Distribution control device

Abstract

To improve CTR in an advertisement distribution system.SOLUTION: A distribution control device 3 includes: a distribution condition generation part 302 which generates a plurality of distribution conditions regarding one advertisement, on the basis of a learning model learned a relationship between distribution conditions and a click rate when distributing advertisement in accordance with those distribution conditions; and a distribution control part 303 which distribute the one advertisement by switching the plurality of distribution conditions generated according to a predetermined algorithm.SELECTED DRAWING: Figure 2

Description

Technical Field

[0001] This invention relates to a distribution control device that controls the distribution of advertisements through a distribution medium.

Background Art

[0002] In recent advertising distribution systems, advertisers can specify the distribution time zone and distribution area according to the set target users (see, for example, Patent Document 1).

Prior Art Documents

Patent Documents

[0003]

Patent Document 1

Summary of the Invention

Problems to be Solved by the Invention

[0004] In such an advertising distribution system, the rate at which the distributed advertisement is actually clicked (CTR (Click Through Ratio); the rate at which users respond to the advertisement) is generally about 0.1% (that is, about one click for every 1000 distributions), and it is important to increase the CTR (how to make users interested in the advertisement). However, a method for reliably increasing the CTR is currently unknown.

[0005] This invention has been made in view of the above circumstances, and an object thereof is to improve the responsiveness of users in an advertising distribution system.

Means for Solving the Problems

[0006] The present invention provides a distribution control device having generation means for generating a plurality of distribution conditions based on a learning model that has learned the relationship between the distribution conditions of an advertisement and the click-through rate when the advertisement is distributed according to the distribution conditions, and distribution control means for switching the plurality of generated distribution conditions according to a predetermined algorithm and executing the distribution of the advertisement.

Effect of the Invention

[0007] According to the present invention, the responsiveness of users in an advertisement distribution system can be improved.

Brief Description of the Drawings

[0008]

Figure 1

Figure 2

Figure 3

Figure 4

Figure 5

Figure 6

Modes for Carrying Out the Invention

[0009] Hereinafter, embodiments of the present invention will be described with reference to the drawings.

[0010] FIG. 1 is a block diagram showing the configuration of an advertisement distribution system 100 including a distribution control device 3 according to an embodiment of the present invention. This advertisement distribution system 100 includes a plurality of terminals 1 connected to a network 101 such as the Internet, a plurality of distribution servers 2, a distribution control device 3 according to this embodiment, and an RTB server 4. The RTB server 4 is a device that manages a bidding system for executing bidding based on a method such as Real time bidding between a person who desires to distribute an advertisement (publisher) and a person who manages a distribution medium having an advertisement space. The terminal 1 is, for example, a smartphone, a tablet terminal, a laptop PC, or a desktop PC, and functions as an advertisement provider or a viewer of a medium such as a web page. The distribution server 2 is a means for distributing a medium such as a web page having an advertisement placement frame. The distribution control device 3 is a means for controlling the distribution of advertisements by requesting the distribution server 2 to distribute advertisements via the RTB server 4 in accordance with a request from the terminal 1 as an advertisement provider. Here, the advertisement may be a notice, an advertisement creative, or other information to be distributed, and may be any information that is essentially distributed to an unspecified number of persons, regardless of the content (content) or form (still image, moving image, banner, text, etc.) of the information to be transmitted or the purpose of the advertisement (profit / non-profit, etc.).

[0011] FIG. 2 is a block diagram showing a configuration example of the distribution control device 3 in this embodiment. The processor 31 is a control center of the entire distribution control device 3. The communication unit 32 is a means for communicating with other devices connected to the network 101. The storage unit 33 includes a non-volatile storage unit such as an HD and a volatile storage unit such as a RAM. Various programs executed by the processor 31 are stored in the non-volatile storage unit. The volatile storage unit is used as a work area by the processor 31.

[0012] In this embodiment, a program for controlling the distribution of advertisements is stored in the non-volatile memory of the storage unit 33. By executing this program, the processor 31 functions as a learning model construction unit 301, a distribution condition generation unit 302, and a distribution control unit 303.

[0013] In this embodiment, the distribution control device 3 transmits a request for advertisement distribution to the distribution server 2 via the RTB server 4 in accordance with a request from the terminal 1. This request for advertisement distribution includes distribution conditions such as the design of the advertisement content, the distribution time zone, and the distribution area. The learning model construction unit 301 collects the CTR in the advertisements distributed according to this distribution condition, and constructs a learning model showing the relationship between the distribution condition and the CTR.

