JP2017188031A - Re-marketing system based on action data and emotion data and management method for the same - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a re-marketing method and system based on action data and emotion data which can more precisely utilize user trend by combining the user's action data with emotion data read from screen operation.SOLUTION: A marketing device includes the steps of: transmitting an action to each user whose prediction conversion rate with respect to a Web site is a set value or larger; receiving and storing a prediction purchase probability of commodities determined by an analysis/prediction device on the basis of the user's reaction to the action and conversion achievement information with respect to the Web site; extracting and analyzing each non-purchaser whose prediction purchase probability of the commodities is a set value or larger; establishing a target user on the basis of a result of the analysis; and distributing to the target user an advertisement. The prediction conversion rate is determined by the analysis/prediction device on the basis of emotion information and action information acquired from screen operation of a user visiting the Web site.SELECTED DRAWING: Figure 3

Description

  The present invention relates to a remarketing system based on behavior data and emotion data, and an operation method thereof, and in particular, behavior data and emotion that can more accurately utilize user trends by combining emotion data that can be read from screen operations with user behavior data. The present invention relates to a remarketing system based on data and an operation method thereof.

  Marketing techniques are widely used that provide product information to users who are viewers through websites and encourage the purchase of the product. The purpose of acquiring many orders for the business that provides product information is, of course, to make the offered product information attractive to users, but the information provision is efficient. In particular, a method of providing information linked to product orders is also important. Even if information is provided to users who are not interested in the product or users who are interested but do not intend to purchase, the probability of actual purchase is low, and the result cannot be expected even if advertising costs are spent. On the other hand, for users who are highly interested in the product and users who are highly motivated to purchase, there is a great expectation that effective information provision will lead to product purchase.

  In order to grasp the user's interest, there is an advertisement distribution method for grasping the history of the website visited by the user and displaying the advertisement of the product related to the website on the screen of the website being browsed by the user. Yes. Thus, it is expected that the advertisement related to the website once the user has withdrawn from browsing will be connected to the purchase of the product by remarketing. However, it is expected that at least the user is interested in the website only by grasping the history of the website visited by the user, but it is impossible to grasp how much the user has browsed the website.

  In order to estimate the degree of interest of the user more accurately, an attempt has been made to grasp the user's emotion. In Patent Document 1, the feature amount is calculated from the content determined to be being browsed, the biometric signal of the user is captured during the browsing period of the content, and the feature amount is calculated from the biometric signal. It is disclosed that a user's emotion corresponding to a feature amount of a content and a feature amount of a biological signal is estimated by collating with registered reference data. Patent Document 2 discloses a method for autonomously generating a prediction model for preference relations regarding products from biological information.

However, in these prior arts, in order to grasp the user's emotions, sensors and devices for assuming the user's skin potential and heart rate are necessary, and it is necessary for advertisement distribution of websites targeting a large number of unspecified users. It is not realistic to apply.
Therefore, there is a need for a remarketing method that can estimate and utilize the user's interests and purchase intentions more accurately without using such special sensors and devices.

Japanese Patent Application Laid-Open No. 2014-222397 JP 2011-066554 A

  The present invention has been made in view of the problems in the above-described conventional remarketing method, and the object of the present invention is to combine user behavior data with emotion data that can be read from a screen operation, and to improve user trends. An object of the present invention is to provide a remarketing system based on behavior data and emotion data that can be used accurately and an operation method thereof.

  In order to achieve the above object, a remarketing system operation method according to the present invention is a remarketing system operation method including a marketing device, an analysis / prediction device, and a user terminal connected by a network. A device that transmits an action to a user whose predicted conversion rate for a website is equal to or greater than a set value; and the marketing device that is determined by the analysis / prediction device based on the user's response to the action. Receiving and storing predicted purchase probability and conversion performance information for the website; and the marketing device extracting and analyzing unpurchased users whose predicted purchase probability of a product is a set value or more; The marketing device is the result of the analysis Determining a target user based on the marketing device, and delivering the advertisement to the target user, wherein the predicted conversion rate is obtained from a screen operation of a user visiting the website It is determined by the analysis / prediction device based on emotion information and behavior information.

It is preferable that the emotion information is acquired in real time by an element including at least one of a mouse speed, behavior, flick, tap speed, and browsing pattern change of the user browsing the website. .
The action includes any one of an A / B test that presents at least two different patterns of web pages to the user, a recommendation that presents information that may be of interest to the user, and a coupon that provides benefits to the user. Is preferred.

The distributed advertisement preferably includes at least one of a display advertisement, a listing advertisement, and a social targeting advertisement.
The predicted purchase probability is determined by the analysis / prediction device based on emotion information and behavior information acquired from a user's screen operation for the action, and the emotion information is at least the user browsing the website It is preferably acquired in real time by an element including any one of the mouse speed, behavior, flick, tap speed, and browsing pattern change.

