WO2016195594A1

WO2016195594A1 - Method and apparatus for selectively allocating data to different online delivery platforms

Info

Publication number: WO2016195594A1
Application number: PCT/SG2016/050250
Authority: WO
Inventors: Vikram Bansal
Original assignee: Aegis Media Asia Pacific Management Pte Ltd
Priority date: 2015-06-02
Filing date: 2016-05-26
Publication date: 2016-12-08

Abstract

The method for selectively allocating data/advertisements to different online delivery platforms for transmission in a communications network, the data including product/service information for delivery to online users, the method comprising applying different attribution models to historical user data to calculate respective probabilities of causing the online users to take action in response to the product/service information being delivered to the online users based on an initial data allocation plan; calculated probabilities with actual action taken by the users to identify a most predictive attribution model; applying the most predictive attribution model to the user chronological journeys to determine strength of each online delivery platform and based on the determined strengths to compute an optimized data allocation plan and apportioning the product/service information to be transmitted based on the optimized data allocation plan, instead of the initial data allocation plan.

Description

Method and Apparatus for Selectively Allocating Data to

Different Online Delivery Platforms

Background and Field

The invention relates to computer technology.

With enormous growth of the Internet and the ever growing network of devices, such as computers, mobile phones etc., allocation of network resource have become a more important challenge. Further, more and more communication service providers are introducing data-intensive services such as interactive video over mobile phones or multi-user conference calling or video conferencing. Indeed, with Internet of Things, there would be greater demand for better optimization of the network resource in order to ensure that demands from such networked devices are met. This is particularly important since capacity provided by the Internet backbone is shared which means that with more and more internet ready devices transmitting and receiving data, there would be greater Internet traffic through the Internet backbone. Thus, it is desirable to provide an apparatus and method of allocating a network resource in a communications network which addresses at least one of the disadvantages of the prior art and/or to provide the public with a useful choice.

Summary

In a first aspect, there is provided a method of selectively allocating data to different online delivery platforms for transmission in a communications network, the data including product/service information for delivery to online users, the method comprising

(i) applying different attribution models to historical user data to calculate respective probabilities of causing the online users to take action in response to the product/service information being delivered to the online users based on an initial data allocation plan; the historical user data including respective user chronological journeys which are computed from tracking exposure of the product/service information to the online users on each online delivery platform;

(ii) comparing the calculated probabilities with actual action taken by the users to identify a most predictive attribution model from the different attribution models;

(iii) applying the most predictive attribution model to the user chronological journeys to determine strength of each online delivery platform in contributing to the user taking action in response to the product/service information;

(iv) iterating through combinations of the online delivery platforms based on the determined strengths to compute an optimized data allocation plan which includes a combination of the online delivery platforms which achieves a highest causation rate; and

(v) apportioning the product/service information to be transmitted based on the optimized data allocation plan, instead of the initial data allocation plan, and allocating the apportioned product/service information to at least some of the delivery platforms of the optimized data allocation plan for transmission to the online users. Since the data is transmitted selectively based on the optimized data allocation plan, network resource of the communication network is freed up for other devices to use. In this way, the described embodiment is able to optimize the performance of the communication network achieving greater versatility, efficiency and productivity.

The first aspect may take the form of a system or apparatus implementation and thus, according to a second aspect there is provided apparatus for selectively allocating data to different online delivery platforms for transmission in a communications network; the data including product service information for delivery to online users, the apparatus comprising

(i) an attributor for applying different attribution models to historical user data to calculate respective probabilities of causing the online users to take action in response to the product/service information being delivered to the online users based on an initial data allocation plan; the historical user data including respective user chronological journeys which are computed from tracking exposure of the product/service information to the online users on each online delivery platform;

(ii) a comparator for comparing the calculated probabilities with actual action taken by the users to identify a most predictive attribution model from the different attribution models;

(iii) a platform comparator for applying the most predictive attribution model to the user chronological journeys to determine strength of each online delivery platform in contributing to the user taking action in response to the product/service information;

(iv) an optimizer for iterating through possible combinations of the online delivery platforms based on the determined strengths to compute an optimized data allocation plan which includes a combination of the online delivery platforms which achieves a highest causation rate; and for apportioning the product/service information to be transmitted based on the optimized data allocation plan, instead of the initial data allocation plan, and allocating the apportioned product/service information to at least some of the delivery platforms of the optimized data allocation plan for transmission to the online users.

