JP5080630B2 - Yield management system, method and program - Google Patents

Description

  The present invention relates to an online advertisement distribution technique, and more particularly to a yield management system, method and program for improving advertisement profits in a publisher or the like.

  With regard to the advertising business of publishers, each publisher places an advertisement on each side of the Internet site. These advertisements are created on the publisher's menu and sold to advertisers through media representatives and advertising agencies. In addition, the advertising space may be left unsold to the advertiser. In that case, there are a publisher that publishes its own advertisement and a publisher that sells surplus advertising space to an ad network. In addition, with regard to the advertising space of a small-sized media company or the advertising space that is difficult to sell by itself, all sales may be entrusted to an ad network. When ad network sells surplus ad inventory (frame) or ad inventory that is not sold in-house, the business model between ad network and publisher is (1) advertising by fixed price CPM, CPC, CPA It is roughly divided into a “purchase model” in which the inventory is purchased by the ad network and (2) a “revenue share model” that receives a certain percentage of the amount sold by the ad network. In any business model, the purchase unit price (CPM, CPC, CPA) and the percentage of revenue share are determined by contract between the publisher and the ad network.

  In recent years, the ad network market in Japan has been activated, and there are many ad network operators. On the other hand, in the United States, in addition to ad network operators, the demand side platform (DSP), which is an ad network for advertisers operated by agencies, ad exchanges that cross-purchase inventory of ad networks and DSPs, and publisher revenue Yield optimization for maximizing, data exchange for collecting behavioral data, etc. are appearing, and the market is increasingly expanding. It is estimated that the ad network market will be further activated and developed in Japan.

JP-T-2004-524593

  When a publisher provides advertisement inventory to an ad network, there are two main problems.

  The first issue is that it is difficult to predict which ad network will provide the most profits. For example, if it is only an ad network that purchases at CPM, it is obvious to which ad network the inventory should be provided for each impression to the consumer's site. However, if an ad network that purchases with CPC or an ad network that performs revenue sharing is added, it becomes difficult to determine which ad network is best to provide inventory. That is, for the CPC ad network, it is necessary to predict the rate of clicks. On the other hand, with respect to the revenue share ad network, the advertising unit price sold by the ad network varies, and it is difficult to predict which advertisement will be displayed. This is because the advertisement distribution server on the ad network side controls which advertisement is displayed.

  The second problem is that the operation load for confirming whether or not the advertisement provided from the ad network can be posted is high. The types of advertisements sold by each ad network (the number of advertisers and deals) are comparable to those of major media, but the operational burden is high for publishers to check whether these ads are available. By signing up with a large number of ad networks, the potential for revenue increases, but considering the workload required to determine whether or not to publish, publishers do not sign up with so many ad networks. is there.

  Yield management systems for optimizing inventory delivery to ad networks have been developed mainly in the United States. However, the conventional yield management system corresponds to the purchase CPM for which the profit is determined for each impression and the bid for each CPM, and does not correspond to the situation where the profit is not determined for each impression. In addition, it cannot respond to the exact situation of whether or not publication is possible as in Japan.

  The present invention has been made in view of the above circumstances, and it is an object of the present invention to provide a yield management system, method, and program capable of reducing operational burden while improving profits by utilizing many ad networks. And

  The yield management system according to the embodiment obtains advertisement content identification information and a log of the number of impressions and clicks of advertisement space from a publisher's WEB site, and counts the number of impressions and clicks per unit time by counting the logs. Calculate the number. The publisher calculates an eCPM (Effective Cost Per Mille) for each campaign for a list of ad networks that can provide advertising inventory, calculates a predicted eCPM for each ad network from the eCPM of each campaign, and based on the predicted eCPM Select an ad network and output the corresponding ad tag.