  A remarketing system according to the present invention made to achieve the above object is a remarketing system having a marketing device, an analysis / prediction device, and a user terminal connected by a network, wherein the analysis / prediction device is A conversion analysis / prediction unit that determines a predicted conversion rate based on emotion information and behavior information of a user who is visiting a website received from a user terminal, and product prediction based on a user's response to an action received from the user terminal A purchase probability analysis / prediction unit for determining a purchase probability, wherein the marketing device associates the predicted conversion rate received from the analysis / prediction device and the predicted purchase probability with a visitor ID to the user's website. And a user management unit that manages the prediction converter A determination unit that determines whether the application rate or the predicted purchase probability is equal to or higher than a set value, and a distribution unit that distributes the action or the advertisement to a selected user terminal based on the determination. , Acquired in real time by an element including at least one of mouse speed, behavior, flick, tap speed, and browsing pattern change during browsing of the website of the user. To do.

The action includes any one of an A / B test that presents at least two different patterns of web pages to the user, a recommendation that presents information that may be of interest to the user, and a coupon that provides benefits to the user. The distributed advertisement preferably includes at least one of a display advertisement, a listing advertisement, and a social targeting advertisement.
The predicted purchase probability is determined based on emotion information and behavior information acquired from the user's screen operation for the action, and the emotion information is at least the speed of the mouse while browsing the website of the user, It is preferably acquired in real time by an element including any one of behavior, flick, tap speed, and browsing pattern change.

  According to the operation method of the remarketing system according to the present invention, in order to acquire emotion information of a user from the use environment of a normal personal computer or mobile terminal such as the speed, behavior, flick, and tap speed of a mouse while browsing a website. In addition, a special device such as a sensor for grasping emotions is not required, and the purchase probability is predicted from the acquired emotion information and behavior information of the user, and the advertisement is distributed to a user with a high predicted purchase probability. Advertisements can be distributed efficiently.

It is a conceptual diagram for demonstrating the operating method of the remarketing system by this invention. It is a conceptual diagram for demonstrating the operating method of the remarketing system by this invention. 1 is a block diagram schematically showing a configuration of a remarketing system according to an embodiment of the present invention. It is a figure for demonstrating the outline of the data table of the remarketing system by embodiment of this invention. 5 is a flowchart for explaining a method of operating a remarketing system according to an embodiment of the present invention. 5 is a flowchart for explaining a method of operating a remarketing system according to an embodiment of the present invention. FIG. 5 is a diagram for conceptually explaining a user analysis stage according to an exemplary embodiment of the present invention. FIG. 5 is a diagram for conceptually explaining a user analysis stage according to an exemplary embodiment of the present invention. FIG. 5 is a diagram for conceptually explaining a user analysis stage according to an exemplary embodiment of the present invention.

Next, specific examples of modes for carrying out the remarketing system and the operation method thereof according to the present invention will be described with reference to the drawings.
The remarketing system according to the present invention is browsing the user's website in addition to the historical information indicating what website the user has browsed before, the attribute information such as the user's age and gender. The user's emotion information is obtained from screen operations such as mouse operation and flicking, and the user's interest and purchase probability for the products and services posted on the website are predicted and determined from these information, and stored. It is a system that can perform efficient advertisement distribution by analyzing user characteristics based on stored data, extracting unpurchased users with a high predicted purchase probability, and distributing commercial advertisements.

FIG. 1 is a conceptual diagram for explaining a method of operating a remarketing system according to the present invention, and is a schematic diagram for explaining a basic method for acquiring emotion information.
Referring to FIG. 1, the method for acquiring emotion information includes steps of tag installation 11, emotion information acquisition 12, behavior / emotion analysis 13, moment detection 14, approach 15, and action response 16.

  In tag installation 11, in order to acquire emotion information from the user's screen operation, a tracking code such as detecting and transmitting the coordinate position of the mouse for the screen operation is implemented on the website. In the embodiment, emotion information may be tracked from a screen operation every 0.03 seconds. By tracking with a short cycle, it can be set so that tracking of the trajectory including fine movement of the mouse and change in speed can be caught.

In the emotion information acquisition 12, the emotion information is acquired from the user's screen operation according to the tag set while the user is browsing the website on the user terminal. Emotion information refers to information that captures emotions such as hesitation, jealousy, or prompt decision included in the operation when the user operates the screen as information on how to operate the mouse or the like. When the user terminal is a personal computer, screen operations are generally performed with a mouse, and emotion information includes the speed and behavior of the mouse. In the case of a touch panel type terminal such as a tablet terminal or a smartphone, emotion information includes flick and tap speeds.
Also, emotion information includes changes in browsing patterns such as how users browse web pages and how to browse displayed advertisement screens, such as how to select and scroll web pages within a website. May be included. The acquired emotion information is transmitted to the analysis / prediction device according to the installed tag.