The described embodiment may be specifically adapted for allocating online advertisements since pushing of online advertisements to users via different online delivery platforms do take up network resources. Consequently, in a third aspect there is provided a method of selectively allocating advertisements to different online delivery platforms for transmission in a communications network to online users, the method comprising:

(i) applying different attribution models to historical user data to calculate respective probabilities of inducing conversions by the online users in response to the advertisements being delivered to the online users based on an initial media plan; the historical user data including respective user chronological journeys which are computed from tracking exposure of the advertisements to the online users on each online delivery platform; (ii) comparing the calculated probabilities with actual conversions to identify a most predictive attribution model from the different attribution models;

(iii) applying the most predictive attribution model to the user chronological journeys to determine strength of each online delivery platform in contributing to the conversions;

(iv) iterating through possible combinations of the online delivery platforms based on determined strengths to compute an optimized media plan which includes a combination of the online delivery platforms which achieves the highest conversion rate; and

(v) apportioning the advertisements to be transmitted based on the optimized media plan, instead of the initial media plan and allocating the apportioned advertisements to at least some of the delivery platforms of the optimized media plan for transmission to the online users. In addition to optimizing the use of network resources in the communication network, it has been found that the described embodiment is able to predict and drive improvements to e-commerce revenue. To achieve this, millions of consumer interaction data may be analysed to obtain insights into disaggregate data for both conversions and non-conversions. The optimized media plan may then be executed in real-time to ensure that the delivery of the advertisements are optimized using a combination of the online delivery platforms to maximize the effectiveness of ensuring greater returns on investment.

Preferably, calculating the respective probabilities for the different attribution models may include adding the probabilities associated with each online platform and user chronological journey to produce respective prediction values associated with corresponding attribution models and comparing the prediction values against a threshold to determine predictions of the corresponding attribution models. The method may further comprise assigning a score for each prediction if the prediction value exceeds a threshold.

Advantageously, the method may further comprise allocating a greater portion of the advertisements to a selected one of the delivery platforms having a highest determined strength. Accordingly, the method may also further comprise checking availability of additional advertisement inventory associated with the selected delivery platforms and if no additional advertisement inventory is available, allocating the greater portion of the advertisements to a next delivery platform with the next higher determined strength.

It is envisaged that the method may further comprise repeating steps (i) to (iv) to compute further optimized media plans at predetermined time intervals, and allocating the advertisements based on further optimized media plans for the respective time intervals. The predetermined time intervals may vary depending on requirements and may include daily, fortnightly or monthly intervals.

The third aspect may take the form of a system or apparatus implementation and thus, according to a fourth aspect, there is provided apparatus for selectively allocating advertisements to different online delivery platforms for transmission in a communications network to online users, the apparatus comprising:

(i) an attributor applying different attribution models to historical user data to calculate respective probabilities of inducing conversions by the online users in response to the advertisements being delivered to the online users based on an initial media plan; the historical user data including respective user chronological journeys which are computed from tracking exposure of the advertisements to the online users on each online delivery platform;

(ii) a comparator for comparing the calculated probabilities with actual conversions to identify a most predictive attribution model from the different attribution models;

(iii) a platform comparator for applying the most predictive attribution model to the user chronological journeys to determine strength of each online delivery platform in contributing to the conversions;

(iv) an optimizer for iterating through possible combinations of the online delivery platforms based on determined strengths to compute an optimized media plan which includes a combination of the online delivery platforms which achieves the highest conversion rate; and for apportioning the advertisements to be transmitted based on the optimized media plan, instead of the initial media plan and allocating the apportioned advertisements to at least some of the delivery platforms of the optimized media plan for transmission to the online users.

A general architecture of the described embodiment may form a fifth aspect in which there is provided apparatus for selectively allocating advertisements to different online delivery platforms for transmission in a communications network to online users, the apparatus comprising:

(i) one or more processors;

(ii) a system memory in electrical communication with the one or more processor;

(iii) a network interface in electrical communication with the one or more processors to electronically couple the apparatus to a communication network, the communication network having a number of computing devices that provides respective display websites for viewing by the online users,

(iv) an attributor applying different attribution models to historical user data to calculate respective probabilities of inducing conversions by the online users in response to the advertisements being delivered to via different online delivery platforms to the display websites based on an initial media plan; the historical user data including respective user chronological journeys which are computed from tracking exposure of the advertisements to the online users on each online delivery platform;

(v) a comparator for comparing the calculated probabilities with actual conversions to identify a most predictive attribution model from the different attribution models;

(vi) a platform comparator for applying the most predictive attribution model to the user chronological journeys to determine strength of each online delivery platform in contributing to the conversions;

(vii) an optimizer for iterating through possible combinations of the online delivery platforms based on determined strengths to compute an optimized media plan which includes a combination of the online delivery platforms which achieves the highest conversion rate; and for apportioning the advertisements to be transmitted based on the optimized media plan, instead of the initial media plan and allocating the apportioned advertisements to at least some of the delivery platforms of the optimized media plan for transmission to the display websites for display to the online users.

