WO2008070834A2 - Procédé d'automatisation de vente sur des chaînes numériques - Google Patents

Procédé d'automatisation de vente sur des chaînes numériques Download PDF

Info

Publication number
WO2008070834A2
WO2008070834A2 PCT/US2007/086784 US2007086784W WO2008070834A2 WO 2008070834 A2 WO2008070834 A2 WO 2008070834A2 US 2007086784 W US2007086784 W US 2007086784W WO 2008070834 A2 WO2008070834 A2 WO 2008070834A2
Authority
WO
WIPO (PCT)
Prior art keywords
attributes
media
user
medias
marketing
Prior art date
Application number
PCT/US2007/086784
Other languages
English (en)
Other versions
WO2008070834A9 (fr
WO2008070834A3 (fr
Inventor
Christopher Raniere
William Schaefer
Brad Smallwood
Original Assignee
Revcube Media, Inc.
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Revcube Media, Inc. filed Critical Revcube Media, Inc.
Publication of WO2008070834A2 publication Critical patent/WO2008070834A2/fr
Publication of WO2008070834A3 publication Critical patent/WO2008070834A3/fr
Publication of WO2008070834A9 publication Critical patent/WO2008070834A9/fr

Links

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
    • G06Q30/00—Commerce
    • G06Q30/02—Marketing; Price estimation or determination; Fundraising

