Information processor, information processing method, and computer program

Abstract

To provide an information processor, an information processing method, and a computer program that visualize a potential customer of a commodity and a service and realize an accurate approach to the potential customer.SOLUTION: A life action estimation system includes a life action estimation device, a member terminal, and a customer terminal. The life action estimation device includes: a survey result acquisition unit that acquires a survey result related to a survey of consumption behavior of a survey target member; an estimation unit that inputs member information of the survey target member, inputs, by using an estimation model trained to output the survey result, membership information of an estimation target member, and estimates a score related to the consumption behavior; and a profile generation unit that generates, on the basis of the score, a profile of the estimation target member.SELECTED DRAWING: Figure 2

Description

[Technical Field]

[0001] The present invention relates to an information processing device, an information processing method, and a computer program for supporting marketing of products and services. [Background technology]

[0002] In order to promote the sale of products and the provision of services, various marketing techniques such as research, advertising, publicity, sales promotion, etc. are implemented. One of these marketing techniques is to estimate potential customers of a product or service and then promote the product or service to these potential customers by means of direct mail, etc., and techniques for estimating such potential customers have also been developed.

[0003] Patent Document 1 describes that when a pharmaceutical company has a demand to identify people who are suffering from a specific symptom, the specific symptom is designated as an index item to be indexed, and by inputting the index item into a customer estimation device, it is possible to estimate not only potential customers for the index item from among all customers, but also potential customers (targets) who are unlikely to be affected at present but are expected to have demand in the future.

[0004] Patent document 2 describes a sales activity support system that more effectively supports sales activities, in which a predictive model is generated using training data that associates the success or failure of a specified objective in a user's sales activity with multiple variables that can be correlated with the success or failure of the objective, and predicts at least one of the success or failure of the objective and the probability of success for a potential client based on information about the potential client for the sales activity.

[0005] Patent Document 3 describes a sales support system that automatically extracts a list of customers to whom product proposals can be made from among existing customers, and enables seller companies to realize one-to-one marketing to their customers using this list of customers to whom product proposals can be made.The system extracts potential customers to whom new product purchase proposals should be made based on the properties of the products, the relationships between products, past purchase history, etc., and further extracts these potential customers to suit the seller's needs based on the results of accounting analysis and trouble analysis. [Prior art documents] [Patent documents]

[0006] [Patent Document 1] Patent Publication No. 2021-43513 [Patent Document 2] Japanese Patent Publication No. 2022-69602 [Patent Document 3] Japanese Patent Application Laid-Open No. 2002-443201 Summary of the Invention [Problem to be solved by the invention]

[0007] In conventional technology, in order to estimate potential customers for a certain product or service, a learning model is used that outputs variables indicating the likelihood of a customer becoming a potential customer for the product or service when customer information is input, thereby estimating potential customers who are expected to have demand in the future. However, the potential customers that a company wants to estimate vary depending on the company's product or service, and it is necessary to set customer information and variables representing potential customers according to the company's needs, making preparations such as collecting customer information difficult. Furthermore, even if customer information is collected, it requires a lot of effort to generate and use a learning model, such as training and tuning the learning model.

[0008] The present invention has been made in light of the above-mentioned circumstances, and its purpose is to provide an information processing device, an information processing method, and a computer program that visualize potential customers for products and services and enable an accurate approach to potential customers. [Means for solving the problem]

[0009] In order to solve the above problem, a first aspect of the information processing device, information processing method, or computer program disclosed herein is characterized by comprising a survey result acquisition unit that acquires survey results regarding the consumption behavior of survey target members, an estimation unit that inputs member information of the survey target members and estimates a score regarding the consumption behavior based on the member information of the member to be estimated using an estimation model trained to output the survey results, and a profile generation unit that generates a profile of the member to be estimated based on the score.

[0010] A second aspect of the information processing device according to the present disclosure is characterized in that the survey results regarding the consumption behavior of the surveyed members include responses regarding whether or not they have engaged in consumption behavior, collected through a questionnaire survey of the surveyed members.

[0011] A third aspect of the information processing device according to the present disclosure is characterized in that the member information includes at least member attribute information and purchase information.

[0012] A fourth aspect of the information processing device according to the present disclosure is characterized in that the consumption behavior includes a purchase of a product or service that is the subject of the survey, or an interest in the product or service.

[0013] A fifth aspect of the information processing device according to the present disclosure is characterized in that the score indicates the degree to which the consumption behavior of the member to be estimated is similar to the consumption behavior of a group of surveyed members who responded that they had consumption behavior in the questionnaire survey.

[0014] A sixth aspect of the information processing device according to the present disclosure is characterized in that the profile is generated for each product or service segment based on at least member information of members whose scores are equal to or greater than a predetermined threshold. [Effects of the Invention]

[0015] According to the first aspect of the present disclosure, an estimation model for estimating a consumption behavior score can be generated using membership information of existing members, which makes it easy to prepare data for generating the estimation model. Also, by generating a profile report of members who are potential customers of a product or service, it is possible to easily visualize anticipated potential customers.

[0016] According to the second aspect of the present disclosure, a survey is conducted by selecting survey subjects from existing members, which makes it easy to set the scope of the survey and the content of the survey. Furthermore, by adopting a binary response method for the survey, i.e., whether or not the survey has engaged in consumption behavior, the response rate is improved and the collection and aggregation of response results can be efficiently performed.

[0017] According to the third aspect of the present disclosure, attribute information and purchasing information contained in existing member information can be used to estimate consumption behavior, making it possible to easily collect data to generate estimation models and estimate scores related to consumption behavior.

[0018] According to the fourth aspect of the present disclosure, it is possible to visualize the profile of potential customers with respect to major consumption behaviors, such as purchase of the product or service being the subject of the survey or interest in the product or service.