In the behaviour / emotion analysis 13, a change in the user's behavior and emotion is analyzed based on the acquired emotion information.
There are various backgrounds for users browsing websites. A user who browses a website for the purpose of purchasing a certain product operates the screen so as to reach the target information without being relatively at a loss because the information to be obtained is clear, such as confirmation of the price and specifications of the product.
Users who are interested in the product, but are wondering whether or not to purchase, may become confused or hesitant about the information to select. appear.
A user who is not interested in the product does not visit the website in the first place, or even if he / she visits, he / she leaves the page without performing a fine screen operation.

As described above, since the user who has a high willingness to purchase is linear in the screen operation in a short time as described above, the user who is at a loss tends to be jealous or stray from the screen in one embodiment. The time from the display to the user's screen operation, the speed of the mouse, and the movement distance of the locus from the start to the end of the movement of the mouse are used as evaluation items, and the emotion information is evaluated by weighting each of the evaluation items. Further, when the web page does not fit on the display screen of the user terminal, the degree of interest may be evaluated using the scrolling speed and scrolling range as evaluation items. The evaluation method is not limited to this, and when there are a plurality of information linked to the web page, the degree of interest may be evaluated depending on which link is selected in which order.
The behavior information includes, as a result of the user's screen operation, the number of clicks, the selected web page, the order of selection of the web page, and the like. In the behavior / emotion analysis 13, the state of the user is combined with the emotion information and the behavior information. Is analyzed.

By analyzing where the mouse operated by the user has stopped and where the scrolling has been accelerated, it is possible to know which screen, which advertisement the user has become more interested in, and what is more interested in. , It is possible to obtain information that is extremely effective for delivering an optimal advertisement at an optimal timing.
According to the set evaluation items, it is possible to improve the accuracy of the evaluation algorithm by repeatedly analyzing a lot of emotion information and action information and performing machine learning. Further, by collecting and analyzing a large number of such emotion information and behavior information of various users, users who perform similar screen operations can be grouped as one segment. In this case, more detailed grouping is possible by considering the attribute information of each user. Furthermore, it is possible to make a purchase prediction as to what kind of screen operation the user of the segment will perform the purchase by conducting a follow-up survey on whether or not the user of each segment has reached the purchase of the product and taking the result into consideration it can.

  The moment detection 14 is a step of detecting moments of individual users who are browsing the website according to an evaluation algorithm. A moment is a moment when a user is at a loss about purchasing a product. A user who is wondering about the purchase is likely to move to an action that leads to purchase of the product by giving an opportunity to promote the purchase.

  The ultimate goal of advertising on websites is to have users purchase products, but even if they do not purchase, they can inquire about products and request catalogs. It can be considered as one of the effects. In this way, when the user performs specific actions such as purchase and catalog request for the presentation of the advertisement and achieves the advertising goal, it is called conversion (CV). The ratio of conversion to the number of website accesses is called the conversion rate (CVR).

  Of course, users who browse websites for purchase have a high conversion rate, but users who are wondering about purchase should also be given the opportunity to invite interest in the product or encourage purchase. And is likely to lead to a conversion. The moment detection 14 is to detect such a user's emotional movement.

  The approach 15 transmits an action to a user terminal as an approach for prompting a user who has detected a moment to change behavior. An action is a tool that gives a user an opportunity to take action, specifically, an A / B test that presents two different web pages to the user, a recommendation that presents information that the user may be interested in, and a user For example, a coupon providing a privilege. The number of actions is not limited to one, and a combination of these may be used.

  The action response 16 is a user response to the action transmitted to the user terminal. Since an action that further intrigues is sent to a user who has an interest in the product to the point where he / she has lost the purchase of the product from the beginning, there is a high possibility that the user will take an action leading to conversion with respect to the action. However, of course, some users do not respond to any action sent. The user's emotion information can also be acquired from the way of dealing with such actions.

  In one embodiment, the behavior / emotion analysis after acquisition of emotion information is performed in real time, and when a moment is detected, an action is transmitted before the user leaves the website, and the response to the user's action is grasped. Emotion information acquisition and behavior / emotion analysis can be used to grasp the response to an action.