It should be appreciated that features relevant to one aspect may also be relevant to the other aspects. Brief Description of the Drawings

Exemplary embodiments will now be described with reference to the accompanying drawings, in which: Figure 1 is a schematic diagram of a communications network according to an embodiment of this invention;

Figure 2 is a schematic diagram of a computer system suitable for implementing any of the devices of the communication network of Figure 1 ;

Figure 3 illustrates a software environment which may be implemented by the computer system of Figure 2;

Figure 4 illustrates an alternative software environment which may be implemented by the computer system of Figure 2;

Figure 5 is a flow chart illustrating steps for computing an optimised data allocation plan for the communication network of Figure 1 ;

Figure 6 illustrates an example of how five attribution models being used in Figure 5 are used to quantify probabilities from three websites; and Figure 7 is a table illustrating results after applying a most predictive attribution model to user data in order to compute which combination of the online platforms achieves a highest conversion or causation rate. Detailed Description of Preferred Embodiment

Figure 1 shows a communication network or network apparatus 100 comprising "n" number of computing devices such as servers 200 hosting web contents and in this embodiment, there is a first server 202 hosting web content for a theme park USP, a second server 204 hosting a search engine such as Google™ or Yahoo!™ and a third server 206 hosting social media such as Facebook™. The communications network 100 further includes a central computing device 208 arranged to receive and transmit data from the "n" number of computing devices, and this will be further elaborated later.

The communication network 100 further includes "m" number of communication devices 300 configured to communicate with the servers 202,204,206 over the internet 102 or over communication links. These communication devices may include laptops, personal computers (PC), mobile phones, personal digital assistants (PDAs), gaming devices, media players, tablets, wearable computers, headset computers, in-vehicle computers etc. In this embodiment and for ease of explanation, the communication devices 300 include a first PC 302, a second PC 304, a third PC 306 and a mobile phone 308.

Figure 2 is a schematic diagram of a computer system 310 suitable for implementing one or more of the communication devices 300, the servers 200, and the central computing device 208 of Figure 1. The computer system 310 includes a processor 312 (which may be commonly referred to as a central processor unit (CPU)) that is in communication with memory devices including secondary storage 314 (such as a Hard Disk Drive), read only memory (ROM) 316, random access memory (RAM) 318, input/output (I/O) device 320, and network connectivity device 322 so that the computer system 310 may communicate with other computer systems. As is known, the processor 312 may be implemented as one or more CPU chips.

It should be appreciated that programming and/or loading executable instructions onto the computer system 310, this would cause the processor 312, the ROM 316 and the RAM 318 to operate and execute the instructions, transforming the computer system 310 in part into a particular machine or apparatus having the novel functionality taught by the present disclosure. It is fundamental to the electrical engineering and software engineering arts that functionality that can be implemented by loading executable software into a computer can be converted to a hardware implementation by known computer engineering design rules. Quite often, a decision has to be made between implementing a solution in software or hardware and this may hinge on considerations of stability of the design and numbers of units to be produced, rather than any issues involved in translating from the software domain to the hardware domain. Generally, a solution that is still subject to frequent change may be preferred to be implemented in software, because re-spinning a hardware implementation is more expensive than re-spinning a software solution. Generally, a design that is stable that will be produced in large volume may be preferred to be implemented in hardware, for example in an application specific integrated circuit (ASIC), because for large production runs the hardware implementation may be less expensive than the software implementation. Often a design may be developed and tested in a software form and later transformed, by using design rules, to an equivalent hardware implementation in an application specific integrated circuit that hardwires the instructions of the software. In the same manner as a machine controlled by a new ASIC is a particular machine or apparatus, likewise a computer that has been programmed and/or loaded with executable instructions may be regarded as a particular machine or apparatus.

Additionally, after the computer system 310 is turned on or booted, the processor 312 may execute a computer program or application. For example, the processor 312 may execute software or firmware stored in the ROM 316 or stored in the RAM 318. In some cases, on boot and/or when the application is initiated, the processor 312 may copy the application or portions of the application (or application modules) from the secondary storage 314 to the RAM 318 or to memory space within the processor 312 itself, and the processor 312 may then execute instructions of the computer application. In some cases, the processor 312 may copy the application or portions of the application from memory accessed via the network connectivity devices 322 or via the I/O devices 320 to the RAM 318 or to memory space within the processor 312, and the processor 312 may then execute instructions of the computer application. During execution, an application may load instructions into the processor 312, for example load some of the instructions of the application into a cache of the processor 312. In some contexts, an application that is executed may be said to configure the processor 312 to do something, e.g., to configure the processor 312 to perform the function or functions defined by the instructions of the application. When the processor 312 is configured in this way by the application, the processor 312 becomes a specific purpose computer or a specific purpose machine.

The secondary storage 314 may include one or more disk drives or tape drives and is used for non-volatile storage of data and as an over-flow data storage device if the RAM 318 is not large enough to hold all working data. The secondary storage 314 may be used to store programs which are loaded into the RAM 318 when such programs are selected for execution. The ROM 316 is used to store instructions and perhaps data which are read during program execution. The ROM 316 is a non-volatile memory device which may have a small memory capacity relative to the larger memory capacity of the secondary storage 314. The RAM 318 is used to store volatile data and perhaps to store instructions. Access to both the ROM 316 and RAM 318 is typically faster than to the secondary storage 314. The secondary storage 314, the RAM 318, and/or the ROM 316 may be referred to in some contexts as computer readable storage media and/or non-transitory computer readable media.