[0019] According to the fifth aspect of the present disclosure, by utilizing the scores related to consumption behavior output by the estimation model, it is possible to easily identify potential customers by utilizing the profiles of high-scoring members among the members subject to estimation.

[0020] According to the sixth embodiment of the present disclosure, by preparing ready-made profile reports of members who could be potential customers for each product or service segment, even when an analysis of potential customers in a certain segment is required, a profile report for that segment can be quickly provided. [Brief explanation of the drawings]

[0021] [Figure 1] FIG. 1 is a bird's-eye view showing an example of the overall configuration of a life action estimation system 1 to which an information processing device, an information processing method, and a computer program according to a first embodiment of the present disclosure are applied. [Figure 2] FIG. 2 is a diagram showing an example of an information processing flow of the life action estimation device 2 according to the first embodiment of the present disclosure. [Figure 3] FIG. 3 is a block diagram showing an example of a functional configuration of the life action estimation device 2 according to the first embodiment of the present disclosure. [Figure 4] FIG. 4 is a block diagram showing an example of member information according to the first embodiment of the present disclosure. [Figure 5] FIG. 5 is a diagram for explaining an example of generation and use of an estimation model according to an embodiment of the present disclosure. [Figure 6] FIG. 6 is a diagram illustrating an example of a profile report according to an embodiment of the present disclosure. [Figure 7] FIG. 7 is a flowchart for explaining an example of the operation of the life action estimation device 2 according to the first embodiment of the present disclosure. [Figure 8] FIG. 8 is a diagram showing an example of an information processing flow of the life action estimation system 1 according to the second embodiment. DETAILED DESCRIPTION OF THE INVENTION

[0022] First Embodiment An information processing device, an information processing method, and a computer program according to a first embodiment of the present disclosure will be described below with reference to the drawings. In the following description, configurations and elements that are the same as or similar to configurations that have already been described will be assigned the same reference numerals and descriptions thereof will be omitted.

[0023] (1) Overall structure of the Life Action Estimation System 1 FIG. 1 is a bird's-eye view showing an example of the overall configuration of a life action estimation system 1 to which an information processing device, an information processing method, and a computer program according to a first embodiment of the present disclosure are applied. The life action estimation system 1 shown in FIG. 1 is an information processing system for visualizing potential customers for products and services and assisting in approaching these potential customers. It includes a life action estimation device 2, a member terminal 3, and a customer terminal 4, which are connected via a network 5 such as the Internet. The life action estimation device 2 may be an independent information processing device, or may be configured with multiple information processing devices. In FIG. 1, any number of member terminals 3 and customer terminals 4 may be connected to the life action estimation device 2.

[0024] The life action estimation system 1 disclosed herein acquires survey results from questionnaires and research conducted on members, scores the consumption behavior of members using the survey results and member information, analyzes the daily actions (hereinafter also referred to as "life actions") of high-scoring members based on their attributes and values ​​for each product or service segment, and provides the results as a profile of the member. Here, consumption behavior includes actions that consumers may take from the time they become aware of a product or service to the time they purchase it, such as awareness, interest, desire, memory, and behavior (purchase). Furthermore, segments are units that subdivide products and services into groups that are easier to approach.

[0025] In this disclosure, a member is defined as a person who participates in a specific service program, such as a loyalty program. In a narrow sense, member information includes information about members of a specific service program, but in this disclosure, it also includes information about members of a specific service program, as well as information about purchases and values. In other words, in this disclosure, member information is defined in a broad sense to include various information about members, such as attribute information, purchase information, and value information about members of a specific service program.

[0026] Member attribute information includes information such as the member's gender, age, marital status, occupation, and address. Member purchase information includes information such as the store where the member purchased the product or service, the purchase date, the product name, and the quantity. Purchase information may also include POS data when the member purchased the product or service at the store. Member value information includes value information such as the member's personal values, economic values, and values ​​regarding food. Value information includes value information collected through member surveys, etc.

[0027] The life action estimation device 2 obtains survey results on the consumption behavior of the surveyed members, inputs the membership information of the surveyed members, and uses an estimation model trained to output the survey results to estimate a score related to consumption behavior based on the membership information of the surveyed members, and generates a profile of the surveyed members based on the score.

[0028] Survey target members are members of a service program who are the subject of surveys such as questionnaires and research. For example, they are a predetermined number of members randomly selected from all members participating in a specific service program, such as a loyalty program. The predetermined number of members corresponds to a sample when all members of the service program are considered to be the population, and the predetermined number of members can be selected while taking sampling error into consideration. For example, a group of members representing approximately 0.2% of all members of the service program can be selected as survey target members.

[0029] The survey results on the consumption behavior of the surveyed members include responses to a questionnaire survey administered to the surveyed members. The questionnaire survey includes questions about the consumption behavior of a certain segment, and is structured so that the respondents can answer with a binary answer indicating whether or not they have engaged in consumption behavior, such as "yes, I purchased" or "no, I did not purchase" in response to a question about whether or not they have purchased a product or service belonging to that segment.

[0030] The estimation model is a learning model trained by machine learning to input member information of surveyed members and output survey results for the surveyed members. The estimation model may be generated by either supervised learning or unsupervised learning, but the first embodiment describes an estimation model trained using training data. In the first embodiment, the estimation model is generated using the results of a survey on the consumption behavior of surveyed members conducted for each segment and the member information of the surveyed members as training data. In other words, the estimation model is trained to input member information of surveyed members and output survey results on consumption behavior. Member information such as the member's gender and sales are input for each member, and the model is trained to output the presence or absence of consumption behavior (yes / no), which is the surveyed member's response to a questionnaire.