Definitions

  • the present method relates generally to marketing campaigns and more specifically to a method of automating media content and placement in online digital channels.
  • Advertisers have long understood that the more they know about their audience, the better they are able to present a compelling offer. In the online world, the marketer has the opportunity to target content to an audience of one person. The more you know about them, the better chance you have of serving a banner or landing page they will respond to.
  • the present method is an integrated, cross-channel software platform for optimizing and managing online advertising campaigns. It considers all of the components shown in Figure 1 simultaneously. It can optimize both media buying decisions and creative serving decisions.
  • the present method of automating marketing on digital channels includes collecting attributes of users and attributes of and users responses to medias displayed via a plurality of different digital channels for marketing stages from first contact to purchase if any on the digital channels. Attributes of user and media are grouped and ordered as a function of user responses. The media displayed to a user is modified at a marketing stage based on one of the user's attributes and response using the groups of attributes.
  • the collection of a user's attributes and responses may be assembled for each user session from first contact to purchase independent of length of time from the first contact to purchase.
  • the user session attributes may include at least one of costs, revenue, targeting, campaign, cookie characteristics, browser information, IP address, time and day, length of session and action taken.
  • the user session attributes may further include at least one of status of the user for the displayed media, frequency of use for all medias, frequency of purchases for all medias, number of purchases for all medias, and net revenue for all medias.
  • the present method of automating marketing on digital channels to met a marketing goal includes collecting attributes of users and attributes of and users responses to medias displayed via a plurality of different digital channels. Attributes of user and media are grouped and ordered as a function of user responses. Expected value and expected cost on at least one digital channel of subgroups of the ordered groups are calculated. The media displayed on the at least one digital channel is modified using the expected value and the expected cost of the subgroups groups to met the marketing goal.
  • the computing of expected value may include determining a type of occurrence to be measured, the rate to be paid for the occurrence and an expected volume for a selected occurrence.
  • the computing of expected cost may include determining the cost based on the type of media and the channel to be displayed and based on one of the frequency of display, length of the display, time of day, week or year of the display, and geography of the display.
  • the present method optimizes display of advertisements for an ad campaign by collecting responses to prior ads displayed via a plurality of different digital channels; ordering attributes that describe ad content and placement; computing the expected value of the full set of candidate ad decisions; making the optimal ad decision with respect to content and placement; and running the new ad on one or more selected channels using a favorable set of targeting parameters.
  • the method operates in a closed loop on a continual basis and thus includes continuing to collect responses to ads displayed via a plurality of different channels; re-ordering attributes that describe ad content and placement if necessary; recomputing the expected value of the full set of candidate ad decisions; determining if the optimal ad decision with respect to content and placement needs to be altered based on the revised attribute ordering; and running the revised new ad on one or more selected channels using a favorable set of targeting parameters.
  • algorithms for weighting or otherwise treating attributes may be employed based on a variety of factors.
  • the digital channels may include paid search engines, television, bill boards, the web, e-mail, mobile devices and video games.
  • Figure 1 is a block diagram of the components of an online campaign and its optimization according to the present disclosure.
  • Figure 2 is a functional diagram of an online advertising campaign and its optimization according to the present disclosure.
  • Figure 3 is a flow chart for creating an initial media plan according to the present disclosure.
  • Figure 4 is a flow chart for trafficking an initial campaign according to the present disclosure.
  • Figure 5 is a flow chart for initial placement optimization, updating and re- trafficking a campaign according to the present disclosure.
  • Figure 6 is a flow chart for landing page and acquisition page optimization according to the present disclosure.
  • Figure 7 is a flow chart for placement optimization according to the present disclosure.
  • Figure 8 is a flow chart for data analysis, archiving and reporting according to the present disclosure.
  • Figure 9 is a block diagram of an advertising campaign optimization process according to the present disclosure
  • Figure 10 is a block diagram of the components of a system which implements an online campaign and its optimization according to the present disclosure.
  • the present method is an integrated, cross-channel software platform for optimizing and managing online marketing or advertising campaigns. It considers all of the components shown in Figure 1 simultaneously. It can optimize both media buying decisions and creative or media content serving decisions.
  • the examples described are for online channels, the method can also be used to place ads on television, on radio, in newspapers or other channels/media in which the consumer response may be received via a different channel/media.
  • the present method covers the case in which a consumer may respond to a television ad via email, telephone or perhaps only later when making a future purchase.
  • the present system automatically determines the best purchasing decisions across multiple channels in order to achieve the goals of a marketing campaign. While different strategies are employed for each channel, the approach is similar:
  • Cost Predict the actual cost to be paid. For a given bid, the actual cost paid can vary significantly based on the competitive landscape. For some placements, this landscape is fairly stable and predictable. For others, it changes rapidly and must be constantly evaluated.