The I/O device 320 may include printers, video monitors, liquid crystal displays (LCDs), plasma displays, touch screen displays, keyboards, keypads, switches, dials, mice, track balls, voice recognizers, card readers, paper tape readers, or other known input and output devices which allows a user to interact with or provides inputs to or obtain outputs from the computer system 310. The network connectivity device 322 may take the form of modems, modem banks, Ethernet cards, universal serial bus (USB) interface cards, serial interfaces, token ring cards, fiber distributed data interface (FDDI) cards, wireless local area network (WLAN) cards, radio transceiver cards that promote radio communications using protocols such as code division multiple access (CDMA), global system for mobile communications (GSM), long-term evolution (LTE), worldwide interoperability for microwave access (WiMAX), near field communications (NFC), radio frequency identity (RFID), and/or other air interface protocol radio transceiver cards, and other network devices. This network connectivity device 322 may enable the processor 312 to communicate with the Internet 102 or one or more intranets and in particular with the other communication devices 300, servers 200 or the central computing device 208. With such a network connection over the communication network 100, it is contemplated that the processor 312 might receive information from the communication network 100, or might output information to the network 100 in the course of performing the above-described method steps. Such information, which is often represented as a sequence of instructions to be executed using processor 312, may be received from and outputted to the network 100, for example, in the form of a computer data baseband signal or signal embodied in a carrier wave.

The baseband signal or signal embedded in the carrier wave, or other types of signals currently used or hereafter developed, may be generated according to several methods known to one skilled in the art. The baseband signal and/or signal embedded in the carrier wave may be referred to in some contexts as a transitory signal or communication signal carrying data or information.

Specifically, the processor 312 executes instructions, codes, computer programs, scripts which the processor 312 accesses from the secondary storage 314, the ROM 316, the RAM 318, or the network connectivity devices 322. While only one processor 312 is shown in Figure 2, multiple processors may be used. Thus, while instructions may be discussed as executed by a processor, the instructions may be executed simultaneously, serially, or otherwise executed by one or multiple processors. Instructions, codes, computer programs, scripts, and/or data that may be accessed from the secondary storage 314, for example, hard drives, floppy disks, optical disks, and/or other device, the ROM 316, and/or the RAM 318 may be referred to in some contexts as non-transitory instructions and/or non-transitory information. The computer system 310 may comprise a secure element and associated near field communication transceiver or RF transceiver for wireless communications, particularly if the computer system 310 is in the form of the mobile phone 308. In an embodiment, the computer system 310 may comprise two or more computers in communication with each other that collaborate to perform a task. For example, but not by way of limitation, an application may be partitioned in such a way as to permit concurrent and/or parallel processing of the instructions of the application. Alternatively, the data processed by the application may be partitioned in such a way as to permit concurrent and/or parallel processing of different portions of a data set by the two or more computers. In an embodiment, virtualization software may be employed by the computer system 310 to provide the functionality of a number of server systems (such as forming one of the servers 200 of Figure 1 ) that is not directly bound to the number of computers in the computer system 310. For example, virtualization software may provide twenty virtual servers on four physical computer systems 310. In an embodiment, the functionality disclosed above may be provided by executing the application and/or applications in a cloud computing environment. Cloud computing may comprise providing computing services via a network connection using dynamically scalable computing resources. Cloud computing may be supported, at least in part, by virtualization software. A cloud computing environment may be established by an enterprise and/or may be hired on an as-needed basis from a third party provider. Some cloud computing environments may comprise cloud computing resources owned and operated by the enterprise as well as cloud computing resources hired and/or leased from a third party provider. In an embodiment, some or all of the functionality disclosed above may be provided as a computer program product. The computer program product may comprise one or more computer readable storage medium having computer usable program code embodied therein to implement the functionality disclosed above. The computer program product may comprise data structures, executable instructions, and other computer usable program code. The computer program product may be embodied in removable computer storage media and/or non-removable computer storage media. The removable computer readable storage medium may comprise, without limitation, a paper tape, a magnetic tape, magnetic disk, an optical disk, a solid state memory chip, for example analog magnetic tape, compact disk read only memory (CD-ROM) disks, flash memory, jump drives, digital cards, multimedia cards, and others. The computer program product may be suitable for loading, by the computer system 310, at least portions of the contents of the computer program product to the secondary storage 314, to the ROM 316, to the RAM 318, and/or to other non-volatile memory and volatile memory of the computer system 310. The processor 312 may process the executable instructions and/or data structures in part by directly accessing the computer program product, for example by reading from a CD-ROM disk inserted into a disk drive peripheral of the computer system 310 with the disk drive peripheral as an example of the I/O device 320. Alternatively, the processor 312 may process the executable instructions and/or data structures by remotely accessing the computer program product, for example by downloading the executable instructions and/or data structures from a remote server through the network connectivity device 312. The computer program product may comprise instructions that enable the loading and/or copying of data, data structures, files, and/or executable instructions to the secondary storage 314, to the ROM 316, to the RAM 318, and/or to other non-volatile memory and volatile memory of the computer system 310.