[0031] The life action estimation device 2 uses a trained estimation model to input member information of the members to be estimated and estimate a score related to consumption behavior. The members to be estimated are a group of members made up of all members participating in the service program or a group of members selected from all members. The members to be estimated are not a group equal to all consumers, but are large in quantity, diverse, and real-time, and are composed of a number of members that can represent the trends of all consumers. For example, if the total number of consumers is approximately 100 million, approximately 100 million people can be selected as the members to be estimated from all members participating in the service program.

[0032] The score related to consumption behavior is an indicator that indicates the degree of similarity between the consumption behavior of the member subject to estimation input into the estimation model and the consumption behavior of the member group of surveyed members who responded "yes" to the questionnaire. A high score indicates that the consumption behavior of the member subject to estimation is similar to the consumption behavior of the member group of surveyed members who responded "yes" to the questionnaire. For example, when the membership information of a certain member subject to estimation is input into the estimation model and a score value higher than a predetermined threshold is output from the estimation model, it is estimated that the member subject to estimation is likely to exhibit consumption behavior similar to that of the member group of surveyed members who responded "yes" to the questionnaire regarding products and services in that segment. In other words, when a high score is output when the membership information of a member subject to estimation is input, it is estimated that the member subject to estimation is likely to purchase products in a certain segment.

[0033] The life action estimation device 2 selects, for each segment, members with high scores related to consumption behavior from among the members to be estimated as a high-score member group, and compiles the basic information, values, and profile image of the high-score member group to generate a profile report.

[0034] The basic profile information section shows the gender and age composition, occupational composition, marital composition, etc. of the high-scoring member group. The values ​​section shows the value clusters into which the members of the high-scoring member group are distributed. The profile image shows the profile of a member who represents the group. When a customer terminal 4 requests a profile report for a specific segment, the life action estimation device 2 sends the corresponding profile report to the customer terminal 4.

[0035] 2 is a diagram showing an example of an information processing flow by the life action estimation device 2. The information processing flow by the life action estimation device 2 can be broadly divided into an estimation model generation flow that generates an estimation model, and an estimation flow that performs estimation using the generated estimation model. The estimation model generation flow and the estimation flow are separate flows, and the estimation model generated by the estimation model generation flow is used to estimate and generate profiles of members who could become potential customers in the estimation flow.

[0036] In the estimation model generation flow, the life action estimation device 2 executes the steps of conducting a survey on survey target members and obtaining the survey results, linking the survey results of the survey target members with the member information of the survey target members, and generating a model from the member information and the survey results.

[0037] In the step of obtaining survey results, the life action estimation device 2 selects survey target members, conducts a questionnaire survey on consumption behavior for each segment, and obtains responses from the survey target members regarding whether or not they have engaged in consumption behavior. For example, when surveying whether or not they have purchased products or services belonging to a specific segment, the survey target members respond with "Yes, I purchased" or "No, I did not purchase" to the question "Have you purchased product XX?". Alternatively, when surveying whether or not they have an interest in products or services included in a specific segment, the survey target members respond with "Yes, I am interested" or "No, I am not interested" to the question "Are you interested in or concerned about product XX?". The life action estimation device 2 conducts a questionnaire survey for all segments and obtains responses regarding the consumption behavior of the survey target members, such as their interests and purchases. Note that the questions asked to the survey target members are not limited to interests and purchases, but may include questions regarding any consumption behavior.

[0038] In the step of linking the survey results of the surveyed members with member information, the life action estimation device 2 obtains the member ID, attribute information, purchasing information, etc. of the surveyed members, and links the attribute information and purchasing information of the surveyed members with the survey results regarding consumption behavior via the member ID of the surveyed members.

[0039] In the step of generating an estimation model from member information and survey results, the life action estimation device 2 generates the estimation model using the results of the survey of surveyed members and the member information of the surveyed members as training data. The estimation model is trained so that when attribute information and purchase information of surveyed members with a specified member ID are input into the estimation model, the estimation model outputs survey results for the surveyed members with the specified member ID. For example, the estimation model is trained so that when the gender, sales, etc. of each member ID are input, the response "purchased" or "did not purchase" is output. Note that, to obtain the optimal estimation model, multiple estimation models may be generated by repeating steps such as preprocessing, modeling, parameter tuning, and model evaluation, and the model with the highest evaluation may be selected from the multiple estimation models generated.

[0040] The life action estimation device 2 executes an estimation flow of the life actions of the estimation target members using the generated estimation model. The estimation flow includes a step of preparing estimation data for the estimation target members, an expanded estimation step of calculating the scores of the estimation target members using the estimation model, an output step of extracting high-scoring members from the estimation target members, and a profile generation step of outputting the profiles of the high-scoring members as graphs or the like.

[0041] In the step of preparing data for estimation, the life action estimation device 2 selects members to be estimated and acquires member information for the selected members to be estimated. The members to be estimated are randomly selected from among the members of the service program. Since the consumption behavior of the members to be estimated is presumed to represent the trends of consumers as a whole, the attributes of the members to be estimated, such as age structure, gender, and occupation, may be selected so that they are similar to the attributes of the consumer population. For example, the gender and age distribution of the members to be estimated may be selected so that there is no bias compared to the gender and age distribution of the consumer population.

[0042] It is desirable to use the same information items as the member information of the members subject to estimation as the member information of the members subject to survey used when training the estimation model. For example, if gender was used as the attribute information of the members subject to survey and sales as the purchasing information when training the estimation model, it is desirable to use gender and sales as the attribute information and purchasing information of the members subject to estimation.