  • the next step is to calculate the optimal bid level.
  • the present system can automatically optimize any number of objective functions at the detailed placement level (e.g., keyword level in the case of paid search). For example, it can optimize net revenue subject to certain constraints that reflect different marketing campaign goals an advertiser may have:
  • the system also considers targeting, which adds an additional level of complexity.
  • the software automatically structures bids based on factors such as day of week, time of day, and geography.
  • the present method can group placements, automatically or by means of explicit targeting rules, so that all placements need not be treated identically. For example, bid updates for a group of high value keywords in paid search may be sent to the search engine several times per day, whereas bid updates for lesser performing keywords may only be sent every two or three days.
  • the present system analyzes each placement in real time, based on current results and trends. It also uses extensive attribute analysis to intelligently combine data from similar placements when individual data is not sufficient. This approach is then combined with the goals of a marketing campaign, from the point of view of the advertiser, to come up with a winning bid strategy.
  • the present method is designed in such a way as to allow different multivariate statistical algorithms to be employed; it is not dependent upon any particular one.
  • the present system not only supports multiple channels, it also explicitly shares learning across them. This is a key differentiator from others in the industry. Advertisers want to spend a minimum of time in learning mode so that a campaign can begin to generate maximum revenue as soon as possible. When channels are considered as separate entities and there is no shared learning (as in the prior art), the learning phase will be prolonged at the expense of revenue.
  • banners of different sizes in a display campaign can, if appropriate, be configured to share learning so as to reduce time in the learning phase and transition more quickly to the optimization phase.
  • Revcube digital channel partners are illustrated as search engines 1000, display networks 1002, e-mail partners 1004, cell phone networks 1006, video game vendors 1008, behavioral networks 1010, and other types not explicitly listed.
  • a viewer or consumer in the course of interacting with these digital media, may be presented with a paid search text ad 1012, banner impression 1014, mobile Short Message Service (SMS) 1016, or other ad format served by Revcube.
  • SMS mobile Short Message Service
  • the consumer may then click or otherwise respond, resulting in transfer to a landing page 1018, web page 1020, conversion page 1022, or other property.
  • the Revcube platform illustrated in Figure 2 consists of a layer 7 switch 1024, or similar network device, that routes and switches traffic to/from the consumer interacting with the digital media.
  • the layer 7 switch manages web cookies and other tagging schemes, and communicates with a content server 1026 where the creative content resides.
  • the content server may be owned and operated by Revcube or may be a service leased from a third party provider. All traffic through the layer 7 switch is sent to log processing 1028 which organizes and formats logs, then forwards them to the rest of the Revcube platform via the transport infrastructure 1030.
  • the creative optimization engine 1032 computes the most effective creative to serve, based on a plurality of factors, and transfers serving decisions to the layer 7 switch.
  • the Revcube platform further consists of a partner interface 1034 which connects to Revcube's digital channel partners. This interface is customized depending upon the nature of the interface and extent of automation offered by each channel partner.
  • the partner interface is connected to the transport infrastructure 1030 as well as to media allocation component 1036, which is responsible for managing all cross channel media buying in conjunction with the bid optimization engine 1038.
  • a campaign manager component 1040 oversees all aspects of proper campaign functioning. It is connected to the transport infrastructure 1030 as well as to the campaign database 1042.
  • the campaign database is populated in part by the ad campaign manager working with the campaign web interface 1044.
  • the ad campaign manager may be an employee of the advertising client or acting on their behalf.
  • the final part of the Revcube platform consists of a data mining component
  • the reporting web interface 1054 is the means by which client executives, managers, analysts or other approved parties can monitor the progress of an ongoing ad campaign.
  • FIG. 3-8 The basic Revcube process flow for campaign setup and management is shown in Figures 3-8.
  • the creation of an initial media plan is illustrated in Figure 3.
  • Campaign 50 is designed for client 52 using inputs 54. These include the product category and the incentive or offer to be advertised, campaign financial information, campaign target information and business rules.
  • Historical campaign information when available, is provided to the campaign learning function 56.
  • This historical information includes creative 57 and associated metadata description, and site targeting 60.
  • the intent is to provide a starting point for developing creative for the current campaign and determining which ad networks to target.
  • the channels illustrated in 58 include paid search, contextual search, display behavior, email and display banner, but also extend to mobile, video games and others not explicitly listed.
  • the client may explicitly request that certain keywords and sites 62 be excluded from the campaign.
  • the media planner 66 creates the initial media plan for all of the relevant channels 70. Other channels not shown may also be included in the initial media plan.
  • Each of the individual channel plans has inputs 72 which include pricing information as well as a value estimator. Biddable inventory uses a bid pricing algorithm to estimate the value of each placement. Paid search and contextual search channels further have a keyword list and associated metadata description. These comprise the initial media buying decision 10 from Figure 1.
  • Each includes initial creative and metadata for the creative.
  • the media planner will select a subset of creative by channel 68 to build into the initial media plan 64.
  • the resulting initial media plan is provided to the trafficking function 76 which is further described in Figure 4.
  • Figure 3 to channel level detail at 78 including allocation across channels.