FIG. 2 is a conceptual diagram for explaining the operation method of the remarketing system according to the present invention, and is a diagram schematically showing a time change of the purchase probability of the user who is browsing the website.
FIG. 2 shows changes over time in the purchase probability of each of three types of users: a user who purchases a product, a user who is unsure about purchase, and an indifferent user. The purchase probability of the user for purchase shows a high value of 90% from the start of browsing the website, maintains a value of 90% or more during browsing, and is relatively stable. Such users have a high conversion rate and lead to conversion.
On the other hand, an indifferent user shows a purchase probability as low as 10% from the start of browsing the website, and while there is some fluctuation during browsing, the purchase probability does not greatly improve and leaves the website as it is.
In the middle of this, a user who is confused about purchase has a large change in purchase probability depending on the browsing contents, and if the purchase probability increases, conversion will occur, and if the purchase probability does not increase, the user will leave the website.

  Even if an approach that sends an action to an indifferent user is implemented, it is not possible to induce an action that leads to conversion by inducing interest. On the other hand, a user who is unsure of purchase is expected to take an action that leads to conversion by taking an approach that captures the moment when the purchase probability increases. Therefore, it is an efficient advertisement distribution method that induces conversion by transmitting an A / B test and advertisement distribution in real time to a user who is particularly at a loss.

FIG. 3 is a block diagram schematically showing the configuration of the remarketing system according to the embodiment of the present invention.
Referring to FIG. 3, the remarketing system according to the embodiment of the present invention includes a marketing device 100, an analysis / prediction device 200, and a user terminal 300 connected via a network 400.

  The marketing device 100 receives the predicted conversion rate and the predicted purchase probability of the website browsing user determined by the analysis / prediction device 200 and manages it together with the visitor ID of the user's website, and the predicted conversion rate or predicted purchase. It is an apparatus that determines whether a probability is equal to or higher than a set value and distributes an action or an advertisement to a user terminal of a selected user based on the determination.

The marketing apparatus 100 includes a control unit 110, an input / output unit 120, a storage unit 130, a display unit 140, a user management unit 150, a determination unit 160, and a distribution unit 170.
The control unit 110 controls the operation of the marketing apparatus 100 itself, and controls each component to perform the following predetermined functions, and is configured by an electronic circuit including at least a processor.
The input / output unit 120 is connected to the analysis / prediction device 200 and the user terminal 300 through the network 400, receives the predicted conversion rate and the predicted purchase probability for each user from the analysis / prediction device 200, and sends actions and advertisements to the user terminal 300. Send.

The storage unit 130 stores a program that controls the operation of the marketing apparatus 100 and stores information such as a predicted conversion rate and a predicted purchase probability for each user, and is configured by a semiconductor memory, a hard disk, or the like. The storage unit 130 may be configured as an internal memory of the marketing apparatus 100 or may be an independent external memory.
The display unit 140 displays information for the operator of the marketing apparatus 100. The information for the operator is data such as a program that controls the operation state and operation of the marketing apparatus 100, a predicted conversion rate, a predicted purchase probability, and the like, and the display unit 140 is embodied by a display or the like.

The user management unit 150 manages the predicted conversion rate and the predicted purchase probability received from the analysis / prediction device in association with the visitor ID of the user's website.
The website can be viewed by anyone if there are no restrictions for specific members. Users browsing the website can be identified by the IP address of the user terminal, so it is possible to acquire and analyze emotion information from the user's screen operation as described above, and to approach by sensing moment It is. However, attribute information such as the age and sex of the user cannot be obtained with only the IP address, and there is a limit to accurately obtaining the predicted conversion rate and the predicted purchase probability.

  Therefore, the user management unit 150 manages the predicted conversion rate and the predicted purchase probability in association with the visitor ID of the user's website. The visitor ID is a user-specific ID that is given when logging in to the website for the first time when the user browses the website, and attribute information such as the user's age and gender is required when logging in. The In one embodiment, when attribute information is input, the input attribute information is transmitted to the marketing apparatus 100 together with the visitor ID.

  The emotion information obtained from the screen operation of the website being browsed by the user is sent to the analysis / prediction device 200 in real time together with the visitor ID for analysis, and the predicted conversion rate and the predicted purchase probability are determined. 100 receives the predicted conversion rate and the predicted purchase probability, and the user management unit 150 manages the predicted conversion rate and the predicted purchase probability in association with the visitor ID and attribute information.

The determination unit 160 determines whether the predicted conversion rate or the predicted purchase probability is greater than or equal to a set value.
In order to efficiently distribute advertisements, it is important to distribute to users who can expect conversion or purchase, and a reference value is set in order to select such users. The set value may be set in advance by an administrator of the marketing apparatus 100 and stored in the storage unit 130. In addition, the reference set value may be obtained by grasping the conversion of the predicted conversion rate or the predicted purchase probability or the actual purchase of the product, and resetting the value to a value at which the actual result can be efficiently expected in the determination unit 160. .
In addition, the determination unit 160 determines whether the predicted conversion rate or the predicted purchase probability for each visitor ID managed by the user management unit 150 is greater than or equal to a reference value. The predicted conversion rate and the predicted purchase probability may be the same reference value or may be set separately.