[0043] In the expanded estimation step, the life action estimation device 2 estimates the consumption behavior of a group of members to be estimated, which is a larger group than the group of members to be surveyed, based on the survey results on the consumption behavior of the members to be surveyed. Specifically, the member information of the members to be estimated is input into an estimation model, and a score related to the consumption behavior is calculated.

[0044] In the output step, the life action estimation device 2 compares the score value of the estimation target member with a predetermined threshold based on the score calculated by the estimation model, and extracts members with scores higher than the threshold as high-score members. High-score members among the estimation target members are estimated to be highly likely to engage in consumption behavior similar to that of a group of survey target members who have "consumption behavior" for products and services belonging to a certain segment.

[0045] In the profile generation step, the life action estimation device 2 outputs a profile report that compiles the profiles of high-scoring members in tables, graphs, profile images, etc., allowing the characteristics of high-scoring members to be visually understood. From the profiles of high-scoring members, it is possible to estimate the profiles of potential customers in the segment. This profile report is sent to the customer terminal 4 in response to a request from the customer.

[0046] The member terminal 3 is an information processing device used by members who participate in a specific service program, and is communicatively connected to the life action estimation device 2 via the network 5, and has the function of receiving survey requests from the life action estimation device 2 and transmitting survey results.

[0047] The customer terminal 4 is an information processing device used by customers who use the life action estimation services provided by the life action estimation system 1, and is communicatively connected to the life action estimation device 2 via the network 5.The customer terminal 4 sends an estimation request for potential customers in a specific segment to the life action estimation device 2, and receives profile reports of high-scoring members in the specific segment from the life action estimation device 2.

[0048] (2) Hardware configuration (2-1) Life Action Estimation Device 2 The life action estimation device 2 includes a processor 21, a memory 22, a storage 23, a communication interface (communication I / F) 24, an input / output interface (input / output I / F) 25, and a bus 26 connecting these components. The processor 21 controls the overall operation of the life action estimation device 2. The processor 21 may be a general-purpose processor such as a CPU, MPU, or GPU, but is not limited to a general-purpose processor and may also be a dedicated processor such as an ASIC or FPGA. The memory 22 is a main storage device and includes RAM, etc. The storage 23 is an auxiliary storage device and includes a non-volatile storage device such as a hard disk drive (HDD) or a solid-state drive (SSD). The communication interface (communication I / F) 24 is a wired or wireless communication interface and is a module for communicating with devices such as the member terminal 3 and the customer terminal 4 via the network 5. The network 5 is, for example, the Internet and may include access networks such as a LAN, a WAN, a mobile communication network, a wired telephone network, FTTH, and a CATV network. An input / output interface (input / output I / F) 25 receives input data from an external device and outputs output data to an external device.

[0049] (2-2) Member Terminal 3 The member terminal 3 comprises a processor 31, memory 32, storage 33, camera 34, communication interface (communication I / F) 35, input / output interface (input / output I / F) 36, touch screen 37, and bus 30 connecting these. Here, storage 33 stores an operating system (OS), a web browser, various applications, and various data. In addition, touch screen 37 comprises an input unit 38 and a display unit 39. The member terminal 3 is, for example, a mobile terminal such as a smartphone, tablet terminal, or notebook personal computer. The member terminal 3 may also be a stationary computer such as a desktop personal computer.

[0050] (2-3) Customer terminal 4 The customer terminal 4 includes at least a processor, a memory, a storage, a communication interface (communication I / F), an input / output interface (input / output I / F), and a bus connecting these components.

[0051] (3) Functional configuration (3-1) Life Action Estimation Device 2 3 is a block diagram showing an example of the functional configuration of a life action estimation device 2, which is an information processing device according to this embodiment. The functional configuration of the life action estimation device 2 of this embodiment includes a management unit 200, a storage unit 210, and a communication control unit 220. The communication control unit 220 controls communication with the member terminal 3 and the customer terminal 4 via the network 5 in accordance with instructions from the management unit 200.

[0052] (3-1-1) Storage section 210 The memory unit 210 includes a survey information memory unit 211, a member information memory unit 212, an estimation model 213, and a profile information memory unit 214. The survey information memory unit 211 is a memory unit 210 that stores information related to questionnaires, and stores the contents of questionnaires sent to members and the results of questionnaire responses from members. The member information memory unit 212 stores a member table that records member attribute information, etc., a purchase table that stores purchase information, and a value table that stores evaluation value information, etc. The member table, purchase table, and value table are linked by member IDs and form a relational database. The estimation model 213 stores a learning model for estimating the consumption behavior of members. The profile information memory unit 214 stores profile reports for each segment generated by the life action estimation device 2.

[0053] FIG. 4 illustrates an example of member information stored in the member information storage unit 212. (a) shows a member table that records member attribute information, (b) shows a purchase table that records member purchasing information, and (c) shows a value table that stores values. The member table records attribute information about members of a service program. The purchase table records products or services purchased by members at stores, etc., and may include, for example, POS data collected from stores. The value table lists member clusters for each of the following categories: values, communication / media, and consumer awareness / behavior. As described below, clusters are categories that indicate the type to which a member belongs for each of the following categories: values, communication / media, and consumer awareness / behavior. The clusters to which a member belongs can be collected and recorded in advance through a member questionnaire. In FIG. 4, the member table, purchase table, and value table are separate tables linked by member ID. However, these tables may be integrated or further divided into multiple tables.

[0054] (4-1-2) Management Department 200 The management unit 200 functions as a survey result acquisition unit 201, a member information acquisition unit 202, an estimation model generation unit 203, an estimation unit 204, and a profile generation unit 205 by the processor 21 executing a program stored in the memory 22.