  • a delivery plan for each channel and affiliate is produced at 80 along with a candidate set of publishers and networks.
  • the function 84 selects the publishers and networks from the candidate set to match the delivery plan, including their respective targeting capabilities, and the whole channel level media plan is encoded in 82.
  • the channel level media plan is next operated on by 86 which contains a strategic component that understands how to traffic to each publisher and channel with respect to targeting ability, available volumes and heuristics for success. This is then used to produce a channel and publisher level campaign 90. After approval of campaign trafficking to the publishers at 88, placements are transmitted to the publishers 92. This transmittal may be automatic or may involve some manual operations, depending upon the capabilities and fee structure of the publishers' interfaces. At 94 there is a confirmation that all campaigns are running and trafficked properly. Rejected campaigns by publishers are reviewed at 96 and may be corrected at 98 or else dropped, depending on the reasons for rejection. The result is a running campaign 100.
  • FIG. 5 Initial placement optimization, updating and re-trafficking of the media plan are shown in Figure 5. Elements in Figure 5 which are similar to those in Figure 4 have the same number.
  • the input to the placement optimization of Figure 5 is the optimized current campaign 170, which is one of the outputs from Figure 7 discussed below.
  • the optimized current campaign is converted to channel level details including channel level split at 78.
  • the identification of which channels are being optimized in this pass is determined at 102 which results in an updated media plan 80.
  • the translation layer and optimization by publisher and channel including targeting ability is performed at 86.
  • the optimization is designed to achieve the client's campaign goals input at 54 of Figure 3. This results in a channel and publisher level campaign at 90.
  • the provision of trafficking to publishers at 92 includes the steps of updating parameters associated with the placements at 104, such as keyword bids for paid search and creative for email. As in Figure 4, there is a confirmation that all campaigns are running and trafficked properly at 94. Rejected campaigns by publishers are reviewed at 96 and may be corrected at 98 or else dropped, depending on the reasons for rejection. The output is the campaign running 100.
  • Landing page and acquisition page optimization also known as post placement optimization, is illustrated in Figure 6.
  • a post placement opportunity arises when a consumer makes an explicit decision that leads him or her to a landing page or acquisition page served by the system. This explicit decision may involve clicking on a paid search ad, responding to an SMS, or perhaps typing in a URL directly.
  • Creative serving via hard targeting 112 may apply in cases where certain "hard" decisions due to business rules override decisions from the optimization algorithms. An example is if a consumer's cookie directs a particular landing page to be served based on past behavior, or if a client requests that all consumers see a certain landing page on Friday nights.
  • the hard targeting function includes updating customer cookies and logging creative request information.
  • the consumer action made in 110 is analyzed with respect to the various components at 114A.
  • origin ID attributes that describe the campaign, behavioral targeting and creative
  • referrer publishers, keyword (KW) information
  • user cookies user cookies
  • browser OS, IP information, date and time.
  • the decision engine determines which landing page or acquisition page to serve 122.
  • the customer may progress through various pages towards eventual acquisition at 122B. This may take several steps during which time the origin ID behavior information 118 may evolve and be used to update customer cookie at 116A and 116B.
  • the progression from 114B through to acquisition 122B is analogous to that described above from 114A to 122A.
  • the subsequent client acquisition process 136 involves steps needed to complete the acquisition and may include collecting contact information or completing a financial transaction.
  • All customer interaction events are logged at 124 and provide a session level view of the customer engagement 126.
  • This session level view allows resulting revenue to be attributed to ad clicks and impressions on the path towards acquisition, even if they happened several days in the past. This is essential for the post placement optimization algorithms to understand cause and effect.
  • the quick access data repository for ongoing optimizations 128 allows for adapting the landing pages and acquisition pages 130 based on observations of successes and failures. Updates to the Revcube serving decision engine 132 as a result of 130 are then propagated 120A and 120B
  • the session level view 126 also provides primary input for data archiving 138, offline data analysis 140, reporting 142 and the Revcube targeting optimization engine 134 which is described in more detail in Figure 7 below.
  • Revcube placement optimization is illustrated in Figure 7.
  • Reports from publishers are acquired at 150 which provide actual impressions and clicks delivered 152 and associated costs 156.
  • These data and the data from before the last update 154 are fed to the costing engine 158 which generates costing information 160.
  • This information is provided as input to the Revcube targeting optimization engine 134.
  • Also provided as an input is the additional targeting information about campaigns running at publishers not necessarily included in reports 162 and the session level view 126 from Figure 6.
  • the Revcube targeting optimization engine requires revenue information from campaign (acquisition values) from 168 and optimization objective functions 166.
  • the latter may be net profit at the most detailed placement level and may be subject to constraints on position, cost per acquisition (CPA), or other constraints.
  • the partial attribution functions 167 rely on session level data and revenue data and are a means of quantifying the sporadic nature of consumer online behavior.
  • the path to acquisition may take place over several days or weeks, involve exposure to multiple ads and multiple visits to Revcube served landing pages.
  • the partial attribution algorithm is a way to assign value to each decision that the Revcube system makes even if it did not lead immediately thereafter to a conversion. Branding effects in banners campaigns is an example where this is relevant.
