and grammar dictionaries provided with conventional word processors (which “dictionaries” often include usage-, punctuation- and/or style-related entries). Alternatively, the language syntax information may be embodied in one or more rules 25 of the rules base 24, or otherwise.

Context registry 30 is a store that includes information related to the respective contexts or circumstances in which the requested user interfaces (e.g., web pages) are and/or will be communicated to and executed on the respective client devices 14-18. That context or circumstance can include, by way of non-limiting example, user “properties” or business attributes (e.g., access requirements, disability settings, market segment, behavioral segment, age, locale, and so forth), client device 14-18 properties (e.g., processor speed, display size, keyboard capabilities, locale, and so forth), communication channel properties (e.g., the speed and type of connection between server 12 and the respective client devices 14-18) and portal properties (e.g., download file size limits, quality-of-service requirements and so forth). That context or circumstance can further include, by way of non-limiting example, the language, country and/or other locale settings and preferences of the user of a device on which the web page is to be displayed. Still other variations in the possible range of values stored in the context registry 30 are possible.

Illustrated digital data processor 12 also includes rules engine 22 of the type conventionally known in the art (albeit configured in accord with the teachings hereof) for use in processing rules from a rules base in order to process data, e.g., in (and/or for storage to) a transactional database in view of a language base and/or context registry or other data sources described above, for example, in connection with events signaled to and/or detected by the engine. In the illustrated embodiment, the rules engine is of the type used for business process management applications, though in other embodiments it may be of the type used for other applications. Preferred such rules engines are of the type described in the aforementioned incorporated-by-reference U.S. Pat. No. 5,826,250, entitled “Rules Bases and Methods of Access Thereof” and U.S. Pat. No. 7,640,222, entitled “Rules Base Systems and Methods with Circumstance Translation” and/or U.S. Pat. No. 8,250,525, entitled “Proactive Performance Management For Multi-User Enterprise Software Systems,” by way of example, the teachings too of which are incorporated by reference herein-all as adapted in accord with the teachings hereof.

The rules engine 22 may be implemented in a single software program or module, or a combination of multiple software modules/programs. Moreover, it may comprise programming instructions, scripts, rules (e.g., rules stored in rules base 24) and/or a combination thereof. And, even though the rules engine 22 of the illustrated embodiment is depicted as executing on just server digital data processor 12, in other embodiments, the engine 22 may execute on or across multiple digital data processors (e.g., 12, 14, 16 and 18). Executing the engine 22 over multiple digital data processors may be accomplished in the manner described in incorporated-by-reference U.S. Pat. No. 8,924,335, by way of non-limiting example.

Custom Content Generation

Described below is a method of operating system 10 and, particularly, server 12 and still more particularly, by way of example, rules engine 22, to execute a procedure 32 to generate digital content pieces customized for delivery to clients devices 14, 16, 18 via servers 13, 15, 17. This may be, for example, at the request of an administrator of device 12 issued via an browser interface (not shown) or a dedicated application (not shown) executing on that device or otherwise; conversely, such invocation of the method may be effected sans operator request, e.g., upon boot-up of server 12 or otherwise, all as per convention in the art as adapted in accord with the teachings hereof.

FIG. 2 depicts a procedure 32 for generation of custom digital content according to one practice of the invention. The procedure is depicted by way of code blocks, each associated with one or more steps of the procedure, as described below. In practice, each of those blocks—which are referred to as “steps” in the discussion that follows—may be implemented by informal or formal groupings of rules and/or code, whether organized in objects, functions, subroutines or other programming constructs, that execute within rules engine 22 or other software execution environment, all as is within the ken of those skilled in the art in view of the teachings hereof. It will be appreciated that the association between steps of the procedure 32 and blocks of FIG. 2 is for purposes of illustration and that, in practice, the specific association between those steps and the objects, functions, subroutines or other constructs that implement them may vary from that shown here and described below, all without deviating from the scope of the invention. Moreover, though, in the illustrated embodiment, the procedure 32 executes within the engine 22 or other environment on server 12, in other embodiments, one or more steps of the procedure 32 may be executed on other devices, e.g., servers 13, 15, 17, again, as is within the ken of those skilled in the art in view of the teachings hereof.