[0055] The survey result acquisition unit 201 selects survey target members from among the members who will be the subject of a questionnaire survey, acquires contact information such as email addresses linked to the member IDs of the survey target members from the member information storage unit 212, and sends the survey prepared in the survey information storage unit 211 to the survey target members. The survey result acquisition unit 201 receives the survey results from the survey target members and stores them in the survey information storage unit 211. The survey target members can be selected by randomly extracting a predetermined number of members from all members and setting them as survey target members.

[0056] The questionnaire is configured so that questions are set for each segment and the surveyed members can input their answers to the questions. Responses are input based on whether or not they have engaged in consumption behavior (binary). The survey result acquisition unit 201 acquires the responses to the questionnaire from the member terminal 3 of the surveyed members and stores them in the survey information storage unit 211. The questionnaire may be distributed using any medium, such as email or paper.

[0057] Segments include, for example, segments related to purchasing trends and segments related to interests. Segments related to purchasing trends are made up of multiple segments that divide the market from the perspective of purchases, with major and medium classifications established. Segments related to interests are made up of multiple segments that divide the market from the perspective of interests. Note that segments related to purchasing trends and interests can also be defined by segments subdivided from major to medium classifications.

[0058] Segments related to purchasing trends include gourmet food, books / comics, home appliances, beauty / cosmetics, accessories / jewelry, beer / wine / alcohol, diet / health, interior / furniture, men's / fashion, and women's / fashion. For example, a segment related to purchasing trends could be set up with accessories / jewelry as a major category, and pendants, bracelets, and rings as medium categories.

[0059] Interest segments include banking and finance, beauty and health, food and dining, home gardening, sports and fitness, lifestyle and interests, media and entertainment, news and politics, shopping, technology, travel, and vehicles and transportation. For example, interest segments may be broadly categorized as technology, with medium categories such as tech enthusiasts and mobile fans, and narrow categories such as audiophiles and high-end enthusiasts.

[0060] For example, questions about purchasing tendencies are set for each segment related to purchasing tendencies as questionnaire questions. Furthermore, questions about interests are set for each segment related to interests. Responses to the questions are entered as binary values ​​of "Yes" or "No" in a predetermined response field. For example, for "Accessories & Jewelry," a segment related to purchasing tendencies, the questionnaire about purchasing tendencies is configured to ask, "Have you purchased a pendant?" by entering "1" for "Yes" and "0" for "No" in the response field. Furthermore, for "Food & Dining," a segment related to interests, the questionnaire about interests is configured to ask, "Are you a regular at cafes?" by entering "1" for "Yes" and "0" for "No" in the response field. Upon receiving the questionnaire response results, the survey result acquisition unit 201 links the segment ID for identifying the segment, the question number, and the response result (0 or 1) to the member ID and records them in the survey information storage unit 211.

[0061] The member information acquisition unit 202 acquires attribute information and purchase information of the survey target member from the member table and purchase table stored in the member information storage unit 212. The attribute information includes gender, age, annual income, occupation, marital status, and residence (rental, condominium / apartment, etc.). The purchase information also includes the purchase date, location, product / service name, price, and quantity of the product or service.

[0062] The estimation model generation unit 203 acquires the member IDs and survey results of the surveyed members from the survey result acquisition unit 201, and acquires the member IDs, attribute information, and purchase information from the member information acquisition unit 202. Then, the estimation model generation unit 203 prepares the survey results linked by the member IDs of the surveyed members, the attribute information, and the purchase information as training data, and trains the estimation model 213 so that when the attribute information and purchase information are input for each member ID, the model outputs survey results.

[0063] The estimation model generation unit 203 may generate multiple models using multiple datasets of training data, and adopt the model with the highest accuracy from among them as the estimation model 213. All or part of the member information and attribute information can be used as training data. For example, the member ID and gender can be obtained from the member information, and the member ID and sales can be obtained from the purchasing information, and the gender and sales can be used as training data. When obtaining sales from purchasing information, if the purchasing table does not have a sales column, sales can be calculated by summing the purchasing data, such as unit price and quantity, recorded in the purchasing table over a specified period.

[0064] The estimation unit 204 acquires the attribute information and purchase information of the members to be estimated from the member information storage unit 212 and inputs them into the estimation model 213. The attribute information and purchase information of the members to be estimated can use the same information items as the attribute information and purchase information of the members to be surveyed that were used in the training data. For example, if the gender and sales of the members to be surveyed are used in the training data, estimation can be performed using the gender and sales from the attribute information and purchase information of the members to be estimated.

[0065] When the attribute information and purchase information of the member subject to estimation are input into the estimation model 213, the estimation model 213 outputs a score related to the consumption behavior estimated from the attribute information and purchase information of the member subject to estimation. This score indicates the degree to which the consumption behavior of the member subject to estimation is similar to the consumption behavior of the group of surveyed members who responded "have engaged in consumption behavior," i.e., the degree to which the member subject to estimation is likely to engage in consumption behavior similar to that of the group of surveyed members. The score ranges from 0 to 1; for example, if the consumption behavior of the member subject to estimation is similar to that of the group of surveyed members, the score will be close to 1, and if it is not similar, the score will be close to zero.

[0066] FIG. 5 is a diagram for explaining the generation of the estimation model 213 and estimation by the estimation model 213. When generating the estimation model 213, the estimation model 213 is trained using the gender, sales, etc. linked to the member ID of a certain segment and the survey result y as training data. y is a binary variable that represents the survey result of the user, and is assigned either "1" or "0". For example, for a certain segment regarding purchasing tendencies, if the survey result is "yes, purchased", y is set to 1, and if "no, purchased", y is set to 0. Furthermore, for a segment regarding interests, if the survey result is "yes, interested", y is set to 1, and if "yes, not interested", y is set to 0.