  • the output of the Revcube targeting optimization engine 134 is an optimized campaign 170 which is input to the flowchart of Figure 5. Costs are assigned at the most detailed level possible (e.g., keywords in paid search) 172 and outputs are sent to data archiving 138, offline data analysis 140, and reporting 142.
  • Costs are assigned at the most detailed level possible (e.g., keywords in paid search) 172 and outputs are sent to data archiving 138, offline data analysis 140, and reporting 142.
  • Primary inputs are session level view 126, detailed costing 172 and revenue information from campaign (acquisition values) 168.
  • Manual analysis and review 186 using data analysis tools 188 such as SAS offer optimization learnings 200 and campaign learnings 204 which can be used to alert the media planner 206 or tune the optimization algorithms 202 as appropriate.
  • the present method computes a holistic view of each customer session 180 including costs, revenues, targeting, campaigns, cookie characteristics, browser information, IP addresses, time of day, the amount of time on sites and actions taken. This information periodically populates a versioned view of data 190 and a comprehensive reporting view of data for a specific campaign 192.
  • a client interface for reporting 194 queries the database via a web interface. There is also a more "advanced" interface which contains certain types of data that are not relevant for customers to view (e.g., maintenance functions).
  • the present method provides a closed loop system capable of designing an initial campaign based on prior campaign information, then updating this information in real-time to further optimize the campaign. This update may be based solely on the present campaign or on other campaigns which may not be directly related to the present campaign.
  • a summary of the process is shown in the flowchart of Figure 9.
  • Responses to prior ads are collected 1202 from digital channels 1200. Attributes are ordered in 1204 based on the specified goals of the campaign (often financial, but may be non- financial).
  • the expected values of all candidate ad decisions are computed at 1206.
  • the optimal decision for ad content and placement is made at 1208. Finally, the new ad is run at 1210.
  • the method illustrated in Figure 9 also handles the transition between learning and optimizing in a novel manner.
  • the prior art uses past experience or simple confidence calculations to decide when to stop learning.
  • the present method automatically calculates the trade off between making the best decision using accumulated information to date, versus continuing to learn in order to make an even better decision in the future. It also focuses learning in the areas where it will do the most good:
  • the present automated method uses Business Intelligence/machine learning.
  • the Switch is a device or software that redirects user between pages.
  • a user clicking on an ad placed by the present system will navigate through The Switch and be redirected to an appropriate landing page. Also, the user may click through The Switch between landing pages, for example, with multiple page forms, and if they purchase something, their final conversion page may contain a hidden link to a switch URL.
  • the Switch for each redirect, writes and updates the cookie, determines the binary decision cube, and determines from the current decision cube whether to randomly choose a valid destination or redirect to the most historically effective destination.
  • All of the input variables for The Switch are URL variables and at a minimum that includes P for promotion and D for destination.
  • the promotion variable P describes a promotion that the user is requesting.
  • the destination variables D are destination groups of, for example, pages, banners, emails or other creative content that are functionally interchangeable.
  • log processing the optimization process
  • the Switch make the decisions of which page and how many pages to display base on the user's profile.
  • the promotion and the destination variables are defined in The Man which creates promotional-specific promo-info XML files, which are used to create hypercubes.
  • the Shaman produces the hypercubes.
  • the URL variables may also include conversion value and conversion type.
  • the conversion variables are only appropriate in the final destination group signifying an actual conversion.
  • the conversion values indicate that an event has taken place that is expected to result in revenue.
  • Information also includes origin ID and user attributes.
  • the user attributes may be stored in the user cookie file.
  • Well-known user attributes includes gender, age, location, income, time and date, date of week, etc.
  • the Switch determines destination based on one or more of the following attributes: user, promotion specific to user and placement.
  • Hypercubes hold the promotion specific decision information on The Switch.
  • Hypercubes There is exactly one hypercube for each promotion ID-destination group on The Switch. Optimization information in the hypercube is generated if one or more attributes in the destination groups have received enough traffic and enough conversions for the data to converge and determine that a specific attribute (or group of attributes) will produce a higher expected value than others for a particular end user at a particular time. Hypercubes also contain promotion destination and destination specific information from The Man, such as promotion name or the name of the destination group. Hypercubes contain all of the destination information for determining which creative element within a destination group can be returned to a user. In addition to the interaction with the end user, The Switch produces log files for The Woodsman and receives the stateful decision files from The Shaman.
  • the Woodsman receives the log files from The Switch via The Spool.
  • Woodsman splits the switch files into promotion specific files and provides them to The Tallyman. Although the format of the incoming and outgoing files of The Woodsman is the same, they contain different information because data is split into promotion- specific folders.
  • a single log file from The Switch may contain, for example, three requests for promotions one, two and five. This will result in the creation of three new folders in the outgoing directory named One, Two and Five. Within each folder a data log would be created containing all the information representing a single request.
  • the files from The Woodsman include the specific promotion, its user attributes and its page attributes.