Event Ingestion

Referring to the drawing, in step 34 the illustrated procedure 32 continuously ingests data from external data stores, streams, and otherwise, in order to identify current external events taken by or with respect to prospective targets of customized digital content pieces.

In the illustrated embodiment, the targets are individuals that receive digital content by way of client devices 14, 16, 18 and, more specifically, by way of apps 14a, 16a, 18a executing respectively thereon. Other embodiment may vary in these regards. Thus, for example, the targets may be businesses or other entities capable of receiving such content by way of such devices. In still other embodiments, the targets may be digital billboards, interactive voice response systems (IVRs), private or public broadcast systems, and/or other “things” capable of receiving digital content, e.g., for display or other presentation.

Digital content pieces may be emails, text messages, tweets, web pages, data files, digital robocalls or ringless voicemail, video or audio messages, or other digital content amenable to customization suitable for delivery to the devices 14, 16, 18 via servers 12, 13, 15, 17 in accord with the teachings hereof. Customization, as used here, refers to customization in terms of content (e.g., modification of words, images or other portions of a content piece), selection (e.g., as between multiple pre-existing content pieces), and/or delivery mechanism (e.g., email vs text message, web page vs data file, etc., all by way of nonlimiting example).

External events, as used here, refer to actions taken by the prospective targets or with respect to those targets. Events taken by and with respect to targets can include, by way of example, emails, tweets or social network postings by them—or ones directed to them. They can also include purchases (e.g., of products, properties or otherwise), dispositions (by way of sales, legal proceedings, or otherwise), life events (births, deaths, weddings, and so forth), all by way of example.

External events ingested in step 34 of the illustrated embodiment include only those that occur outside the context of the procedure 32. Thus, for example, generation and transmission to a target of a piece of customized digital content by the procedure 32 does not constitute an event that is ingested in step 34 and does not, in and of itself, trigger generation of a next piece of content (although, when an external event does subsequently occur with respect to that target, the fact of transmission of the prior piece of content may inform customization of the next piece of content generated and transmitted to him/her/it).

In step 34 of the illustrated embodiment, the procedure 32 identifies events concerning prospective targets from external data stores, streams and otherwise. The data stores can be public or private databases or other data repositories, including, for example, data warehouses. These can include transactional databases (e.g., reflecting service calls, communications or other interactions with a call center, purchases, and other events), databases of public records and so forth, all by way of non-limiting example. Streams can be social media feeds, web feeds, mobile app feeds, email feeds or other sources of information and other postings by or about (e.g., with respect to) the potential targets. In the discussion that follows, the sources of information regarding external events, are referred to for simplicity as “data stores and streams,” regardless of actual source. Accessing such data stores and streams is within the ken of those skilled in the art in view of the teachings hereof. Ingestion of those stores and streams to identify events can be by way of natural language processing, parsing, scraping and other knowledge discovery techniques, each selected in accord with the particular data store and/or stream from which events are to be identified—all as is within the ken of those skilled in the art in view of the teachings hereof.

In some embodiments, the target of every external event identified in step 34 is deemed a prospective target for purposes hereof and the event is processed accordingly. In the illustrated embodiment, however, only those events that concern previously known targets are processed—or, put another way, ingestion step 34 of the illustrated embodiment includes winnowing from the external data stores, streams and otherwise, events by or with respect to persons, business and/or other things not already included in a potential target database (not shown) or otherwise. By way of non-limiting example, the potential targets database of a business enterprise can be a database of customers and potential customers. Implementation of step 34, whether processing events to identify prospective targets or processing only those concerning already known such targets, is within the ken of those skilled in the art in view of the teachings hereof.

Step 34 of the illustrated embodiment is executed continuously—that is, not in a batch mode. Put another way, it processes, parses, scrapes and otherwise discovers information contained in data stores and streams with such frequency as to identify events in those stores and streams—events which, for the most part, occur asynchronously with respect to one another—substantially in real time with their respective occurrences. Of course, for events that do not occur, in the first instance, in such store or stream, this is real time with respect to the occurrence of their memorializations in such stores or streams. This enables the downstream steps of procedure 32 to customize digital content for the prospective targets to whom those events in real time and, thereby, ensure that content is relevant to those targets.