Figure 3 illustrates a software environment 330 that may be implemented by the processor 312 of the computer system 310. The processor 312 executes operating system software 334 which is typically stored in the secondary storage 314 and that provides a platform from which the rest of the software operates. The operating system software 334 may provide a variety of drivers in order for hardware of the computer system 310 to work with the operating system software 334. The operating system software 334 may be coupled to and interact with application management services (AMS) 336 that transfer control between applications running on the computer system 310. Examples of applications running on the computer system 310 are a web browser application 338, a media player application 340, JAVA applets 342, and computer applications (App) such as various application modules for implementing the described embodiment which will be elaborated later. The web browser application 338 may be executed by the processor 312 (or more generally the computer system 310) to browse content and/or the Internet 102, for example, to access the web content provided by one of the servers 200 via the internet 102. The web browser application 338 may permit a user to enter information into forms, select links to retrieve and view web pages, download information from the internet (or from other communication devices 300 or computing devices 200) and for storage in the secondary storage 314 or to upload information from the secondary storage 314 to the internet 102 or when requested by other communication devices 300 or computing devices 300. Thus, it can be appreciated that the communication network 100 includes devices 200,300 which interact or communicate with each other, and in particular exchange data.

The media player application 340 may be executed by the processor 312 of the computer system 310 to play audio or audiovisual media such as audio or audiovisual files from the secondary storage 314, the I/O devices (optical disks), or streamed or downloaded from the internet 102. The JAVA™ applets 342 may be executed by the processor 312 of the computer device 310 to provide a variety of functionality including games, utilities, and other functionality. Similarly, the computer system 310 may also have other computer applications 344 stored in the memory devices such as the secondary storage 314 which may be run by the processor 312 to perform various tasks or provide other functionalities. Figure 4 is a schematic diagram which illustrates an alternative software environment 350 that may be implemented by the processor 312 of the computer system 310. The processor 312 executes operating system kernel (OS kernel) 352 and an execution runtime 354. The processor 312 executes computer applications 356 that may execute in the execution runtime 354 and may rely upon services provided by an application framework 358. The applications 356 and the application framework 358 may rely upon functionality provided via libraries 360.

Referring again to Figure 1 , with introduction of next generation internet services which are data intensive (and thus, requiring more bandwidth), and with millions of intemet communication devices 300 accessing the internet at any one time, this puts a strain on the capacity of the communication network 100, particularly the Internet backbone where bandwidth is shared. Thus, there should be a better way of transmitting data in order to improve or optimize network resource to meet the demands of next generation services.

For simplicity, let's imagine first, second and third users of the first, second and third PC 302, 304, 306 respectively initiate web browser applications 338 to access the web contents of the first server 202 at the same time and in theory, the first server 202 should deliver the entire landing page of the web contents to the web browser applications of the first, second and third PC 302, 304, 306. The web contents being displayed or exposed to the first, second and third users are thus the same but in reality, the viewing habits and interest of these users vary and a particular content which interest one user may not interest the other. Thus, certain information being transmitted and displayed or exposed to one of the users may cause that user to take further action, such as clicking on one of the links or images to find out more about the delivered content or perhaps cause the user to purchase tickets for the USP theme park via the first server 202 but for the other users, they may not take any further actions for various reasons. Thus, for that user which is likely to take further action on the web contents of the first server 202, certain web contents may be delivered, while for the other users, certain data of the web content may not be delivered thus freely up network resources for others.

Moreover, accessing the web contents of the first server 202 directly is but one online delivery platform and there are other types of online delivery platforms such as the search engine provided by the second server 204 and the social media content provided by the third server 206. Thus, these different online platforms may appeal to or interests the other users of the first, second and third PCs 302,304,306. Certainly, for users who are not aware of the USP theme park, the other online platforms offer alternative avenues of reaching out to more users or potential customers. Thus, it is proposed to track the online behavior or habit of the users and to determine possible combinations of the online delivery platforms which provides for the highest probability of the users taking further actions to compute an optimized network resource allocation plan. In this way, the data may be transmitted selectively and delivered to various target users, resulting in a much more efficient performance for the communications network 100.

The present embodiment would be further explained by using digital or online advertisements as an example data, since more and more advertisements use images and video which demand greater bandwidth to deliver, than say text transmissions. Specifically, reference is made to a campaign for the USP theme park to explain how the network resource may be optimized for USP's advertisements.

Figure 5 is a flow chart illustrating steps to compute an optimized data allocation plan (or optimized media plan) automatically for the USP theme park, and in this embodiment, the computation is performed by the central computing device 208, or central server. It should be appreciated that the computation may also be performed by the first server 202 or by other servers or computing devices.