The figure which shows the online advertisement delivery system of embodiment The block diagram which shows the yield management system of the embodiment The figure which shows the JS log when there is CDNURL The figure which shows the JS log when there is no CDNURL Diagram showing delivery log The figure which shows JS log totaling every hour when there is CDNURL and the case is unspecified The figure which shows the creative hash and creative master table for data aggregation The figure which shows the JS log compilation every hour of the state where the matter is specified The figure which shows the JS log compilation every hour when there is no CDNURL Figure showing distribution log totaling every hour Figure showing integrated hourly aggregated data Diagram showing campaign availability data Diagram showing distribution contract data Diagram showing placement master Diagram showing performance data for each advertiser Diagram showing performance data for each creative Diagram showing advertiser master Diagram showing Ad Network Campaign Master Diagram showing ad network campaign creative master Flow chart showing the overall flow of optimization processing Flow chart showing predicted eCPM calculation processing Flow chart showing predicted eCPM calculation processing in the case of revenue share Flow chart showing ad network priority calculation processing Figure showing an example of the posting availability screen

  Embodiments described herein relate generally to an online advertisement distribution technique using an ad network, and more particularly, to a yield management system, method, and program for improving advertisement profits at a media company or the like. Prior to the description of a specific embodiment, the contents of a contract between a publisher and an ad network operator, and the advertisement menu pattern in the ad network will be described.

Table 1 shows the contents of the contract between the publisher and the ad network provider.

  When the business model is Revenue Share (RS), a certain percentage (%) of the amount sold by the ad network to the advertiser is paid to the publisher. When the business model is purchase (buy-out), the total fixed unit price of CPM, CPC, and CPA set by the ad network for each site or each advertising space is paid to the publisher. There are two types of house advertisements: paid and free. This means whether there is no advertisement issued by the ad network when the ad network and the media company have a purchase contract, and whether or not there is a charge when the company's own advertisement is issued. Of course, the revenue share is free.

  The amount of inventory provided by publishers to the ad network is also included in the contract. There are cases where there is no limit on the amount of inventory provided to the ad network, and cases where an upper limit or a lower limit of the number of impressions is imposed on the amount of inventory provided to the ad network. Unrestricted means that there is no limit on how much inventory the publisher will provide to the ad network. In revenue share, there is no limit in many cases, but in purchase, in most cases, an upper limit or a lower limit is set for the number of impressions. These upper and lower limits are set for each site or advertising space as to how much inventory the publisher provides to the ad network. There are cases where both an upper limit and a lower limit are set, and only one of the upper limit and the lower limit is set.

Table 2 shows advertisement menu patterns in the ad network.

  The unit price of the advertisement menu includes CPM (Cost Per Mille), CPC (Cost Per Click), and CPA (Cost Per Action). In CPM, the amount of money is determined according to the distribution per 1000 imp (impression). In CPC, the amount is determined according to the click. In CPA, the amount of money is determined according to an action such as purchase or document request.

  As a guarantee form of the advertisement menu, there are a case of “billing” and a case of “guarantee”. In the case of “billing”, the amount paid by the advertiser is obtained by multiplying the CPM unit price, the CPC unit price, or the CPA unit price by the actual amount. In the case of “guaranteed”, the amount is guaranteed such that 10 million impressions are 1 million yen in the advertisement menu. In this case, the ad network generally distributes a certain amount larger than the guaranteed amount, and the excess is not charged to the advertiser. From the viewpoint of the publisher, the excess is distributed free of charge.

  Hereinafter, embodiments of the present invention will be described with reference to the drawings.

  FIG. 1 is a diagram illustrating an online advertisement distribution system according to an embodiment. As shown in the figure, this system has a yield management system 1, an advertisement management system 3, an advertisement placement status investigation system 4, a media company ad server 5, a consumer terminal 6, an ad network 7, and a CDN (Contents Delivery Network) 8. These are connected to a physical network (for example, the Internet).

  The yield management system 1 aggregates impressions and clicks for each advertisement (creative) when inventory is provided to the ad network, determines an appropriate ad network from among the plurality of ad networks 7, and uses the ad tag as the consumer terminal 6. Deliver to the browser. The advertising space of the site displayed in the browser is managed by the publisher ad server 5, and the advertising content displayed in the advertising space of the site is provided from the CDN 8. A user terminal 2 is connected to the yield management system 1, and through the management screen, it is possible to manage whether or not to publish, manage an ad network for depositing inventory and advertising space, and report delivery results.