As shown in the drawing, processing by step 34 of the illustrated embodiment produces a list of IDs of prospective targets, as well as a data records reflecting the events discerned with respect to them during ingestion step 34. The former are referred to as “trigger” IDs in the drawing. They can be generated, on the fly, e.g., by hashing or otherwise—though, in the illustrated embodiment, the IDs of those potential targets are assigned based on those for the same respective persons, businesses or other targets, in a pre-existing known customer database or other data store of known prospective targets (not shown). Data records generated in step 34 can encode the identified external events (along with the respective prospective target ID) in any conventional manner known in the art as adapted in accord with the teachings hereof. IDs generated by step 34 are processed in the next-best-action step 36 discussed below, while the data records generated in step 34 reflecting current external events are passed to a pre-fetch data store 38 and memorialized there.

Trickling

In step 40, the illustrated procedure 32 generates IDs of prospective targets on a round-robin basis from among a list of identifiers of known respective prospective targets—e.g., from the aforesaid customer database or other data store of known prospective targets. This is done concurrently with step 34, though, it need not be synchronous with it.

As with step 34, step 40 is executed continuously (i.e., not in batch mode), albeit at a rate (e.g., in terms of IDs per hour, per day or so forth) paced to cycle through the IDs of all such known respective targets once per desired contact interval. The latter is the desired average period of time over which the operator of procedure 32 considers communicating to or to receive communication from the average prospective target. Thus, for example, if the operator considers communicating to/from the average target once per day, the step 40 is executed at a rate paced to cycle through IDs of all respective targets once per day. Of course, the desired contact interval will vary from embodiment to embodiment, as is within the ken of those skilled in the art in view of the teachings hereof.

This round-robin “trickling” of IDs, as this operation of step 40 is referred to in the drawing and elsewhere herein, is within the ken of those skilled in the art in view of the teachings hereof. In the illustrated embodiment, step 40 includes winnowing from the list of generated IDs those of respective prospective targets for whom a current external event has been identified in step 34 within the same desired contact interval.

Just as with IDs generated in step 34, those generated in step 40 are passed to next-best-action step 36 for processing by it.

Event History

The generation of custom digital content by illustrated procedure 32 is a function of the history of events with respect to each target. This can include current external events identified for those targets in step 34 and maintained in pre-fetch store 38. It can also include a prior history of external events for those targets maintained in database 42, labelled, “interaction history,” whether rolled over from the pre-fetch store 38 into database 42 or gathered in that database in the first instance from transactional database 26, data warehouses, or streams or other sources, e.g., using natural language processing, parsing, scraping and other knowledge discovery techniques, e.g., as discussed above in connection with step 34, or otherwise. In the illustrated embodiment, that history also includes “internal” events—to wit, data records regarding custom digital data content transmitted to the targets, e.g., in step 48.

Events can be stored in database 42 in a conventional manner known in the art, as adapted in accord with the teachings hereof. This can be by array, linked list or other data structures a maintained on server 12 or otherwise; although, in the illustrated embodiment, database 42 comprises a relational database under management (e.g., for purposes of data retrieval) of a relational database management system that forms part of the server 12 middleware (not shown). Other embodiments may vary in these regards, e.g., storing the event histories in online databases, non-relational, or otherwise, all as is within the ken of those skilled in the art in view of the teachings hereof.

Data store 38 serves as a cache for database 42 and, through the action of rules 25 or other programming instructions that work with store 38, prefetching from that database 42 and holding for more rapid access during next-best-action step 36 event data for respective targets that are about to be processed during that step 36. This can include current external event data identified for those targets in step 34 and both prior internal and external event data for those targets retrieved from the database 42 upon generation of target IDs in steps 34 and 40. In some embodiments, those target IDs are passed directly to those rules/instructions in connection with execution of steps 34, 40; in other embodiments, they are passed in connection with execution of step 36. Specifics of which such prior internal/external event data to retrieve is implementation specific and, more particularly, dependent on factors evaluated in next-best-action step 36 in determining and generating sets of actions for those targets, as discussed below. Implementation of the data store 38 and rules/instructions for pre-fetching event data for caching therein—including the use of arrays, linked lists or other data structures to store the prefetched event data—is within the ken of those skilled in the art in view of the teachings hereof.