[0067] In Figure 5, for example, when member ID = AAA, gender = M (male), and sales = 1,000 (yen) are input into estimation model 213 as training data for a certain segment, "y = 1" is output, and when member ID = BBB, gender = M (male), and sales = 100 (yen) are input, "y = 0" is output. Similarly, when gender, sales, etc. are input for other member IDs, the parameters of estimation model 213 are adjusted so that "y = 0" or "y = 1" is output.

[0068] The estimation model generation unit 203 uses the teacher data prepared in this manner to train the estimation model 213 for each member ID, and generates a trained estimation model 213. In order to generate an estimation model 213 with optimal accuracy, multiple models can be generated by repeating evaluation steps such as preprocessing to prepare teacher data, modeling processing to generate a model using the prepared teacher data, parameter tuning processing to adjust the parameters of the generated model, and comparing estimated values ​​using the generated model with actual measured values, and the model with the highest accuracy can be selected from these as the estimation model 213.

[0069] 5 also illustrates an estimation flow for estimating a score value Y related to consumption behavior from the member information of the member to be estimated using the generated estimation model 213. For example, for member ID=FFF, when gender=M (male), sales=3,000 (yen), etc. are input into estimation model 213, a score value Y=0.8 is output, and for member ID=GGG, when gender=F (female), sales=0 (yen), etc. are input into estimation model 213, a score value Y=0.7 is output. Similarly, when gender and sales, etc. are input for other member IDs, a score value Y=0 to 1 is output.

[0070] The score calculated for each purchasing tendency segment can be called an in-market score, which indicates the likelihood that a certain target member will purchase products included in that segment. The purchasing tendency score indicates the degree of similarity between the consumption behavior of the group of surveyed members who have "purchased" that segment, and the higher the purchasing tendency score, the higher the similarity between the group of surveyed members who have "purchased." Furthermore, the score calculated for each interest segment can be called an affinity score, which indicates the degree of interest a certain target member has in the products, services, or genres included in that segment. The interest score indicates the degree of similarity between the consumption behavior of the group of surveyed members who have "interested" that segment, and the higher the interest score, the higher the similarity between the group of surveyed members who have "interested."

[0071] The profile generation unit 205 extracts estimation target members whose score values ​​output by the estimation model 213 are equal to or greater than a predetermined threshold as high-score members, and generates a profile report from the membership information, purchase information, or value information of the high-score members. For example, 0.8 is set as the threshold for determining a high score, and a group of estimation target members whose score values ​​are 0.8 or greater for a product or service in a certain segment is extracted. A profile report of the group is generated by analyzing the attributes, values, profile image, etc. that the high-score members have in common. The generated profile report is stored in the profile information storage unit 214.

[0072] It is estimated that a group of members with a high score in terms of consumption behavior for a certain segment are likely to behave in a way that indicates that they are interested in or purchasing products or services of that segment. Therefore, it is estimated that a profile report of a group of high-scoring members represents the purchasing tendencies and interests of potential customers regarding products or services of that segment.

[0073] Figure 6 shows an example of a profile report. A profile report is generated for each segment and includes, for example, basic information (1), basic information (2), value clusters, and a profile image. Basic information (1) of the profile report shows, for example, a graph by gender and age of high-scoring members who are likely to purchase products or services from a certain segment. The left side shows a graph by gender and age of all members subject to estimation, and the right side shows a graph by gender and age comparing all members subject to estimation with high-scoring members with high scores. Basic information (1) displays the analysis result that "85% of high-scoring members are women, and the majority are women in their 30s at 29% (19% higher than the overall comparison)."

[0074] Basic information (2) in the profile report shows graphs comparing all members targeted for estimation with high-scoring members for each occupational category and marital category. The occupational category graph compares all members targeted for estimation with high-scoring individuals for categories such as company employee / civil servant, temporary / part-time worker, self-employed, full-time housewife, student, pensioner, other, and unspecified. The analysis results show that "high-scoring individuals are more likely to be temporary / part-time workers (+3% compared to the overall figure)." Similarly, the marital category graph compares high-scoring individuals with all members targeted for estimation for categories such as unmarried, married, and unspecified.

[0075] The value cluster (PERSONA) in the profile report in Figure 6 shows the values ​​held by all members subject to estimation and high-scoring individuals. Value clusters (PERSONA) indicate the typical consumer types for products and services, and Figure 6 divides consumers into clusters 1 (CL1) to 15 (CL15). Consumer types CL1 to CL15 are established, including the discerning leader type (CL1), the information-sensitive trend leader type (CL2), the safe status quo maintainer type (CL3), the interference-averse individualist type (CL4), and the earnest hard worker type (CL5). The graph on the right of the value cluster in Figure 6 compares all members subject to estimation with high-scoring members for each of the CL1 to CL15 clusters, showing the proportion of members belonging to each cluster.

[0076] Figure 6 shows the value cluster (PERSONA) as an example of a value cluster, but other value clusters such as value cluster (FINANCE) and value cluster (FOOD) can also be presented. The value cluster (FINANCE) classifies consumer types regarding financial products, such as those who leave financial products to experts (CL1), those who are not interested in investments or financial products (CL2), those in the working generation who aim to build assets (CL3), and those who are conservative and reluctant to use financial products (CL4). The value cluster (FOOD) classifies consumer types regarding food, such as those who prioritize health and safety as a food preference (CL1), those who enjoy trendy gourmet foods (CL2), those who eat at their own pace and in their own style (CL3), those who enjoy eating out freely (CL4), and those who have a high appetite and a behavioral gap (CL5).