The distribution unit 170 distributes the action or advertisement to the selected user terminal based on the determination of the determination unit 160.
The distribution unit 170 extracts a user whose predicted conversion rate is equal to or higher than a set value so as to efficiently acquire emotion information, and distributes the action. Actions include an A / B test that presents two different web pages to the user, a recommendation that presents information that may be of interest to the user, a coupon that provides benefits to the user, etc. Sent.

  Further, the distribution unit 170 distributes advertisements by extracting users whose predicted purchase probability is equal to or higher than a set value in order to perform efficient remarketing. Advertising includes display advertisements that are embedded and displayed as part of a web page, listing advertisements that are displayed in conjunction with user keyword search results, social targeting advertisements that use the net segmented according to hobbies and preferences, etc. And any one of these may be combined.

Referring to FIG. 3, the analysis / prediction apparatus 200 includes a control unit 210, an input / output unit 220, a storage unit 230, a display unit 240, a conversion analysis / prediction unit 250, and a purchase probability analysis / prediction unit 260.
The analysis / prediction device 200 analyzes the movement and behavior of the user's emotion based on the emotion information and behavior information acquired from the user's screen operation transmitted from the user terminal 300, and predicts the user's predicted conversion rate and the predicted purchase of the product. It is a device that determines the probability.

The control unit 210 controls the operation of the analysis / prediction apparatus 200 itself, and controls each component to perform the following predetermined function, and is configured by an electronic circuit including at least a processor.
The input / output unit 220 is connected to the marketing device 100 and the user terminal 300 through the network 400, receives emotion information from the user terminal 300, and transmits a predicted conversion rate and a predicted purchase probability for each user to the marketing device 100.
The storage unit 230 stores a program that controls the operation of the analysis / prediction apparatus 200 and stores information such as a predicted conversion rate and a predicted purchase probability for each user, and is configured by a semiconductor memory, a hard disk, or the like. The storage unit 230 may be configured as an internal memory of the analysis / prediction device 200 or may be an independent external memory.
The display unit 240 displays information for the operator of the analysis / prediction device 200.

  The conversion analysis / prediction unit 250 determines a predicted conversion rate based on emotion information and behavior information of a user who is visiting the website received from the user terminal 300. The behavior information is information about the behavior associated with the user visiting the website. The number of visits to the website, the number of items that have been put into the cart to purchase products, the number of bounces that have been visited but have left, Includes the number of clicks made during the visit. Furthermore, the number of impressions (Imp number) indicating the number of advertisements displayed while the user visited the site may be included.

The behavior information is updated with the behavior of the user visiting the website. The behavior information is received from the user terminal 300 together with the emotion information and stored in the storage unit 230. The stored behavior information includes past visit records and also serves as history information.
The acquisition of emotion information is performed according to the user's screen operation in accordance with the method described with reference to FIG.
The conversion analysis / prediction unit 250 evaluates the degree of interest by analyzing the degree of decision or hesitation of the user based on the emotion information, and evaluates the degree of interest based on the number of visits or the number of bounces based on the behavior information. To determine the predicted conversion rate.

The purchase probability analysis / prediction unit 260 determines the predicted purchase probability of the product based on the user's response to the action received from the user terminal 300. The action is a tool that gives the user an opportunity to take action as described above, and is an A / B test, a recommendation, a coupon, or the like. The emotion information at that time is acquired depending on how the user performs a screen operation on the action transmitted from the marketing apparatus 100 to the user terminal 300. The acquisition of emotion information follows the method described with reference to FIG.
Also, action information is acquired together with emotion information for the action, received from the user terminal 300 and stored in the storage unit 230. The action information regarding the action may be information shared with the action information associated with the user visiting the website, or may be stored separately.

The purchase probability analysis / prediction unit 260 evaluates the degree of interest by analyzing the degree of user's decision or hesitation based on the emotion information about the action, and based on the number of visits or the number of carts entered based on the action information about the action. The degree of purchase is evaluated, and these are combined to determine the predicted purchase probability.
The analysis / prediction device may be combined as a single device, and as shown by the analysis / prediction device-1 and analysis / prediction device-2 in FIG. Also good. Moreover, it is good also as a structure divided | segmented according to a function and including the analysis / prediction apparatus of emotion information, and the analysis / prediction of action information. The configuration of the analysis / prediction device can be variously configured by combining numbers and functions.