At 501 , user data is collected over a certain period of time, such as a few months or a year, for example. At a start, conversion metrics may be identified for the USP theme park and such metrics may take the form of revenue on USP's web hosted on the first server 202 or clicks on specific buttons on their web properties. A common off-the-shelf tool may be Doubleclick Campaign Manager (DCM) by Google™ to traffic online display ads and monitor campaign performance. DCM comes with sales tags to track sales metrics, and counter tags to track metrics such as click impressions. The DCM may be running on another server or on the central server 208.

New tags may then be added when USP expands their web properties or when new clients adopts a similar campaign.

During the data collection or harvesting period, the DCM would traffic online adverstiments (or ads) for the USP theme park. Specifically, from the DCM ad server which in this case is the central server 208, ads are communicated to browsers 338 of target web audiences via various online platforms or channels in a preliminary or initial media plan and in this embodiment, and for simplicity, the online platforms are the search engine on the second server 204 and the social media platform on the third server 206 . As ads are shown to web audiences, DCM drops cookies that can individually identify the browsers 338 on which the ads were shown.

Using information from the cookie, DCM keeps a log of the browsers 338 on which the ads were shown, as well as time, placement and publisher used to show the ad. The data collected is then analysed at 502. It should be mentioned that the data maybe transferred to a third party server for analsysis but in this embodiment, the data is maintained in the central server 208 for processing. The data may also be sorted and processed on cloud computing facilities such as Dentsu Aegis™ Network's propreitary Data Labs or Amazon™ Web Service's S3 and Cloud Compute.

The data collected for analaysis could be ernomous. For example, over a year period, line item records of more than 600 million ad impressions served for USP was extracted and prepared. The line items record many aspects of a user's interaction with the displayed ads - advertising impression clicks, website visits, and value of online purchaese made - all identifications of whether the user took further actions in response to the displayed ads. These records allow construction of a chronological sequence of ads shown to each browser 338 representing each user/consumer's online experience of the brand and ad exposures at a disaggregate level both for converted and non- converted customer journeys. Individual browsers are identified according to an encrypted version of their cookie identifier.

More specifically, on a daily basis, the central server 208 excutes an analytical module or analyser to analyse the records to identify browsers 338 that have seen the campaign's ads or performed a conversion event. In other words, the tags have recorded identifiers of cookies that have been dropped on the browsers 338 that users have clicked on an event button or made a purchase on the USP's website. Using these identifiers, the central server 208 is arranged to search the records from the last 30 days for the browsers' interactions with USP's online digital advertisements via the various online platforms 202,204,206. This constructs a 30-day chronology of all the online digital advertisements via the various online platforms by USP that have been shown or displayed to users using the browsers 338. Further, information about whether there was a conversion is appended to each browser's chronology, to illustrate which online platform or combinations of online platforms have induced the conversion, and in this case, the conversion means a user purchasing a ticket for the USP theme park. In other words, the historical user data is now presented as multiple browser chronologies or user journeys.

The central server 208 next executes an attribution modelling module or attributor which applies selected attribution methods to the browser chronologies at 504. These attribution methods or models use the historical user data to quantify the effectiveness of each online platform in the preliminary media plan. In this embodiment, five attribution methods are selected and they are:

1. Ordinary Least Squares (OLS) regression

2. Logistic regression

3. Shapely Value regression

4. Classic Probabilistic model

5. Markov Chain model

It should be appreciated that the type and number of attributon methods may vary depending on requirements and other types of attribution methods may be used or added such as Heuristic-based methods such as first-click and last- click.

Each attribution method quantifies the probability of conversion arising from having an online platform on the browser's chronology of ad exposures. For a given chronology of ad exposures, the probabilities associated with online platforms in the chronology are summed to produce a prediction on whether a conversion was made on that browser. If an individual chronology's summation of probabilities crosses a threshold, it is assumed that the model predicts a conversion. This prediction is then validated against the ground truth, which is based on the historical user data obtained from step 502.

Figure 6 illustrates an example of how the five selected attribution methods quantifies the probabilities from three websites A, B and C relating to one user (user ID 23456), with the three sites A, B, C being analogous to the web contents hosted by the first, second and third servers 202,204,206. Arrows 600 are pictorial representation of the probabilities of conversion predicted by the attribution methods and the longer the arrow, the higher the probability of conversion if the user is exposed to the ad on that website. For example, the OLS regression method predicts that the user (with ID:23456) has a higher probability of conversion if he is exposed to the ad by USP on website B, than on website C, since the arrow 602 associated with website B is longer than the arrow 604 for website C. If the prediction is correct, a score of 1 is awarded, whereas a score of zero means that the attribution model predicted the conversion incorrectly.

The threshold may be an average summation of the probabilities from a sampling of chronologies. For instance and ease of explanation, if the summation of probabilities across a chronology is 0.06 and the summation of probabilities across another chronology is 0.04, the threshold is then set at 0.05.

This process is repeated multiple times for all the historical user data and using each of the selected attribution methods, and the predictions from all the attribution methods are compared with each other by a comparison module. The attribution method that most often makes the correct prediction is chosen as the one that is the most predictive at setp 506. In this embodiment, the most predictive model is the Classic Probabilitic Model.