  The yield management system 1 cooperates with the advertisement management system 3 through the data cooperation function. The advertisement management system 3 stores various masters such as a product category master and an advertiser master, and publication performance data provided from the advertisement placement situation survey system 4, and these are stored in the yield management system 1 through a data linkage function. Send. The information provided from the advertisement management system 3 is used for ad network optimization processing and automatic processing for determining whether or not to publish. A basic campaign information file, a basic creative information file, a creative material file, and the like are uploaded from the ad network 7 to the yield management system 1, and these are also used for an optimization process and an automatic process for determining whether to post.

  The characteristics of the yield management system 1 are (1) optimization not only considering the CPM purchase ad network but also the CPC / CPA purchase ad network and revenue share ad network, and (2) posting on the ad network. This is to reduce the operational load for determining availability. A specific configuration example of such a yield management system 1 will be described.

  FIG. 2 is a block diagram illustrating the yield management system of the embodiment. The system 1 includes an optimization server 10, a data collection server 11, and a batch server 12.

  In general, when a publisher provides inventory to an ad network, the publisher cannot determine which ads are actually posted and clicked on, but in the yield management system 1, the posted ads The number of impressions, the number of clicks, and, if possible, information for identifying the creative (for example, creative ID) can be measured. In order to perform such measurement, a script (for example, JavaScript (registered trademark): JS) is used. The script is inserted in advance into the WEB site of the target media company (hereinafter referred to as “media company site” or “site”).

  The data collection server 11 acquires an inner HTML file of a publisher site by communicating with the script. By analyzing innerHTML, there is a possibility that an advertisement actually delivered by the ad network can be identified. Also, a record of events measured by the script is collected as the JS log 111. The JS log 111 is used for counting the number of impressions and the number of clicks. The JS log 111 is transferred to the data collection server 11 when the page being browsed at the time of advertisement distribution is closed or moved to another page.

  The batch server 12 acquires the JS log 111 from the data collection server 11 and aggregates the data to predict eCPM (effective Cost Per Mill, effective CPM). eCPM represents revenue per 1000 impressions.

  In the data aggregation, the JS log 111 acquired from the data collection server 11 is analyzed, and the number of impressions and the number of clicks are aggregated for each advertising space, for each campaign, and for each site.

  The optimization server 10 calculates the eCPM for each current ad network based on the data aggregated in this way and the unit price for each item of each ad network acquired at the time of publication availability, Use for. The optimization server 10 compares the eCPM of each ad network each time a consumer user displays a page of a publisher site, selects an ad network having the highest eCPM, and selects an ad tag of the ad network. To deliver.

  On the other hand, in the data linkage, the data of the advertisement placement situation survey system 4 is linked daily through the advertisement management system 3. Based on the linked data, we will automate the determination of availability. When determining whether or not to publish at a media company, etc., the advertisement actually posted by almost all media companies in Japan is compared with the advertisement acquired from each ad network at the time of application for publishing. It is determined whether an advertisement to be published exists in the past performance of a certain publisher or in the past performance of a standard publisher in Japan. If there is a past record, the posting is automatically set to be possible, and only the items that do not have a past record on the posting availability screen are determined artificially. As a result, the operational load of determining whether to publish is reduced.

  Hereinafter, specific configurations and processing procedures will be described for each function of data collection, data aggregation, and data linkage.

  The optimization server 10 includes an optimization module 101, a KVS (Key Value Storage) GATEWAY 103, and a KVS 104. The optimization module 101 delivers an ad network ad tag currently selected to the browser 6 when the consumer site is loaded into the browser 6. Specific processing of ad network optimization by the optimization module 101 will be described later. The optimization module 101 also generates and stores a distribution log 102 representing information about the distributed ad network.