Next Best Action Generation

In step 36, the illustrated procedure 32 generates sets of one or more actions for each respective prospective target whose ID is generated in steps 34 and 40. The step 36, which operates continuously (i.e., not in batch mode), generates those respective sets upon receipt of output from steps 34 and/or 40. In addition, in some embodiments, step 36 can generate such sets (thereby forcing update of the cache 46, as discussed below) whenever there is a change, e.g., in transactional database 36 regarding in demographic or other characteristics of a prospective target, when a target-specific watchdog timer fires (e.g., indicating that it is a target's birthday or that no activity has occurred vis-à-vis the target in a designated period, e.g., 30 days, or otherwise). Moreover, the step 36 of some embodiments can additionally be invoked to generate such a set whenever a special piece of digital content—e.g., an emergency message—must be generated for and sent to a target.

Each action in those sets identifies a digital content piece to be transmitted to the respective target and a delivery mechanism by which to transmit it. The illustrated embodiment generates three actions for each set (or communication plan—referred to as “comm plan” in the drawing), though, other embodiments may vary in this regard. Moreover, because each action is dependent upon the current and prior history of events of the respective prospective target, the actions in the plan for one such target are likely to differ from those for another.

Thus, for example, in an implementation that sets an upper limit of two on the number of actions in each communication plan, those for a first prospective target for whom an event was recently identified in step 34 might include a first action identifying a digital banner ad for display upon the target's next visit to a designate URL and a second action identifying an SMS to send to the target's phone number; whereas actions for a second prospective target who has no history of web visits and for whom no SMS phone number is known might include first and second actions each of which identify a different respective digital advertisement to send to that target's email address.

Step 36 of the illustrated embodiment executes a next-best-action algorithm of the type known in the art to generate each action in a communication plan for a given prospective target, though, other embodiments may use other algorithms or methodologies for plan determination. Depending upon the implementation of that algorithm in a given instance, determination of what actions—i.e., what digital content pieces and respective delivery mechanisms—are generated by the algorithm is a function of current external event data identified for that target, if any, in step 34 and both prior internal and external event data for that target retrieved from database 42 into prefetch data store 38, as discussed above. It can also be dependent on demographic and other characteristics of the target, e.g., as prefetched to store 38 from interaction history database 42, transactional database 26 or otherwise, all as is within the ken of those skilled in the art in view of the teachings hereof.

It will thus be appreciated that, because each target has a different current and prior event history, the algorithm generates different action recommendations—and, therefore, communication plans—for each target. Moreover, depending on the nature of the next-best-action algorithm utilized in a given embodiment, those recommendations can pertain solely to selection of pre-existing content pieces (e.g., from a database 44 of digital content pieces and other potential actions) and/or modification of such content pieces, e.g., with varied pricing, images, discounts, and so forth. They can also pertain to selection of alternate delivery mechanisms, e.g., as where the prior event history of a given prospective target augurs in favor of, say, email over text messaging, by way of non-limiting example.

As noted above, in the illustrated embodiment, multiple actions are generated for each communication plan. To effect this, the method of step 36 utilizes the next-best-action (or other) algorithm to generate multiple recommendations and, based on probability estimates, or other estimated success values assigned to those recommendations by the algorithm or otherwise, stores them in priority order in linked lists, arrays or other data structures employed by the method to encode those plans for storage in cache 46. In some embodiments, a communication plan is a list of records containing the following fields:

• • TargetID
  • Priority—importance of the message compared to other messages
  • Value—value of the message
  • ActionID
  • TreatmentID
  • Destination (Email, Branch, Call-Center, . . . )
  • TimeToCommunicate
  • Personalized Message

Implementation of the next-best-action step 36, as described above, including encoding the actions generated by it into communication plans, is within the ken of those skilled in the art in view of the teachings hereof.