  The user terminal 300 includes a control unit 310, an input / output unit 320, a storage unit 330, and a display unit 340. In FIG. 1, one user terminal 300 is shown. However, in the embodiment, each terminal of a plurality of users is connected. The user terminal is embodied by a personal computer, a tablet terminal, a smartphone, or the like that can display website information.

FIG. 4 is a diagram for explaining an outline of a data table of the remarketing system according to the embodiment of the present invention.
FIG. 4 is an example of a data table including the predicted purchase probability determined by the purchase probability analysis / prediction unit 260 of the analysis / prediction apparatus 200. Data table is associated with each visitor ID, attribute information such as gender, age, hobbies, keywords extracted from product groups that users mainly browse, number of bounces, number of visits, number of carts, etc. In addition to behavior information including information representing behavior, an interest score representing a degree of interest obtained from emotion information is included, and a predicted purchase probability determined based on the information is recorded in association with each other.

  Thus, the data table is used for data management for each visitor ID, and is created and stored by the marketing device 100 and the analysis / prediction device 200, respectively. The data table shown in FIG. 4 is one embodiment, and the items to be associated are not limited to this, and can be selected in various ways. The data tables of the marketing device 100 and the analysis / prediction device 200 may be separate or the same.

  Referring to FIG. 4, the user with visitor ID “543210” visited this website 20 times and bounced twice, but the interest score was 0, indicating that the user has low interest. The predicted purchase probability is as low as 12.7%. On the other hand, the user with the visitor ID “567890” has visited 20 times and has a high interest score of 4 and the predicted purchase probability is as high as 77.5%. Such users are expected to make purchases by sending valid advertisements.

  FIG. 5 is a flowchart for explaining the operation method of the remarketing system according to the embodiment of the present invention, and is a diagram showing a method of predicting the conversion rate and grasping the results when the user visits.

In step S510, the user visits the website, and the user terminal 300 displays the web page on the display unit 340.
5 assumes that the user has logged in at the previous visit, registered attribute information such as age and gender, and has already acquired the visitor ID. Further, it is assumed that the visitor ID and attribute information of the user are transmitted to the marketing device 100 and the analysis / prediction device 200 and are already stored when the visitor ID is acquired.

  While the user visits the website and browses the web page, the user terminal 300 acquires emotion information and behavior information from the user's screen operation according to the tag embedded in the web page, and transmits it to the analysis / prediction apparatus 200 (Step S515). In one embodiment, the user terminal 300 is a personal computer, and emotion information captures the movement of the mouse, and is acquired and transmitted as a mouse position coordinate every 0.03 seconds.

  In step S515, emotion information is acquired from the state of movement of the mouse, and behavior information is acquired from the result selected by the movement of the mouse. In the behavior information, in addition to information on the user's behavior such as the number of clicks of the user and which web page was browsed in which order, the number of impressions indicating the number of advertisements displayed on the website being browsed by the user ( Imp number) may be included. The user terminal 300 transmits the acquired emotion information and behavior information to the analysis / prediction device 200 together with the visitor ID for specifying the user.

The analysis / prediction device 200 analyzes emotion information and behavior information received from the user terminal, calculates the movement speed and movement distance of the mouse, and calculates the average speed from the start of movement, the position of the speed change, the total movement distance, etc. The data is obtained and updated to the latest data (step S520).
The analysis / prediction device 200 obtains a predicted conversion rate based on the analysis results of emotion information and behavior information, and updates data such as the number of impressions and the number of clicks received from the user terminal 300 (step S525).
The data updated in steps S520 and S525 are all stored in the storage unit 230 in association with the visitor ID.

The user terminal 300 transmits the visitor ID and the behavior information to the marketing device 100 as the emotion information and the behavior information are transmitted to the analysis / prediction device 200 (step S530).
Based on the received visitor ID, the marketing device 100 receives data such as a predicted conversion rate, an impression count, and a click count corresponding to the visitor ID from the analysis / prediction device 200, along with the behavior information received from the user terminal 300. Update / save in association with the visitor ID (step S535).

  In FIG. 5, data is shown to be received from the analysis / prediction device 200 based on the visitor ID received from the user terminal 300, but every time the analysis / prediction device 200 obtains a predicted conversion rate and updates the data, the visit is performed. Data updated together with the person ID may be transmitted to the marketing apparatus 100. In this case, step S530 is not necessary.