[0014] FIG. 3 is a diagram illustrating the processing of the learning model construction unit 301. In the illustrated example, advertisement campaigns C1 to C3 for performing advertisement distribution are carried out, and in each campaign, a plurality of sets Ca of distribution conditions and CTRs are collected. Then, a learning model L that associates the distribution condition with the CTR is constructed based on these sets Ca of distribution conditions and CTRs.

[0015] The distribution condition generation unit 302 is a means for generating a plurality of distribution conditions based on the learning model L. In a preferred embodiment, the distribution condition generation unit 302 includes an input unit 304 and a setting unit 305. FIG. 4 is a diagram illustrating the processing of the distribution condition generation unit 302. In the illustrated example, the distribution condition generation unit 302 generates a plurality of distribution conditions D1 to D3 from one distribution condition D based on the learning model L. The distribution condition D is the distribution condition received by the input unit 304, for example, the distribution condition included in the request for advertisement distribution received from the terminal 1. The distribution condition generation unit 302 modifies each item of the distribution condition D based on the learning model L so that the CTR increases, and generates a plurality of distribution conditions D1 to D3 (in the illustrated example, three distribution conditions are output, but the number of distribution conditions is arbitrary).

[0016] For example, in the learning model L, if in the advertisement banner to be distributed, an image of food or drink is largely placed in the center, it is assumed that a learning result is obtained that there is a tendency for a higher CTR to be more likely to be obtained when distributed towards a wide area at night. In this case, the distribution condition generation unit 302 generates a distribution condition that designates an image of a content design in which a photo of a size larger than the image specified by the user is placed as a variation of the advertisement banner image. Also, for the distribution time zone, one with a higher ratio at night is specified. Also, for the distribution area, an area adjacent to the area specified by the user is added and specified.

[0017] FIG. 5 is a diagram showing a specific example of the processing of the distribution condition generation unit 302. In this example, the distribution condition D includes items such as content design Da, distribution area Db, distribution time zone Dc, distribution medium Dd, and user attribute De. Here, the content design Da includes what is also called a so-called advertisement creative, and it may be content that is distributed to a substantially unspecified number of people, such as a still image banner advertisement or a video advertisement, for commercial or private use (by general users, event announcements, recruitment, and other messages), and the content and purpose thereof are not limited. In distribution condition D1, each of these items is switched to items D1a to D1e, in distribution condition D2, each of these items is switched to items D2a to D2e, and in distribution condition D3, each of these items is switched to items D2a to D2e. The plurality of distribution conditions obtained in this way are set in the distribution control unit 303 by the setting unit 305.

[0018] The distribution control unit 303 switches the plurality of distribution conditions generated by the distribution condition generation unit 302 according to a predetermined algorithm and transmits them to the plurality of distribution servers 2 via the RTB server 4. The timing for switching the plurality of distribution conditions and the algorithm for determining the next distribution condition to be switched are arbitrary. For example, every time a predetermined period elapses, the next distribution condition to be switched may be randomly selected.

[0019] The RTB server 4 transfers the distribution conditions included in the distribution request to each distribution server 2 and conducts an auction for purchasing the advertising spaces of the media provided by each distribution server 2 for advertising distribution.

[0020] The distribution server 2 that bids in this auction conducts distribution according to the distribution conditions. For example, the operation of the distribution server 2 that receives the distribution condition D1 in FIG. 5 is as follows. First, in the distribution condition D1, since the distribution medium Dd1 is the top 100 sites, the said distribution server 2 is a distribution medium belonging to the top 100 sites. In the distribution condition D1, the distribution time zone Dc1 is in the morning, the distribution area Db1 is only the 23 wards of Tokyo, and the user attribute De1 is all user attributes. Therefore, when there is an access from a user within the 23 wards of Tokyo in the morning, the said distribution server 2 distributes the advertisement of the content design Da1.

[0021] The distribution server 2 monitors the occurrence status of clicks at the distribution destination for the advertisements distributed in this way, calculates the CTR for each distribution condition, and notifies the distribution control device 3 of the CTR for each distribution condition. The distribution control device 3 constructs the learning model L based on the CTR reported from each distribution server 2 in this way.

[0022] The inventor of the present application has found that when continuously operating by finely changing the variations of banner images and other creatives, the distribution timing, and the distribution area for substantially the same advertising case (one campaign), the CTR will increase dramatically. However, it is troublesome and not very realistic to manually implement such changes in distribution conditions.