  When the publisher's site is displayed in the browser 6 and an advertisement is displayed in the advertisement space of the site, the above-described script is executed in the browser 6. The script communicates with the data collection server 11 using, for example, the HTTP protocol. The data collection server 11 includes a data collection module 110 and generates and stores a JS log 111 representing information acquired by communication with a script of the browser 6. The batch server 12 periodically acquires the JS log 111 from the data collection server 11 and analyzes it.

  The batch server 12 includes a data totaling function unit 120, a data cooperation function unit 121, and a database 122. The data totaling function unit 120 includes an analysis module 123, a first batch group 126 for data collection, a second batch group 128 for data collection, a storage unit 129 for KVS upload data, and a KVS data upload unit 130. It comprises.

  The analysis module 123 acquires and analyzes the JS log 111 from the data collection server 11 and outputs the JS log 124 with CDN URL or the JS log 125 without CDN URL. The CDN URL is the URL of the CDN 8 that provides the advertisement content. When the CDN URL is found from the JS log 111, the advertisement can be acquired from the publisher site, and the campaign (project) can be identified. is there. When the CDN URL is not found from the JS log 111, the ad network can be identified, but the campaign or the advertisement creative cannot be identified.

  FIG. 3 shows an example of the JS log 124 when there is a CDN URL. The JS log 124 includes a date / time ID, a placement ID, a business ID, a click presence / absence, a CDN URL, and a click URL. The date / time ID represents the date / time when the advertisement is displayed. The provider ID is an ID of the provider of the ad network and can identify the ad network. The presence / absence of click represents whether or not the advertisement is clicked. Note that the above-described script preferably pretends that a click has been made by detecting a mouse-over event instead of an actual click. On the other hand, FIG. 4 shows the JS log 125 when the CDN URL is absent. As can be seen from the figure, when there is no CDN URL, the CDN URL and the click URL are not included.

  In parallel with the data collection based on the communication with the script, the data totaling function unit 120 acquires the distribution log 102 from the optimization server 10. As shown in FIG. 5, the distribution log has a date / time ID, a placement ID, and an operator ID. Such a distribution log 102 can accurately grasp to which ad network the distribution is made. On the other hand, the above script may not be able to accurately measure the number of impressions / clicks for each ad network, and the distribution log 102 from the optimization server 10 is used to correct this. The distribution log 102 from the optimization server 10 is preferably a distribution log 132 excluding distribution to a crawler such as a search engine. The above correction is applied to both the JS log 124 when the CDN URL is present and the JS log 125 when the CDN URL is absent.

  The first batch group 126 for data collection generates hourly aggregated data 127 based on the JS log 124 when there is a CDN URL and the JS log 125 when there is no CDN URL. The unit time is not limited to 1 hour.

  A description will be given of JS log tabulation when a CDN URL exists. FIG. 6 shows an hourly JS log total in the case where there is a CDN URL and the case is not specified. This log total includes a processing date / time ID, a collection date / time ID, an operator ID, a publisher ID, It has a site ID, a size ID, a placement ID, a CDN URL, an impression number / h, and a click number / h. A case and an advertisement creative are specified based on the CDN URL in the JS log aggregation. For this purpose, reference is made to the data aggregation creative hash and the creative master table shown in FIGS. 7A has a CDN URL, a creative hash value, and a material file storage path, and the creative master table in FIG. 7B has a creative ID, a campaign ID, and a CDN URL. The creative master table is uploaded in advance from the ad network 7 to the yield management system 1 and stored in the database 122.

  As shown in FIG. 8, the JS log aggregation for each hour after the case and the advertisement creative are specified in the case where there is a CDN URL is processing date / time ID, collection date / time ID, business operator ID, publisher ID, site ID, size ID, placement ID, campaign ID, creative ID, number of impressions / h, and number of clicks / h. As a result of referring to the data aggregation creative hash and the creative master table based on the CDN URL, the campaign (project) ID and the creative ID are specified.

  FIG. 9 is a diagram showing hourly JS log tabulation when there is no CDN URL. Processing date / time ID, collection date / time ID, company ID, publisher ID, site ID, size ID, placement ID, With impressions / h and clicks / h. Since there is no CDN URL, the project and the advertisement creative cannot be specified, and the campaign ID and the creative ID are not included.