In step S540, the marketing apparatus 100 determines whether or not the predicted conversion rate of the user is equal to or higher than a preset set value, and if it is equal to or higher than the set value, transmits an action to the user. The action includes any of an A / B test, a recommendation, and a coupon, and may be selected in relation to a web page viewed by the user or information clicked.
The user's feelings change while browsing the web page, and even when the predicted conversion rate is less than the set value at the start of browsing, it may exceed the set value during browsing. In such a case, the content of the action may be selected in relation to information corresponding to the coordinates of the mouse when the predicted conversion rate exceeds the reference value.

In addition, when the marketing apparatus 100 analyzes user information in advance to collect users having similar tendencies in attribute information, behavior information, predicted conversion rate, and the like into segments, and sends an action, You may make it transmit also to this user.
Although not shown, when the predicted conversion rate of the user is less than the set value in step S540, the process returns to step S515, and the process is repeated while the user stays on the website.

When the user responds to the action, the user terminal 300 acquires emotion information and behavior information from the user's screen operation for the action, and transmits it to the analysis / prediction apparatus 200 together with the visitor ID (step S545).
Although not shown, the transmitted information is analyzed and updated by the same processing as the steps S520 to S535, and the process returns to step S540.

  When the user leaves the website in step S550, the information of the departure is transmitted to the analysis / prediction device 200, and the analysis / prediction device 200 confirms whether the user has converted while visiting the website, and the conversion is performed. If so, the conversion performance information is updated and stored in association with the visitor ID. The analysis / prediction device 200 updates the predicted purchase probability of the product based on the user's emotion information and action information regarding the action (step S555). Here, conversion is set in advance as an action to be counted as a conversion among the actions taken by the user in relation to the contents and actions of the website, and when the action is performed by the user, it is determined that the conversion has been performed. You may do it.

In step S560, the marketing device 100 receives and updates the predicted purchase probability and the conversion performance information from the analysis / prediction device 200, and stores them in association with the visitor ID.
The process from step S510 to step S560 in FIG. 5 is a process for one visit from when the user visits the website until the user leaves, but this process is repeated each time the user visits the website. As a result, emotion information, action information, and conversion performance information for the user's screen operation are accumulated, and the analysis / prediction apparatus 200 performs machine learning, thereby improving the accuracy of analysis / prediction.
In FIG. 5, the predicted purchase probability is shown to be updated when the user leaves the website. However, if the user stays on the website for a long time and performs various screen operations for the action, The predicted purchase probability may be obtained and updated every time.

  FIG. 6 is a flowchart for explaining the operation method of the remarketing system according to the embodiment of the present invention. Retargeting and target for extracting a user who is likely to purchase from unpurchased products and setting it as a target Shows how to deliver advertisements to users.

In step S610, the marketing apparatus 100 extracts unpurchased users of products from user data for each visitor ID managed by the user management unit 150.
In step S620, the marketing apparatus 100 extracts users whose predicted purchase probability is equal to or higher than a preset value from the extracted unpurchased users, and transmits the visitor ID of the corresponding user to the analysis / prediction apparatus 200. .
In step S630, the analysis / prediction device 200 extracts analysis data of emotion information and behavior information related to the received visitor ID and transmits the analysis data to the marketing device 100.

The marketing apparatus 100 analyzes the analysis data of the attribute information, emotion information, and behavior information of each user, extracts users having the same attribute information and keywords based on the analysis results, and determines them as the same target user (step S640). .
In step S650, marketing device 100 distributes the advertisement to the user terminal of the determined target user. Advertisements to be distributed include display advertisements, listing advertisements, social targeting advertisements, and the like.

  Although not shown, the marketing device 100 grasps the record of whether conversion has been performed or the purchase has been made in response to the transmitted advertisement, accumulates information for each user, and distributes user analysis and advertisements. It may be used to determine a target user to perform. The analysis / prediction device 200 acquires emotion information and behavior information from the response to the advertisement and uses them to improve the accuracy of analysis / prediction.

7 to 9 are diagrams for conceptually explaining a user analysis stage according to an embodiment of the present invention.
FIG. 7 schematically shows the elements for analyzing the user. The analysis is preferably performed separately for users who have made conversions such as requesting materials and purchasing products, and users who have not yet purchased or signed up. Then, for each user, browse the web page that was viewed, which keyword of the web page you were interested in, what product you viewed, age, gender, etc. Whether to browse while wondering whether or not, and what is the purchase probability for the product, is a criterion for analyzing the user.
With respect to the browsing page, the browsing time and the number of times may be used as criteria for determination.

FIG. 8 shows a method of scoring and classifying users who have visited a website as one method for analyzing users.
FIG. 8 shows a case where scoring is performed from the three aspects of the last site visit date, the number of visits to the website, and the purchase probability of the products posted on the website.
User A visits within one week, but the number of visits is small, and the purchase probability is 50%, so that the user A is generally classified as a good user. User B has a large number of visits and a high purchase probability, so it is generally classified as a good user. User C does not visit for more than a month and purchase probability is 20% or less. being classified. Thus, it is effective to evaluate the user's analysis from many aspects and make a comprehensive judgment.