  • the Tallyman pre-processes the promotion- specific files from The Woodsman and distributes the results to The Shaman and to The Man database.
  • the Tallyman reads the promotion specific log entries from The Woodsman and based on the information within the log file and from the promo information, outputs information to the appropriate output file.
  • Each log entry gets converted into a request object, which has all the information needed for each data file that The Tallyman outputs. Additionally
  • the Tallyman decrypts all the hex data within the log entry.
  • the data from the log entries get broken into user and placement attributes levels and other data such as click stream.
  • the Tallyman provides Plog files to The Shaman and provides costs, revenue and dust files to be stored by The Man.
  • the Plog file contains one entry for each switch request, one entry for each conversion and one entry for each destination group in the click stream for each conversion. If The Tallyman were, for example, to process logs where the user listed two destinations groups within a conversion, another visitor at one destination without a conversion and a third visited three destination groups in the conversion, there would be a total of ten entries.
  • the Plog in this example would contain six entries for a request - one entry for conversion and three entries for proactive conversion.
  • the Cost data file is used for exporting data into the database.
  • the Cost data file exports into the database of The Man either directly or through The Middleman.
  • the data delineate each distinct cost event. In CPC advertising campaign, this is typically a click and, in CPM campaign step, an ad impression (display of an ad, such as a banner).
  • the Dust file is for importation into SAS or similar data analyzing tool for human review of raw data.
  • the analysis of these files provides a degree of offline, human review of many of the decision that The Shaman makes.
  • the analysis may also be performed automatically with an output for human review and the Dust file eliminated.
  • the Shaman receives the Plogs from The Tallyman, analyzes and optimizes and provides the decision hypercubes to The Switch and TLC files to The Man database/web server.
  • the Shaman optimizes which content the user sees or to navigate through a promotion based on which content is expected result on the highest profit.
  • the Shaman also optimizes advertising placement by determining which are the most profitable.
  • the Shaman utilizes user click history as well as associated pages, user and click attributes. It reads this information from the Plog files from The Tallyman, and analyzes them and outputs multiple dimensional decision-matrices that are used to determine the optimum content of a particular set of attributes. For example, if a particular set of landing pages exhibit different behavior or different type of users, this learned information will be applied to decisions made for serving content. If a particular landing page results in more profit when viewed by Macintosh users on Thursday, this information will be encapsulated in the decision cube file BDC provided to The Switch and as a result The Switch will behave differently in selecting sights and transfers.
  • the Shaman also creates a matrix of expected values that is useful in deciding which advertising results in the highest profits. These expected revenue values will be compared to costs per placement to determine what the actual profit per placement is. These PLC files are provided to The Man directly or through The Middleman.
  • the Middleman receives costs, revenue and PLC files and injects them into a media allocation database.
  • costs, revenue, and expected values are available for all placements of a single database.
  • Applications that interact with the media allocation database have access to this data as soon as it is injected.
  • the Man database/web server includes The Man web interface, The Man database, and the media allocation database.
  • the Man web interface is the web front end to the databases.
  • the Man web interface is for media planner and trafficking, utility and a link to reporting the trafficking portion of The Man.
  • the Man web interface allows the creation of creative, targeting and in case of searches keywords in order to automatically create and submit ads to advertising partners.
  • FIG. 10 a system having main components which may be individual processors or individual programs is capable of performing the method of optimization of the present disclosure. Other allocations of resources and performances may be used.
  • Figure 10 is just one example of an implementation.
  • the present method of automating marketing on digital channels includes collecting attributes of users and attributes of and users responses to medias displayed via a plurality of different digital channels for marketing stages from first contact to purchase if any on the digital channels. Attributes of user and media are grouped and ordered as a function of user responses. The media displayed to a user is modified at a marketing stage based on one of the user' s attributes and response using the groups of attributes.
  • the collection of a user's attributes and responses may be assembled for each user session from first contact to purchase independent of length of time from the first contact to purchase.
  • the user session attributes may include at least one of costs, revenue, targeting, campaign, cookie characteristics, browser information, IP address, time and day, length of session and action taken.
  • the user session attributes may further include at least one of status of the user for the displayed media, frequency of use for all medias, frequency of purchases for all medias, number of purchases for all medias, and net revenue for all medias.
  • the present method of automating marketing on digital channels to met a marketing goal further includes calculating expected value and expected cost on at least one digital channel of subgroups of the ordered groups.
  • the media displayed on the at least one digital channel is modified using the expected value and the expected cost of the subgroups groups to met the marketing goal.
  • the computing of expected value may include determining a type of occurrence to be measured, the rate to be paid for the occurrence and an expected volume for a selected occurrence.
  • the computing of expected cost may include determining the cost based on the type of media and the channel to be displayed and based on one of the frequency of display, length of the display, time of day, week or year of the display, and geography of the display.