Next Best Action Plan Queuing

Each set of actions, or communication plans, generated in step 36 and including the ID of the respective target (as well, of course, of the action(s) that make up the plan) is queued to next-best-action cache 46 at the conclusion of step 36. Queuing can be performed as part of step 36 and/or by through the action of rules 25 or other programming instructions associated with cache 46. Regardless, the queuing is done continuously (i.e., not as part of a batch mode process) as each communication plan is generated in step 36. Moreover, the queuing of the illustrated embodiment is on a FIFO (first in, first out) basis, thereby, better insuring that the communication plans generated for prospective targets will be carried out, e.g., in step 48, closer in real time vis-à-vis the events that triggered creation of those plans in the first instance per steps 34-36.

Notwithstanding FIFO queuing, the cache maintains no more than one plan per prospective target—even if multiple events ingested for a given target in step 34 and/or if action of steps 34 and 40 result in generation of multiple plans for that target in step 36. To this end, when a plan is queued to the cache 46 (whether by rules/code implementing step 36 or otherwise) any prior plans currently in that cache for the same target are removed. This can be done through a brute-force search of the queue with the target ID in a newly generated plan, though, preferably, a hash of such IDs or other index-based search within the ken of those skilled in the art is used to speed the search and replacement.

Implementation of the FIFO cache 46 in a manner that supports the searching for and replacement plans for the same target, whether via arrays, linked lists or other structures, is within the ken of those skilled in the art in view of the teachings hereof.

Executing Action Plans (Alternative 1—Pushing Plans to Servers for Execution)

In step 48 (labelled “governor” in the drawing), the illustrated procedure 32 retrieves and acts on action plans in the cache 46. This is done continuously (i.e., not on a batch mode basis) and on a FIFO basis as discussed above.

To that end, in step 48 the illustrated procedure 32 retrieves a next communication plan from the cache 46 and unpacks the one or more actions contained within it. In the illustrated embodiment, these are typically provided in an ordered priority determined in the next-best-action step 36, as discussed above.

The method of step 48 selects the most highly recommended of those actions and determines whether it can be carried out—and, specifically, whether the digital content piece identified in that action is available (if not, itself, encoded in the plan), whether a server necessary to deliver that content piece via the delivery mechanism identified in the plan is available, and whether delivery of that content piece by that delivery plan (and server) to the target identified by the action at the time of execution of step 48 on that action would violate quotas (e.g., contact policies, action constraints and/or treatment constraints) associated with that target and/or that delivery plan. A pseudo-code example showing how such a quota violation determination can be made is provided below in connection with the alternate embodiment.

Those quotas (or policies), which may be hard-coded into the rules/code that implements step 48 but which, preferably, are maintained in a database or other data set accessible to those rules/code, are of the type known in the art establishing numerical, rate, time/date or other limits on content transmissions. By way of non-limiting example, such a policy can define numerical daily or weekly limits on the number of text messages to non-corporate targets; another could restrict usage of a given delivery mechanism on a per-period or time-of-day basis; and so forth, all as within the ken of those skilled in the art in view of the teachings hereof. In the illustrated embodiment, the quotas (or policies) are be shared with the next-best-action step 36, as indicated in the drawing.

Instead of or in addition to those associated with targets and/or delivery plans, other embodiments may utilize quotas associated with digital content pieces (e.g., limiting the number of times a given content piece can be transmitted) or otherwise, again, as is within the ken of those skilled in the art in view of the teachings hereof. On the other hand, in some embodiments, quotas are not invoked for certain actions or action plans—to wit, those including special pieces of digital content—e.g., an emergency message.

If the method of step 48 determines that the selected action in the communication plan can be carried out and would not violate quotas, it “pushes” that plan for execution. More specifically, the method invokes a server 13, 15, 17 appropriate for the delivery mechanism identified in the selected action, passing to it the digital content piece identified in that action for digital transmission to the target's client device 14, 16, 18. This, too, is done continuously (i.e., not on a batch basis) to better insure delivery of the content piece while it remains relevant. At the same time, step 48 memorializes transmittal of that content piece by that mechanism to database 42. In the illustrated embodiment, step 48 also memorializes the event to the pre-fetch store 38.

On the other hand, if the selected action from the communication cannot be carried out, e.g., because of lack of availability of the content, a server or violation of a quota, the method of claim 48 selects the next most highly recommended action for the prospective target from the action plan retrieved from FIFO 46.