FIG. 9 shows a method for analyzing users from the aspect of website visit rate and purchase probability.
The users mapped to the area A with a low purchase probability are groups of users who visit but do not purchase. Users located in the region B with a high purchase probability and a high visit rate are very interested in the product and are a group of users with a high conversion rate. Although the visit rate is low, users in the region C with a high purchase probability are users who can expect conversion if they visit. FIG. 9 shows an example of analysis of health foods. It is assumed that an age group of 60 years or older has a high interest and purchase probability, and a purchase probability of a user having a relatively high annual income is high.
By analyzing users in this way and grouping them based on their characteristics, it is possible to appropriately target users who may lead to conversion or purchase when an advertisement is sent, and to perform efficient advertisement distribution Become.

  As mentioned above, although embodiment of this invention was described in detail, referring drawings, this invention is not limited to the above-mentioned embodiment, In the range which does not deviate from the technical scope of this invention, it changes variously. Is possible.

DESCRIPTION OF SYMBOLS 1 Remarketing system 100 Marketing apparatus 110,210,310 Control part 120,220,320 Input / output part 130,230,330 Storage part 140,240,340 Display part 150 User management part 160 Judgment part 170 Delivery part 200 Analysis / prediction Device 250 Conversion Analysis / Prediction Unit 260 Purchase Probability Analysis / Prediction Unit 300 User Terminal

Claims (8)

An operation method of a remarketing system including a marketing device, an analysis / prediction device, and a user terminal connected by a network,
The marketing device transmits an action to a user whose predicted conversion rate for a website is a set value or more;
The marketing device receives and stores the predicted purchase probability of the product determined by the analysis / prediction device based on the user's response to the action and the conversion performance information for the website;
The marketing device extracting and analyzing unpurchased users whose predicted purchase probability of the product is a set value or more;
The marketing device determines a target user based on the result of the analysis;
The marketing device comprises delivering an advertisement to the target user;
Operation of the remarketing system, wherein the predicted conversion rate is determined by the analysis / prediction device based on emotion information and behavior information acquired from a screen operation of a user visiting the website Method.   The emotion information is acquired in real time by an element including at least one of a mouse speed, a behavior, a flick, a tap speed, and a change in a browsing pattern of the user browsing the website. The operation method of the remarketing system according to claim 1.   The action includes any one of an A / B test that presents at least two different patterns of web pages to the user, a recommendation that presents information that may be of interest to the user, and a coupon that provides benefits to the user. The remarketing system operating method according to claim 1, wherein:   The remarketing system operating method according to claim 1, wherein the distributed advertisement includes at least one of a display advertisement, a listing advertisement, and a social targeting advertisement. The predicted purchase probability is determined by the analysis / prediction device based on emotion information and behavior information acquired from a user's screen operation for the action,
The emotion information is acquired in real time by an element including at least one of a mouse speed, a behavior, a flick, a tap speed, and a change in browsing pattern of the user browsing the website. The operation method of the remarketing system according to claim 1. A remarketing system having a marketing device, an analysis / prediction device, and a user terminal connected by a network,
The analysis / prediction device includes a conversion analysis / prediction unit that determines a predicted conversion rate based on emotion information and behavior information of a user visiting the website received from the user terminal, and an action received from the user terminal. A purchase probability analysis / prediction unit that determines the predicted purchase probability of the product based on the user's response,
The marketing device includes a user management unit that manages the predicted conversion rate and the predicted purchase probability received from the analysis / prediction device in association with a visitor ID of the user's website, and the predicted conversion rate or the predicted value. A determination unit that determines whether the purchase probability is equal to or higher than a set value; and a distribution unit that distributes the action or advertisement to the selected user terminal based on the determination;
The emotion information is acquired in real time by an element including at least one of a mouse speed, a behavior, a flick, a tap speed, and a change in browsing pattern of the user browsing the website. Remarketing system characterized by that. The action includes any one of an A / B test that presents at least two different patterns of web pages to the user, a recommendation that presents information that may be of interest to the user, and a coupon that provides benefits to the user.
The remarketing system according to claim 6, wherein the delivered advertisement includes at least one of a display advertisement, a listing advertisement, and a social targeting advertisement. The predicted purchase probability is determined based on emotion information and behavior information acquired from a user's screen operation for the action,
The emotion information is acquired in real time by an element including at least one of a mouse speed, a behavior, a flick, a tap speed, and a change in browsing pattern of the user browsing the website. The remarketing system according to claim 6.

Images