Next, at step 508, the central server 208 executes instructions of a platform comparison module or script, or a platform comparator is configured to detemine strength of the online platforms in causing or inducing the user to take further actions or by responding to the online ad being displayed to the user. This comparison is performed by running the chosen (most predictive) attribution model on the USP historical user data for the past 30 days which produces a table akin to that illustrated in Figure 7 (only part of the data is shown). For each day, the platform comparison module computes an effectiveness reading 652 of the probability of conversion arising from having a particular online platfrom in the initial media plan. As an example, the table in Figure 7 illustrates the computed readings 652 for five online channels 650, namely Facebook™ , Innity™, Pantip™, Google™ and Trip Advisor™ , with Facebook™ hosted by the third server 206 and Google™ hosted by the second server 204. Innity™, Pantip™ and Trip Advisor™ are hosted on other web servers which are not shown in Figure 1 , but these online platforms are included to illustrate that a plurality of online platforms may be used. The computed effectiveness readings 652 provide a measure of the effectiveness or strengths of each online platform as determined by the most predictive attribution method.

The central server 208 next executes instructions of an optimisation module or optimiser at step 510 which feeds the data or statistics that quantify the effectiveness of each online platform 650 from step 508 into an optimisation algorithm. The optimisation algorithm may be coded in R computer language which is suitable for statistical computing. However, other computer programs may also be used.

The optimisation algorithm is defined as:

Maximise ROI

Optimisaton model:

Minimise Risk

where

Maximise RO

Minimise Ris

where:

E: Expectation

R_p : Return on the media plan

Ri : Return on advertising asset /^'

Wi : weighting of advertising asset /^' (that is, the proportion of asset "/^'" the media plan)

Wj : weighting of advertising asset j (that is, the proportion of asset "j" the media plan) p : variance of the media plan

¹ : variance of advertising asset /

P'-j is the correlation coefficient between the returns on advertising assets /^' and j.

i : is a variable; The effect of the optimisation algorithm is that the algorithm iterates through every possible combination of the online platforms 650 in the initial media plan and based on the computed effectiveness readings to arrive at a combination of online platforms 650 that drives the highest conversion rate while maintaining the lowest fluctuation in conversion rate. In general, this also means that the combination of online platforms 650 would induce or drive the highest causation rate for the user to take further actions when exposed to the ads on a particular combination of online platforms 650. For example, in this embodiment, the optimisation algorithm may recommend using the two online platforms Google™ and Trip Advisor™, and the USP' own website hosted by the first server 202 as the best combination to achieve an optimised resource allocation. This combination of online platforms for delivering the data or information to users may also be ranked so that between these online platforms, one online platform may be allocated more resourse than the other, to determine the strengths or effectiveness of one online platform against the others.

Based on the above, an optimised media plan, as an example of an optimised data allocation plan, is computed and replaces the initial media plan for use by the USP theme park to manage its advertising resouces. For example, more ads could be channelled through the Google™ search engine for delivery and display to users when there are keyword hits since the Google™ search engine is part of the optimised media plan with a higher ranking than Trip Advisor™ for example. Further, the optimised media plan indicates that Trip Advisor™ has a higher ranking than the USP website, then more advertisement resources may be channelled to the Trip Advisor™ website.

The optimised media plan is then deployed at step 514 with the central server 208 making use of the selected online platforms 650 that form the optimised resource allocation plan for delivery of the advertisment (or product/service data) to the users and also to communicate the resource allocation between the selected online platforms 650. In this way, USP is able to be more selective in the way USP theme park targets certain users with advertisements and as a result, the USP theme park may allocate resources in a much more effective manner. Thus, this frees up the network resources of the communication network 100 for other parties to use. Preferably, after the optimised data allocation plan is deployed, the central server 208 continues to analyse historical user data as suggested in step 502 so that any new user data (between the initial media plan and deployment of the optimised media plan) is then combined with the historical user data to contruct further chronological sequence of ads. Steps 504 to 512 may then be executed again to derive a further optimised network resource allocation plan so that the network resource is further optimised. In this way, the communication network is able to optimise its resources dynamically and efficiently. Ideally, such re- optimisation may be performed daily, but for practical reasons, such re- optimisations may be performed over a longer period of time, such as every two weeks or monthly.

The described embodiment also has an unexpected use which is for budget allocation. Specifically, the optimised resource allocation plan computed in step 512 may be used by USP them park to decide how much budget to allocate to each online platforms 650 for delivery of the ads to users.

Case study: USP theme park tactical campaign

As an example of the usefulness of the described embodiment for use in budget allocation, a campaign was ran to promote a special deal for a particular theme park with the following allocation of budget based on an initial media plan to collect historical user data.