Executing Action Plans (Alternative 2—Permitting Servers to Pull Plans for Execution)

In other embodiments, step 48 of the illustrated procedure 32 permits a server 13, 15, 17 to request (or “pull”) a list of actions destined for it for one or more prospective targets and, thereby, to execute those actions when any of those targets next interacts with the server. For servers, like web servers, for which interactions demand real-time response, this minimizes the risk that interaction with any given prospective target will occur before the next best action for that target is executed.

According to this alternative embodiment, step 48 retrieves and acts on action plans in the cache 46 along the lines described above, as modified by the steps below:

• • In response to a request from a server, the method of step 48 pulls a set of n prioritized actions for a particular delivery mechanism (or server 13, 15, 17) by executing a request in the following form on the cache 46
    • Call pull(destID, n)→set of n highest priority actions or less if there are no actions available:
      • Select * from Cache where DestinationID=destID and TimeToCommunicate<=now( ) ordered by priority desc limit n;
  • The subsequent ResultSet is checked in step 48 to insure it abides by:
    • Contact policies (e.g., table of customerID×DestinationID with counts for a rolling time window (e.g., last 30 days))
    • Action constraints (e.g., table of ActionID×DestinationID with counts for a rolling time window (e.g., This action can only be sent via email 100,000 times per day))
    • Treatment constraints (e.g., table of TreatmentID with counts for a rolling time window (a constraint on a particular variation of an action, e.g., a limit of 500 per day on the direct mailing variant of an action to send cash-off coupons to customers who recently opened credit accounts))
  • In the discussion immediately above:
    • Counts can be replaced by values to meet budget constraints. (10,000 possible responses)
    • When an action does not pass a constraint, it's skipped. If there are fewer than ‘n’ actions left, a new query is executed against the cache 46.
  • The result-set is returned to the requesting server, and every prospective target in this set is submitted for re-evaluation, e.g., by sending their respective target ID's to the next-best-action step 36, thereby, triggering generation of a further action plan for each.

Throughput

Illustrated system 10 can be operated so as to generate at transmit 100,000 or more digital content pieces per day and, more preferably, 1,000,000 or more such digital content pieces per day, still more preferably over 10,000,000 such pieces per day. Execution of the procedure 32 and its constituent steps as discussed above to achieve such rates is within the ken of those skilled in the art in view of the teachings above.

Conclusion

Systems and methods according to the invention achieve the desired objects, among others. It will be appreciated that the embodiments shown in the drawings and discussed in the text above are merely examples of the invention and that other embodiments, incorporating changes thereto within the ken of those skilled in the art, fall within the scope of the invention.

Claims

1. A method of mass generation of customized digital content, comprising executing on a digital data processing system steps of: A. identifying, continuously, current external events taken by or with respect to a plurality of respective prospective targets of customized digital content pieces, where each external event is one taken (i) by the respective prospective target, or (ii) with respect to that prospective target other than as part of this method of customized digital content generation,B. generating, continuously and upon identification of a current event taken with respect to a said respective prospective target, a set of one or more actions, each identifying a digital content piece and a digital delivery mechanism therefor, where each such action is generated based on (i) one or more current external events identified in step (A) for the respective prospective target, and (ii) any of prior information about, and events taken by or with respect to, that respective prospective target as reflected in a database of information and prior events (the “database”),C. queuing to a digital cache each set of actions continuously and upon its generation in step (B),D. retrieving, continuously and on a first-in-first-out basis, sets queued to the digital cache, and selecting, continuously and upon retrieval of each such set, an action for generation of digital content customized for the respective prospective target of that action, where the action is selected from the set of one or more actions based on quotas associated with (i) that prospective target and (ii) the delivery mechanism identified with that action,E. invoking, continuously and on selection of a said action in step (D), a content server for the delivery mechanism identified with that selected action to transmit to the respective prospective target the digital content piece identified with that selected action,F. responding to a request by a content server associated with a said delivery mechanism by i) pulling a set of n prioritized actions for that said delivery mechanism, where the pulling step includes executing a request in the following form on the digital cache call pull(destID, n)→set of n highest priority actions or less if there are no actions available: select * from Cache where DestinationID=destID and TimeToCommunicate<=now( ) ordered by priority desc limit n,where destID and DestinationID are an identification associated with that said delivery mechanism, and TimeToCommunicate is a time-based parameter,ii) transmitting the set of prioritized actions to the requesting content server, andG. responding, with the requesting content server, to interaction between the associated said delivery mechanism and a prospective target by executing a said action generated for that prospective target in the set of prioritized actions in order to generate digital content therefor.