  FIG. 10 is a diagram illustrating distribution log aggregation for each hour, and includes a date / time ID, a business ID, a publisher ID, a site ID, a size ID, a placement ID, and the number of impressions / h. The first batch group 126 of data collection integrates the hourly JS log aggregation when the CDN URL is present and the hourly JS log aggregation when the CDN URL is absent. By performing the above-described correction by the distribution log aggregation, integrated hourly aggregated data 127 is generated. As shown in FIG. 11, the total data 127 includes processing date / time ID, collection date / time ID, business operator ID, publisher ID, site ID, size ID, placement ID, campaign ID, creative ID, number of impressions / h, and Has clicks / h. If no CDN URL exists, the campaign ID and the creative ID are unknown because the case and the creative cannot be specified. In this case, in the calculation process of the predicted eCPM described later, calculation for each campaign (case) cannot be performed, but eCPM calculation itself can be performed.

  The second batch group 128 of data collection stores the aggregated data 127 for each hour in the storage unit 129 for KVS upload data. The KVS data upload unit 130 reads the data for KVS upload from the storage unit 129 and uploads the data to the KVS GATEWAY 103 of the optimization server 10. On the other hand, the second batch group 128 for data collection stores hourly aggregated data 127 in the database 122 for report aggregation processing and the like.

  The database 122 stores publication availability data, ad network campaign master, distribution contract data, and placement master for each campaign as contract information between the publisher and the ad network provider. As shown in FIG. 12, the publication availability data for each campaign includes a site ID, a campaign ID, and publication availability judgment data. The ad network campaign master will be described later. As shown in FIG. 13, the distribution contract data includes a distribution contract ID, a media company ID, an operator ID, a contract type (purchase or RS), a contract target media company ID, a contract target site ID, and a contract target placement ID. . As shown in FIG. 14, the placement master has a placement ID, an affiliated media company ID, an affiliated site ID, and a size ID.

  Further, by the data cooperation function, the data cooperation batch group 131 of the data cooperation function unit 121 acquires the various masters 30 and the publication results 31 from the advertisement management system 3 and stores them in the database 122. The data of the publication performance 31 includes the publication performance data for each advertiser as shown in FIG. 15 identified by the site URL and the advertiser ID, and for each creative as shown in FIG. 16 identified by the site URL and the creative hash. There is publication result data. The various masters 30 include an advertiser master as shown in FIG. 17 having an advertiser ID and an advertiser name.

  The information uploaded from the operator of the ad network 7 to the yield management system 1 includes an ad network campaign master, an ad network campaign creative master, and a creative material file. Such information is also stored in the database 122. As shown in FIG. 18, the ad network campaign master has information such as campaign ID, business operator ID, contract type (CPM or CPC or CPA), contract unit price, campaign period FROM, and campaign period TO. As shown in FIG. 19, the ad network campaign creative master has a creative ID, a CDN URL, and a belonging campaign ID.

  Hereinafter, each of the ad network optimization process and the publication availability process will be described in detail with reference to flowcharts. The ad network optimizing process is mainly realized by the optimizing module 101 of the optimizing server 10, and the publication permission / inhibition process is realized by the management screen 13.

  FIG. 20 is a flowchart showing the overall flow of the optimization process. In S101, first, processing based on the FQ (frequency) of the ad network is performed. For example, when the FQ of the user in the past two hours for each ad network exceeds a predetermined threshold, the eCPM value of the ad network is set to zero. Next, as shown in S102, the predicted eCPM is calculated for each ad network. Specific contents of this processing will be described later. Next, as shown in S103, it is determined whether or not there is an ad network to be prioritized. If there are a plurality of priority ad networks, the priority is calculated in S104, and the ad tag of the ad network with the highest priority is distributed. If there is only one ad network to be prioritized in S103, the ad tag of the ad network is distributed. On the other hand, if there is no ad network to be prioritized, it is determined whether or not there is an ad network that eCPM presumes. If there is no ad network estimated by eCPM, the ad network having the highest predicted eCMP calculated in S102 is set as the optimum ad network, and the ad tag is distributed. On the other hand, if there is an ad network whose eCPM is estimated, the learning rate X% is randomly distributed so that an ad network that is not distributed at all due to a low eCPM does not appear (S109, S110). Even when the FQ flag of the ad network is established, it is distributed to a random ad network.