[0077] The profile image in Figure 6 shows a typical person who is likely to engage in consumption behavior toward that segment by extracting information common to high-scoring members. The person image includes items such as basic attributes, lifestyle values, communication / media, and consumer awareness / behavior. Basic attributes include information on age, annual income, occupation, spouse, and residence, and this information can be used to extract the gender and age group, occupation category, and marital status in which high-scoring members are concentrated in basic information (1) and (2). In addition, for lifestyle values, communication / media, and consumer awareness / behavior, clusters in which high-scoring members are concentrated can be generated by aggregating information previously recorded in the values ​​table.

[0078] (4-2) Member terminal 3 and customer terminal 4 The member terminal 3 according to this embodiment has a function for using the service program in which the member participates, as well as a function for receiving a questionnaire survey sent from the life action estimation device 2, inputting responses to the questionnaire, and transmitting the responses to the life action estimation device 2. The customer terminal 4 is an information processing device that has a function for applying to the life action estimation device 2 for a service that implements life action estimation, or for receiving a profile report that is the result of the life action estimation.

[0079] (5) Operation of Life Action Estimation System 1 7 is a flowchart for explaining an example of the overall operation of the life action estimation system 1 of this embodiment, and illustrates an estimation model generation flow for generating the estimation model 213 and an estimation flow for executing estimation. In the estimation flow, estimation is performed using the estimation model 213 generated in the estimation model generation flow.

[0080] In the estimation model generation flow, the life action estimation device 2 acquires survey results for survey target members (step S51). The life action estimation device 2 randomly extracts a specified number of survey target members from the members recorded in the member table and generates a list of survey target members. The life action estimation device 2 acquires the contact information (email addresses, etc.) of the survey target members from the member table, sends a pre-prepared questionnaire to the member terminals 3 of the survey target members, and receives survey responses from the survey target members. The questionnaire includes questions about consumption behavior such as purchasing behavior and interest behavior for each segment to be surveyed, and members respond to each question with a binary value regarding consumption behavior, such as "made a purchase / did not purchase" or "interested / not interested." The life action estimation device 2 receives these responses from members as the survey results.

[0081] The life action estimation device 2 acquires the member information of the member to be surveyed (step S52). The life action estimation device 2 acquires the member table and purchase table from the member information storage unit 212, and uses the member ID of the member to be surveyed to extract attribute information such as the member's gender from the member table, extracts purchase data such as the unit price and amount of goods or services related to the segment from the purchase table, and calculates the sales amount for the segment.

[0082] The life action estimation device 2 generates an estimation model 213 using the member survey results acquired in step S51 and the member information and purchase information acquired in step S52 as training data. The life action estimation device 2 inputs the member information and purchase information of the member and trains the estimation model 213 so as to output the survey results of that member. The life action estimation device 2 generates multiple estimation models 213 for multiple pieces of training data, and after repeated tuning and evaluation, selects the estimation model 213 that has obtained the optimal evaluation and sets it as the estimation model 213 for the estimation stage.

[0083] In the estimation flow, the life action estimation device 2 selects members to be estimated and acquires member information for the members to be estimated (step S54). As in the case of members to be surveyed, the life action estimation device 2 acquires the member table and purchase table from the member information storage unit 212, and using the member ID of the member to be estimated, extracts member attribute information (gender, etc.) from the member table, extracts purchase data (quantity, amount, etc.) of products or services related to the segment from the purchase table, and calculates the sales amount for the segment.

[0084] The life action estimation device 2 estimates the score of the estimation target member using the estimation model 213 generated in the estimation model generation flow (step S55) and acquires the score (step S56). The life action estimation device 2 calculates a score for each estimation target member for all segments (including major and medium classification segments) in association with the member ID.

[0085] The life action estimation device 2 uses a score table (not shown) that associates the member IDs and scores of members to be estimated to extract, for each segment, members with a score of 0.8 or higher as high-scoring members, and generates profiles of the high-scoring members for each segment (step S57). The life action estimation device 2 uses the member IDs of the high-scoring members to extract basic information to be included in the profile report from the member table. For example, data on the gender and age of all members and high-scoring members is extracted from the member table to create basic information (1). Furthermore, data on the occupation and marital status of all members to be estimated and high-scoring members is extracted from the member table to create basic information (2). Furthermore, the life action estimation device 2 generates value clusters by extracting cluster information from the value table using the member IDs. The life action estimation device 2 also extracts communication / media data and consumer attitude / behavior data from the value table and synthesizes the basic information (1) and (2) and the value clusters to create a profile image.

[0086] When a profile report for a specific segment is requested from the customer terminal 4, the life action estimation device 2 sends the profile report for that specific segment to the customer terminal 4. Based on the profile report, the customer can visualize potential customers for a specific product or service, enabling them to approach consumers in an appropriate manner. Furthermore, since the potential customers to be approached are clearly identified, promotions can be carried out through communication media such as apps, email, and direct mail by post, or through digital advertising media. In this way, it becomes possible to approach customers using reach methods tailored to the target.

[0087] As described above, the life action estimation system 1 sets segments that cover the entire market, executes the estimation model generation flow and estimation flow for each of these segments, and prepares a profile report on the life actions of members, thereby enabling quick and easy provision of member profile reports for each segment. Customers who receive a profile report can estimate the consumption behavior and trends of potential customers for products and services in a certain segment by analyzing the profiles of members with high scores based on the profile report for the desired segment. Furthermore, appropriate marketing approaches can be adopted for estimated potential customers.

[0088] Second Embodiment The life action estimation device 2 in the first embodiment generates an estimation model 213 for each segment, extracts high-scoring members for each segment using the estimation model 213, and generates profiles of the high-scoring members. Meanwhile, consumers often engage in multiple consumption behaviors from when they become aware of a product or service to when they purchase it, such as awareness of the product or service, interest, desire, memory, and action (purchase). For example, by analyzing consumption behavior at multiple times, such as interest in a certain product or service and purchase, it may be possible to estimate potential customers with higher accuracy.