2. The method of claim 1, wherein the step of identifying current external events includes ingesting one or more incoming streams of events.

3. The method of claim 2, wherein the step of ingesting streams of event includes ingesting streams from any of the web, social media, call centers, transaction servers, and data warehouses.

4. The method of claim 1, wherein the current external events occur asynchronously with respect to one another.

5. The method of claim 1, wherein the step of identifying current external events includes identifying those events substantially in real time with their respective occurrences.

6. The method of claim 1, wherein the step of identifying current external events includes memorializing said external events to the database.

7. The method of claim 1, comprising generating, continuously and concurrently with the step of identifying current external events, identifiers of respective prospective targets, where those identifiers are generated on a round-robin basis from among a list of identifiers of known respective prospective targets.

8. The method of claim 7, wherein the step of generating identifiers of respective prospective targets includes winnowing from the list of identifiers of known respective prospective targets those targets for whom a current external event has been identified.

9. The method of claim 7, wherein the step of generating identifiers includes generating identifiers of respective prospective targets asynchronously with respect to Identification of current external events with respect to those respective prospective targets.

10. The method of claim 7, wherein step (B) includes generating, continuously and upon generation of an identifier of a respective prospective target, a set of one or more actions based on prior information about, and events taken by or with respect to, that respective prospective target as reflected in the database.

11. The method of claim 1, comprising invoking one or more said content servers to transmit over 100,000 digital content pieces to different respective targets per day.

12. The method of claim 11, comprising invoking one or more said content servers to transmit over 1,000,000 digital content pieces to different respective targets per day.

13. The method of claim 12, comprising invoking one or more content servers to transmit over 10,000,000 digital content pieces to different respective targets per day.

14. The method of claim 1, comprising memorializing to the database, on invocation of a said content server to transmit a said digital content piece to a said prospective target, a fact of that transmission to that prospective target.

15. The method as in any one of claims 1 and 10, in which step (B) includes generating the actions additionally based on a library of possible actions.

16. The method as in any one of claim 1 and 10, in which the step of queuing a set of actions to the digital cache for a respective prospective target includes removing from the cache any prior stored sets of actions for that target.

17. A method of mass generation of customized digital content, comprising executing on a digital data processing system steps of: A. identifying, continuously, current external events taken by or with respect to a plurality of respective prospective targets of customized digital content pieces, where each external event is one taken (i) by the respective prospective target, or (ii) with respect to that prospective target other than as part of this method of customized digital content generation,B. generating, continuously and concurrently with the step of identifying current external events in step (A), identifiers of respective prospective targets, where those identifiers are generated on a round-robin basis from among a list of identifiers of known respective prospective targets,C. generating, continuously and upon identification of a current event taken with respect to a said prospective target, a set of one or more actions, each identifying a digital content piece and a delivery mechanism therefor, where each such action is generated based on (i) one or more current external events, if any, identified in step (A) for the respective prospective target, and (ii) any of prior information about, and events taken by or with respect to, that respective prospective target as reflected in a database of information and prior events (the “database”),D. queuing to a digital cache each set of actions continuously and upon its generation in step (C),E. retrieving, continuously and on a first-in-first-out basis, sets queued to the digital cache, and selecting, continuously and upon retrieval of each such set, an action for generation of digital content customized for the prospective target of that action, where the action is selected from the set of one or more actions based on quotas associated with (i) that prospective target and (ii) the delivery mechanism identified for that action,F. invoking, continuously and on selection of a said action in step (D), a content server for the delivery mechanism identified for that selected action to transmit to the prospective target the digital content piece identified with that selected action,F. responding to a request by a content server associated with a said delivery mechanism by i) pulling a set of n prioritized actions for that said delivery mechanism, where the pulling step includes executing a request in the following form on the digital cache call pull(destID, n)→set of n highest priority actions or less if there are no actions available: select * from Cache where DestinationID=destID and TimeToCommunicate<=now( ) ordered by priority desc limit n,where destID and DestinationID are an identification associated with that said delivery mechanism, and TimeToCommunicate is a time-based parameter,ii) transmitting the set of prioritized actions to the requesting content server, andG. responding, with the requesting content server, to interaction between the associated said delivery mechanism and a prospective target by executing a said action generated for that prospective target in the set of prioritized actions in order to generate digital content therefor.