Landscapes

  • Business, Economics & Management (AREA)
  • Engineering & Computer Science (AREA)
  • Accounting & Taxation (AREA)
  • Development Economics (AREA)
  • Strategic Management (AREA)
  • Finance (AREA)
  • Game Theory and Decision Science (AREA)
  • Entrepreneurship & Innovation (AREA)
  • Economics (AREA)
  • Marketing (AREA)
  • Physics & Mathematics (AREA)
  • General Business, Economics & Management (AREA)
  • General Physics & Mathematics (AREA)
  • Theoretical Computer Science (AREA)
  • Management, Administration, Business Operations System, And Electronic Commerce (AREA)

Abstract

Un procédé d'automatisation de vente sur des chaînes numériques comporte la collecte d'attributs des utilisateurs et d'attributs des réponses des utilisateurs aux contenus multimédia diffusés par le biais d'une pluralité de différentes chaînes numériques pour des phases de vente depuis un premier contact jusqu'à un éventuel achat sur les chaînes numériques. Des attributs d'un utilisateur et des contenus multimédia sont regroupés et classés en fonction des réponses de l'utilisateur. Le contenu multimédia diffusé à un utilisateur est modifié lors d'une phase de vente sur la base d'un de ses attributs et de sa réponse au moyen des groupes d'attributs. Les contenus multimédia diffusés peuvent être modifiés sur la ou les chaînes numériques en utilisant la valeur attendue et les coûts attendus de groupes de sous-groupes des attributs regroupés afin de satisfaire l'objectif de vente.
PCT/US2007/086784 2006-12-07 2007-12-07 Procédé d'automatisation de vente sur des chaînes numériques WO2008070834A2 (fr)

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
US86894906P 2006-12-07 2006-12-07
US60/868,949 2006-12-07