Table 1 :

Initial allocation based on Optimised allocation based on initial media plan optimised media plan

Yahoo! 25% 57%

Facebook 23% 14%

Trade Desk 52% 29% The campaign period was between 31 Jan and 18 Feb of a particular year, using three online delivery platforms - Yahoo!™, Facebook™, and Trade Desk™. The initial media budget allocated without the recommendation of the described embodiment was used for the period between 31 Jan and 9 Feb (i.e. non-optimised period) and the allocation is shown in Table 1. Based on the user data collected during this period, the flow of Figure 5 was used to derive an optimised media plan and the strengths allocated to the online platforms at the end of step 512 were Yahoo!™ (2.03), Facebook™ (2.26) and Trade Desk™ (1.76). In other words, Facebook™ provided the greatest strengths in driving conversions for the particular theme park. However, since there was no additional advertising inventory available for Facebook™, the budget was allocated to the next best choice based on their strength i.e. Yahoo!™, resulting in the budget being reallocated according to the recommended allocation of Table 1 between 10 Feb and 18 Feb. Of course, if thre is advertising inventory available for Facebook™, then Facebook™ would be allocated a higher budget in line with its higher score.

It was found that between the non-optimised period and the optimised period for that year, revenue improved by 41 % for the optimised period as compared to the non-optimised period. Further, year-on-year, for the same period, there was a lift in sales of about 6% with optimisation. Thus, the described embodiment may also be used to drive improvements to revenue.

To elaborate, when the described embodiment is specifically applied to financial budgeting, the optimisation module iteratively reallocates media financial budget according to the ecommerce revenue accrued to each online (media) platform by the attributon modelling module, so as to strike a balance between maximising conversion revenue and minimising uncertainty in the revenue that a media plan is expected to deliver.

After the optimised media plan is deployed, and when the advertising tags collect data on new impressions served, such 'new' data is added to the historical user data at step 501 of Figure 5 and the steps 502 to 512 repeated to generate an updated media plan. This way, the solution also offers a scalable solution to deploy the most-up-to-date statistics to keep USP's media plans updated. The optimised media plan then directs the choice and level of investment in the various online delivery platforms to promote the USP theme park to online consumers.

The described embodiment should not be construed as limitative. For example, if the described embodiment is used for financial budgeting, baseline revenue forecast may be determined initially so as to provide a baseline to compare improvements which may be achieved through the use of the optimised media plan.

In the described embodiment, the analytical module of the central server 208 analyse the records daily to identify browsers 338 that have seen the campaign's ads or performed a conversion event. This time interval may vary accordingly.

It should be appreciated, while the described embodiments uses the example of advertisements as an example of a data, the described embodiment is also applicable for other types of data which relates to product/service information.

Having now fully described the invention, it should be apparent to one of ordinary skill in the art that many modifications can be made hereto without departing from the scope as claimed.

Claims

1. A method of selectively allocating data to different online delivery platforms for transmission in a communications network, the data including product service information for delivery to online users, the method comprising

(v) apportioning the product/service information to be transmitted based on the optimized data allocation plan, instead of the initial data allocation plan, and allocating the apportioned product/service information to at least some of the delivery platforms of the optimized data allocation plan for transmission to the online users.

2. Apparatus for selectively allocating data to different online delivery platforms for transmission in a communications network; the data including product/service information for delivery to online users, the apparatus comprising (i) an attributor for applying different attribution models to historical user data to calculate respective probabilities of causing the online users to take action in response to the product/service information being delivered to the online users based on an initial data allocation plan; the historical user data including respective user chronological journeys which are computed from tracking exposure of the product/service information to the online users on each online delivery platform;

3. A method of selectively allocating advertisements to different online delivery platforms for transmission in a communications network to online users, the method comprising:

(v) apportioning the advertisements to be transmitted based on the optimized media plan, instead of the initial media plan and allocating the apportioned advertisements to at least some of the delivery platforms of the optimized media plan for transmission to the online users.

4. A method according to claim 3, wherein calculating the respective probabilities for the different attribution models includes adding the probabilities associated with each online platform and user chronological journey to produce respective prediction values associated with corresponding attribution models and comparing the prediction values against a threshold to determine predictions of the corresponding attribution models.

5. A method according to claim 4, further comprising assigning a score for each prediction if the prediction value exceeds a threshold.

6. A method according to any of claims 3 to 5, further comprising allocating a greater portion of the advertisements to a selected one of the delivery platforms having a highest determined strength.

7. A method according to claim 6, further comprising checking availability of additional advertisement inventory associated with the selected delivery platforms and if no additional advertisement inventory is available, allocating the greater portion of the advertisements to a next delivery platform with the next higher determined strength.

8. A method according to any of claims 3 to 7, further comprising repeating steps (i) to (iv) to compute further optimized media plans at predetermined time intervals, and allocating the advertisements based on further optimized media plans for the respective time intervals.

9. A method according to claim 8, wherein the predetermined time intervals include daily, fortnightly or monthly.

10. Apparatus for selectively allocating advertisements to different online delivery platforms for transmission in a communications network to online users, the apparatus comprising:

11. Apparatus for selectively allocating advertisements to different online delivery platforms for transmission in a communications network to online users, the apparatus comprising:

(i) one or more processors;