18. The method of claim 17, comprising invoking one or more content servers to transmit over 100,000 digital content pieces to different respective targets per day.

19. The method of claim 17, wherein the step of generating identifiers includes generating identifiers of respective prospective targets asynchronously with respect to Identification of current external events with respect to those respective prospective targets.

20. The method of claim 17, wherein the step of identifying current external events includes ingesting one or more incoming streams of events.

21. The method of claim 20, wherein the step of ingesting streams of event includes ingesting streams from any of the web, social media, call centers, transaction servers, and data warehouses.

22. A method of mass generation of customized digital content, comprising executing on a digital data processing system steps of: A. identifying, continuously, current external events taken by or with respect to a plurality of respective prospective targets of customized digital content pieces, where each external event is one taken (i) by the respective prospective target, or (ii) with respect to that prospective target other than as part of this method of customized digital content generation,B. generating, continuously and upon identification of a current event taken with respect to a said respective prospective target, a set of one or more actions, each identifying a digital content piece and a digital delivery mechanism therefor, where each such action is generated based on (i) one or more current external events identified in step (A) for the respective prospective target, and (ii) any of prior information about, and events taken by or with respect to, that respective prospective target as reflected in a database of information and prior events (the “database”),C. queuing to a digital cache each set of actions continuously and upon its generation in step (B),D. responding to a request by a content server associated with a said delivery mechanism by i) pulling a set of n prioritized actions for that said delivery mechanism, where the pulling step includes executing a request in the following form on the digital cache call pull(destID, n)→set of n highest priority actions or less if there are no actions available: select * from Cache where DestinationID=destID and TimeToCommunicate<=now( ) ordered by priority desc limit n,where destID and DestinationID are an identification associated with that said delivery mechanism, and TimeToCommunicate is a time-based parameter,ii) transmitting the set of prioritized actions to the requesting content server, andE. responding, with the requesting content server, to interaction between the associated said delivery mechanism and a prospective target by executing a said action generated for that prospective target in the set of prioritized actions in order to generate digital content therefor.

23. A method of mass generation of customized digital content, comprising executing on a digital data processing system steps of: A. identifying, continuously, current external events taken by or with respect to a plurality of respective prospective targets of customized digital content pieces, where each external event is one taken (i) by the respective prospective target, or (ii) with respect to that prospective target other than as part of this method of customized digital content generation,B. generating, continuously and upon identification of a current event taken with respect to a said respective prospective target, a set of one or more actions, each identifying a digital content piece and a digital delivery mechanism therefor, where each such action is generated based on (i) one or more current external events identified in step (A) for the respective prospective target, and (ii) any of prior information about, and events taken by or with respect to, that respective prospective target as reflected in a database of information and prior events (the “database”),C. queuing to a digital cache each set of actions continuously and upon its generation in step (B),D. responding to a request by a content server associated with a said delivery mechanism by i) pulling a set of n prioritized actions for that said delivery mechanism, where the pulling step includes executing a request in the following form on the digital cache call pull(destID, n)→set of n highest priority actions or less if there are no actions available: select * from Cache where DestinationID=destID and TimeToCommunicate<=now( ) ordered by priority desc limit n,where destID and DestinationID are an identification associated with that said delivery mechanism, and TimeToCommunicate is a time-based parameter,ii) transmitting the set of prioritized actions to the requesting content server, andE. responding, with the requesting content server, to interaction between the associated said delivery mechanism and a prospective target by executing a said action generated for that prospective target in the set of prioritized actions in order to generate digital content therefor.