Publications (3)

Publication Number Publication Date
WO2008070834A2 true WO2008070834A2 (fr) 2008-06-12
WO2008070834A3 WO2008070834A3 (fr) 2008-07-31
WO2008070834A9 WO2008070834A9 (fr) 2009-07-09

Family

ID=39493097

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/US2007/086784 WO2008070834A2 (fr) 2006-12-07 2007-12-07 Procédé d'automatisation de vente sur des chaînes numériques

Country Status (1)

Country Link
WO (1) WO2008070834A2 (fr)

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US11295345B1 (en) * 2019-03-28 2022-04-05 Airbnb, Inc. Cross-channel personalized marketing optimization

Citations (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20050125355A1 (en) * 2003-11-19 2005-06-09 Gene Fein Digital media inventory control, distribution and destruction system
US20050262024A1 (en) * 2001-03-30 2005-11-24 Intertainer, Inc. Digital entertainment service platform

Patent Citations (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20050262024A1 (en) * 2001-03-30 2005-11-24 Intertainer, Inc. Digital entertainment service platform
US20050125355A1 (en) * 2003-11-19 2005-06-09 Gene Fein Digital media inventory control, distribution and destruction system

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US11295345B1 (en) * 2019-03-28 2022-04-05 Airbnb, Inc. Cross-channel personalized marketing optimization
US12086835B2 (en) 2019-03-28 2024-09-10 Airbnb, Inc. Cross-channel personalized marketing optimization

Also Published As

Publication number Publication date
WO2008070834A9 (fr) 2009-07-09
WO2008070834A3 (fr) 2008-07-31

Similar Documents

Publication Publication Date Title
US8554602B1 (en) System and method for behavioral segment optimization based on data exchange
US20170364953A1 (en) Offer management system and methods for targeted marketing offer delivery system
US11216850B2 (en) Predictive platform for determining incremental lift
George et al. Modeling the consumer journey for membership services
US20080201188A1 (en) Niche-oriented advertising networks platform and methods of operating same
US20120046996A1 (en) Unified data management platform
US20150095166A1 (en) System, method and computer program for providing qualitative ad bidding
US20060026062A1 (en) System and method for optimizing advertising marketplace operations
US20080052278A1 (en) System and method for modeling value of an on-line advertisement campaign
US20110040636A1 (en) Learning system for the use of competing valuation models for real-time advertisement bidding
US20120158456A1 (en) Forecasting Ad Traffic Based on Business Metrics in Performance-based Display Advertising
US20060173744A1 (en) Method and apparatus for generating, optimizing, and managing granular advertising campaigns
US20020169654A1 (en) Method and system of determining differential promotion allocations
US20030115099A1 (en) Method of automated online media planning and buying
US20160210656A1 (en) System for marketing touchpoint attribution bias correction
US8311884B2 (en) System for allocating advertising inventory in a unified marketplace
Malthouse et al. A Multistakeholder Recommender Systems Algorithm for Allocating Sponsored Recommendations.
US20210382952A1 (en) Web content organization and presentation techniques
Donahue et al. Artificial intelligence in cloud marketing
US8805715B1 (en) Method for improving the performance of messages including internet splash pages
Burke et al. Rethinking marketing research in the digital world
WO2001048666A1 (fr) Systeme, procede et modele d'operation commerciale optimisant les performances de messages publicitaires ou de messages dans des media mesurables interactifs
Adikari et al. Adaptive Ad Network Selection for Publisher‐Return Optimization in Mobile‐App Advertising
US20220005063A1 (en) System and Methods for Delivering Targeted Marketing Offers to Consumers via Mobile Application and Online Portals
WO2008070834A2 (fr) Procédé d'automatisation de vente sur des chaînes numériques

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 07865383

Country of ref document: EP

Kind code of ref document: A2

NENP Non-entry into the national phase in:

Ref country code: DE

122 Ep: pct application non-entry in european phase

Ref document number: 07865383

Country of ref document: EP

Kind code of ref document: A2