  FIG. 21 is a flowchart showing a predicted eCPM calculation process. As shown in S201, a list (NW) of ad networks that can be distributed to the requested advertising space is acquired. As shown in S202, the following processing is performed for each case for each acquired ad network.

  First, as shown in S203, it is determined whether or not the purchase is for retargeting. If the Revenue Share ad network is set to purchase by retargeting and this request (user) has the cookie, the eCPM of the retargeting item is calculated based on the cookie, The value of eCPM is used. Repurchase purchases generally result in high transaction unit prices.

  Next, as shown in S204, it is determined whether or not the purchase is made. When it is not purchase, as shown in S205, eCPM calculation processing in the case of revenue share (RS) is performed. Processing in this case will be described later.

  In the case of purchase, eCPM is calculated based on the purchase price. First, as shown in S206, it is determined whether it is CPM or not. When it is not CPM, that is, when it is CPC or CPA, conversion to purchase CPM is performed by performing the processing of S208 to S211. First, in the case of CPC, as shown in S208, a purchase CTR (Click Through Rate) based on the results is obtained, and as shown in S209, purchase CPM = purchase CPC × purchase CTR × 1000. If it is neither CPM nor CPC, as shown in S210, CTR and CVR (Conversion Rate) of the requested ad space in the ad network are acquired, and purchase CPM = purchase as shown in S211. CPA × CVR × CTR × 1000.

  Next, as shown in S212, it is determined whether the company advertisement is also charged. If you have a purchase contract and do not place an advertisement on the ad network side and place a house advertisement, the house advertisement is usually paid, but depending on the ad network, such house advertisement is free. Are performed in S214 and S215.

  First, when the company advertisement is paid, as shown in step S213, the purchase CPM of the requested advertising space is set as the predicted eCPM. The price of the purchased CPM is calculated from the measured number of impressions and the CPM unit price. On the other hand, if the house advertisement is free, as shown in S214, a manual entry ratio of the requested advertisement space for the ad network is acquired, and as shown in S215, predicted eCPM = purchase CPM. X (1-in-house coverage).

  Finally, as shown in S216, the predicted eCPM of the ad network is obtained based on the eCPM of each case. For example, the harmonic average of eCPM of each case is set as the predicted eCPM. The eCPM of each case may vary greatly, and it is possible to obtain the predicted eCPM of the ad network in consideration of this.

  FIG. 22 is a flowchart showing a predicted eCPM calculation process in the case of revenue share. As shown in S <b> 301, a list of cases that can be distributed in the current frame (placement) is acquired. The following processes from S302 to S312 are repeated for each case. Further, as shown in S302, the following processing is executed based on CPX in the publisher company net. CPX represents one of CPM, CPC, and CPA.

  First, as shown in S303, it is determined whether the item is a guarantee item. When the case is a guarantee case such as an impression guarantee, for example, 1.1 million impressions are actually distributed with a guarantee of 1 million impressions. Since the over amount in that case is not paid to the publisher, the amount is subtracted in advance. In other words, if it is a guaranteed case, a process of obtaining an average over rate for each ad network is performed as shown in S304, and a process of CPX = CPX / average over rate is performed as shown in S305.

  Next, as shown in S306, it is determined whether or not it is CPM. In the case of CPM, eCPM = CPM is set as shown in S307. On the other hand, when it is not CPM, that is, when it is CPC or CPA, eCPM is calculated by the processing of S309 to S312. First, if it is determined in step S308 that CPC is performed, a process for calculating CTR for each case is performed as shown in step S309, and eCPM = CPC × CTR × 1000 is set as shown in step S310. When it is determined in S308 that it is not CPC, that is, in the case of CPA, as shown in S311, the manually input CTR / CVR in the ad network of the requested advertising space is acquired, and as shown in S312, eCPM = CPA × CVR × CTR × 1000.