[0023] Therefore, a plurality of patterns of items (banners, delivery timings, delivery areas) of delivery conditions for which CTR improvement can be expected based on the results of machine learning are preset in advance, and a system is constructed that automatically and appropriately combines these and switches and delivers them at a predetermined timing (for example, every day). This is the delivery control device 3 of the present embodiment. As a result, it was confirmed that the CTR becomes 10 times (a value that cannot be considered in conventional common sense) compared to the case where such switching of delivery settings is not performed.

[0024] As described above, one embodiment of the present invention has been described, but other embodiments are also conceivable for the present invention. For example, as follows. The technical elements according to each embodiment can be combined as appropriate.

[0025] FIG. 6 is a time chart showing the processing of the delivery control unit 303 in another embodiment of the present invention. In this embodiment, when the delivery control unit 303 executes delivery using one delivery condition for a period of a predetermined length T (for example, one day), it determines whether or not the CTR related to the delivery is equal to or greater than a threshold value (step S1). If it is not equal to or greater than the threshold value, it uses other delivery conditions among the plurality of delivery conditions (step S2) and executes delivery for a period of a predetermined length T. On the other hand, if it is equal to or greater than the threshold value, it executes delivery for a period of a predetermined length T using the one delivery condition. According to this embodiment, when the CTR tends to decrease, the delivery conditions are switched, so it is possible to avoid the continuous decrease of the CTR and effectively increase the CTR.

[0026] The delivery condition generation unit 302 can be made to automatically generate various items such as banner advertisement images, delivery areas, delivery timings, and target user attribute variations. Alternatively, the user may input some items of the delivery conditions, and the delivery condition generation unit 302 may complement the remaining items. Alternatively, the user may specify values (variations) for all items, and the delivery condition generation unit 302 may combine these values to configure the delivery conditions.

[0027] In the above-described embodiment, the distribution control device 3 is interposed between the terminal 1 and the RTB 4, but the distribution control device 3 may be included in the distribution server 2. Alternatively, the distribution control device 3 may be included in a server that has both the functions of the RTB 4 and the distribution server 2.

[0028] In short, in an information processing system composed of one or more processors or information processing devices according to the present invention, for one advertisement, a step of generating a plurality of distribution conditions based on a learning model that has learned the relationship between the distribution conditions and the click-through rate when the advertisement is distributed according to the distribution conditions, and a step of switching the generated plurality of distribution conditions according to a predetermined algorithm and executing the distribution of the one advertisement may be executed.

Explanation of Signs

[0029] 100... advertisement distribution system, 1... terminal, 2... distribution server, 3... distribution control device, 4... RTB server, 31... processor, 32... communication unit, 33... storage unit, 301... learning model construction unit, 302... distribution condition generation unit, 303... distribution control unit, 304... input unit, 305... setting unit.

Claims

1. Generating means for generating a plurality of distribution conditions based on a learning model that has learned the relationship between the distribution conditions for one advertisement and the click-through rate when the advertisement is distributed according to the distribution conditions; Distribution control means for switching the plurality of generated distribution conditions according to a predetermined algorithm and executing the distribution of the one advertisement; A distribution control device having the above.

2. The distribution conditions include items regarding the design of the advertisement content, the distribution time zone, and the distribution area. The distribution control means updates the combination of the design of the advertisement content, the distribution time zone, and the distribution area at a predetermined timing. The distribution control device according to Claim 1.

3. Input means for receiving the input of at least one of the plurality of distribution conditions; Setting means for generating the plurality of distribution conditions based on the input of the at least one distribution condition; The distribution control device according to Claim 1, having the above.

4. When the distribution control means executes the distribution using one distribution condition for a period of a predetermined length, it determines whether the click-through rate related to the distribution is equal to or higher than a threshold value. If it is not equal to or higher than the threshold value, it executes the distribution for a period of a predetermined length using another distribution condition among the plurality of distribution conditions. On the other hand, if it is equal to or higher than the threshold value, it executes the distribution for a period of a predetermined length using the one distribution condition. The distribution control device according to Claim 1.

5. A program for causing a computer to: generate a plurality of distribution conditions based on a learning model that has learned the relationship between the distribution conditions for one advertisement and the click-through rate when the advertisement is distributed according to the distribution conditions; switch the plurality of generated distribution conditions according to a predetermined algorithm and execute the distribution of the one advertisement. ​

Images