  Next, as shown in S313, the following processing is performed based on the calculated eCPM of each case. First, as shown in S314, it is determined whether there is a past record. When there is no past performance, as shown in S315, the predicted eCPM = (harmonic average of each eCPM) × (1—the in-house coverage ratio of the requested ad space ad network). When there is a past record, as shown in S316, a process of calculating the appearance probability based on the past record is performed. Then, as shown in S317, predicted eCPM = Σ (each eCPM × each occurrence probability).

  With the above processing, the eCPM in the case of revenue share (RS) can be calculated.

  FIG. 23 is a flowchart showing the priority calculation processing of the ad network. As shown in S401, the following processes S401 to S405 are performed on the ad network in which the upper limit and the lower limit of the number of impressions (imp number) are set. In S402, the distribution pace (%) is obtained by comparing the current number of impressions with the upper and lower limits of the number of impressions. In S403, it is determined whether this distribution pace exceeds 100%. If it exceeds 100%, the predicted eCPM is set to 0 as shown in S404, and the priority is set to the lowest. In S405, for an ad network whose distribution pace is smaller than 100%, a higher priority is assigned in order from the lowest distribution pace.

  FIG. 24 is a diagram illustrating an example of a posting permission / inhibition screen.

  When a publisher provides inventory to an ad network, it is necessary to determine whether or not to publish. In Europe and the United States, it is often the responsibility of each advertiser to take responsibility for the advertisements that appear on the media. Therefore, when providing inventory to advertisers, it is necessary for publishers to check in detail what kinds of advertisements are to be placed and whether or not to place them. The same applies to the case where inventory is provided to an ad network, and depending on the ad network, the number of cases is large, and checking them all is an operational burden on a publisher.

  Therefore, the yield management system 1 centrally manages whether or not the ad network can publish, and links the data with the advertisement placement status survey system 4. For example, an ad that has been inquired about whether or not the ad network can publish is a net advertisement of the company. Is automatically permitted if it has been published in the past, or if it has been published in the past on a site with strict criteria for publication.

  In the posting availability screen of FIG. 24, as indicated by 201, the ad network, the site, the status of posting availability judgment, the answering deadline for posting availability, the advertiser, the last updater name, the product, the last update date, the campaign unit or Display items such as creative unit display switching. In addition, as shown in 202, for each site, regarding the check of whether to publish, the total of the number of unchecked campaigns, the number of unchecked creatives, the number of campaigns close to the response deadline (for example, two days ago) and the number of creatives are displayed. To do.

  A user (for example, a person in charge of a publisher) who performs an operation through the posting permission / inhibition screen is for instructing collectively permission or rejection of posting of all campaigns or creatives for the page on the site including the advertising space. Button 203 is displayed.

  Furthermore, after identifying an ad network, a site, and a campaign, a list of campaign information and creative information is displayed. For individual campaigns, whether or not to publish can be instructed by a button 205, and for individual creatives, whether or not to publish can be instructed by a button 206.

  Here, as a function for realizing whether or not to publish, the white list information of the advertiser or the business type can be obtained from the publication result data 31 for the corresponding site itself or the trusted site set in the site based on the site ID. get. Then, in the order of the advertiser list, the business type advertiser list, and the trusted site advertiser list, it is determined whether or not it matches the target to be determined whether or not to be published. If it matches any of the above, it is determined that there is no need to judge whether or not the posting is permitted, and the posting is automatically set to be possible. The priority may be lowered in the order of the advertiser list, the business type advertiser list, and the trusted site advertiser list, and the determination may be made different according to the priority of the matching list. By displaying only campaigns or creatives that require artificial determination on the posting availability screen of FIG. 24, it is possible to greatly reduce the work of determining whether to allow posting.

  According to the embodiments of the present invention described above, it is possible to provide a yield management system, method and program capable of reducing the operational burden while improving profits by utilizing many ad networks.

  In addition, you may implement this invention as a program for functioning a general purpose computer as the yield management system demonstrated above. A computer-readable recording medium in which the program is recorded may be read from the recording medium reading device and executed by the CPU.

  Although several embodiments of the present invention have been described, these embodiments are presented by way of example and are not intended to limit the scope of the invention. These novel embodiments can be implemented in various other forms, and various omissions, replacements, and changes can be made without departing from the scope of the invention. These embodiments and modifications thereof are included in the scope and gist of the invention, and are included in the invention described in the claims and the equivalents thereof.

1 ... Yield management system;
2 ... User terminal (management screen);
3 ... Advertisement management system;
4 ... Advertisement placement status survey system;
5 ... Publisher Ad Server;
6 ... Consumer terminal (browser);
7… Ad network;
8 ... CDN;
9 ... Network

Claims (6)

Log acquisition means for acquiring the identification information of the advertising content distributed by the ad network and the number of impressions and clicks of the advertising space from the WEB site of the publisher,
A counting means for identifying the campaign of the advertisement displayed in the advertising space based on the identification information of the advertising content, and calculating the number of impressions and the number of clicks per unit time by counting the logs per unit time;
A database that stores CPM (Cost Per Mille) unit price for each campaign for a list of ad networks that the publisher can provide advertising inventory;
An eCPM (Effective Cost Per Mille) is calculated for each campaign based on the number of impressions and clicks per unit time and the CPM unit price, and the predicted eCPM of each ad network is calculated from the eCPM of each campaign, and the prediction optimizing means for selecting an ad network based on eCPM and outputting the corresponding ad tag;
A yield management system comprising: The database further stores publication performance data of advertisers or creatives published in the past,
When a matching advertiser or creative is present in the publication performance data for a campaign or creative submitted from the ad network, a publication that allows the campaign or creative submitted from the ad network to be posted. The yield management system according to claim 1, further comprising support means for determining availability.   When the contract between the publisher and the ad network provider is CPC (Cost Per Click) or CPA (Cost Per Action), the optimization means converts the profit from the CPC or CPA into the profit from the CPM. The yield management system according to claim 1.   2. The yield according to claim 1, wherein the optimization unit calculates the predicted eCPM based on an appearance probability when a contract between a publisher and an ad network provider is revenue share. Management system. The data collection means obtains the identification information of the advertising content distributed by the ad network and the log of the number of impressions and clicks of the advertising space from the website of the publisher,
Based on the identification information of the advertising content, the counting means identifies the campaign of the advertisement displayed on the advertising space, and counts the number of impressions and clicks per unit time by counting the log every unit time. Calculated,
The database stores a CPM (Cost Per Mille) unit price for each campaign for a list of ad networks that the publisher can provide advertising inventory for,
An optimization unit calculates an eCPM (Effective Cost Per Mille) for each campaign based on the number of impressions and clicks per unit time and the CPM unit price, and calculates the predicted eCPM of each ad network from the eCPM of each campaign. A yield management method comprising calculating, selecting an ad network based on the predicted eCPM, and outputting a corresponding ad tag. Computer
Log acquisition means for acquiring identification information of advertisement contents distributed by the ad network and logs of impressions and clicks of the advertising space from the WEB site of the publisher,
A counting means for identifying the campaign of the advertisement displayed in the advertising space based on the identification information of the advertising content, and calculating the number of impressions and clicks per unit time by counting the log for each unit time;
A database that stores CPM (Cost Per Mille) unit price for each campaign for a list of ad networks that the publisher can provide advertisement inventory for,
An eCPM (Effective Cost Per Mille) is calculated for each campaign based on the number of impressions and clicks per unit time and the CPM unit price, and the predicted eCPM of each ad network is calculated from the eCPM of each campaign, and the prediction an optimizing means for selecting an ad network based on eCPM and outputting a corresponding ad tag;
Program to function as.

Images