[0089] The life action estimation device 2 according to the second embodiment is an information processing device for estimating potential customers based on multiple consumption behaviors, generating an estimation model for each of the multiple consumption behaviors, estimating scores for the multiple consumption behaviors using each estimation model, and generating profiles of high-scoring members based on the scores for the multiple consumption behaviors. The multiple consumption behaviors may be, for example, consumption behaviors that occur at different times among time-series purchasing behaviors such as cognition, interest, desire, memory, and behavior (purchase).

[0090] Fig. 8 is a diagram showing an example of an information processing flow of the life action estimation system according to the second embodiment. As shown in Fig. 8, the life action estimation device 2 according to the second embodiment executes a program construction flow and a distribution flow. In the second embodiment, two consumption behaviors, namely, intention to move and actual moving results, will be described as examples.

[0091] In the program construction flow, the life action estimation device 2 executes the analysis of basic data and program construction. In the analysis of basic data, as in the first embodiment, survey target members are selected and two types of basic data are created for the survey target members. One type of basic data is the binary response result of "intent to move / not intending to move" obtained from a questionnaire survey of the survey target members. The other type of basic data is the binary performance result data of "actually moved / not having moved" obtained from the usage history of moving companies. The usage history of moving companies can be obtained from the purchase information of the survey target members.

[0092] The life action estimation device 2 then uses two types of basic data to construct a questionnaire-based estimation program and a relocation record-based estimation program. The questionnaire-based estimation program is a program that inputs membership information and purchase information of surveyed members, generates a questionnaire-based estimation model trained to output survey results that the surveyed members "intend to relocate," and estimates a relocation intention score using the questionnaire-based estimation model. On the other hand, the relocation record-based program is a program that inputs membership information and purchase information of surveyed members, generates a relocation record-based estimation model trained to output survey results that the surveyed members "have moved," and estimates a relocation record score using the relocation record-based estimation model.

[0093] In the distribution flow, the life action estimation device 2 performs analysis of members to be estimated, extraction of high-scoring members, distribution of promotional emails, and profile output.

[0094] In analyzing the members to be estimated, the life action estimation device 2 uses the questionnaire-based estimation program and the relocation record-based estimation program constructed in the program construction flow to obtain two types of scores for the members to be estimated, namely, a relocation intention score and a relocation record score.The life action estimation device 2 selects members to be estimated from the members and obtains a relocation intention score and a relocation record score by inputting the member information and purchase information of the members to be estimated into the questionnaire-based estimation program and the relocation record-based estimation program, respectively.Next, members with a high relocation intention score and members with a high relocation record score are extracted from the members to be estimated.

[0095] The life action estimation device 2 can select an approach to potential members based on the relocation intention score and the relocation record score. For example, a promotional email can be sent to potential customers assumed to be from a group of members with both a high relocation intention score and a high relocation record score. Also, a promotional email can be sent to potential customers assumed to be from a group of members with a low relocation intention score but a high relocation record score. The timing and content of each promotional email can be determined depending on the assumed potential customer. For example, for potential customers assumed to be from a group of members with a high relocation intention score and a high relocation record score, promotional emails can be sent both at the timing of "interest," which determines the intention to relocate, and at the timing of "action (purchase)," which determines the actual relocation. On the other hand, for potential customers assumed to be from a group of members with a low relocation intention score but a high relocation record score, promotional emails can be sent at the timing of "action (purchase)," which is a consumption behavior.

[0096] The life action estimation device 2 can generate profile reports of high-scoring members based on the relocation intention score and the relocation record score, separate from the promotional email distribution. For example, a profile report can be generated for each group of members with high relocation intention scores and a group of members with high relocation record scores. Alternatively, a profile report can be generated for a group of members with both high relocation intention scores and high relocation record scores. Alternatively, a profile report can be generated for a group of members with low relocation intention scores but high relocation record scores.

[0097] When selecting members with high relocation intention scores and high relocation record scores, for example, members whose relocation intention score and relocation record score are equal to or greater than a threshold of 0.8 may be selected as high-scoring members. Alternatively, the thresholds for the relocation intention score and the relocation record score may be set to different values, and for example, if emphasis is placed on relocation record, the threshold for the relocation record score may be set to a higher value.

[0098] According to the second embodiment, potential customers are estimated based on a plurality of consumption behaviors, so that it is possible to adopt a variety of approaches to potential customers based on a combination of consumption behaviors. Furthermore, when potential customers are estimated based on a plurality of consumption behaviors occurring in a time series, it is possible to effectively implement an approach that matches the timing of the consumption behaviors.

[0099] <Third embodiment> In the first embodiment, for each segment, members with high scores in consumption behavior among the members subject to estimation were selected as a high-score member group, and basic information, values, and profiles of the high-score member group were generated. However, consumers may be interested in products and services not only in one segment, but in multiple related segments. For example, consumers who are interested in beauty and cosmetics may also be interested in accessories and jewelry.

[0100] In the third embodiment, high-score members common to multiple segments are extracted as common high-score members, and basic information, values, and profiles of the common high-score members are generated. For example, estimation target members who are included in both the high-score members in the beauty / cosmetics segment and the high-score members in the accessories / jewelry segment are extracted as common high-score members, and a profile report of the common high-score members is provided. This profile report can be used to estimate potential customers who are interested in or have the potential to purchase products and services in both beauty / cosmetics and accessories / jewelry. Multiple segments are not limited to two segments; any combination of segments can be used. For example, when estimating potential customers related to segments A to C, segments A to C can be combined using any logical formula, such as (segment A + segment B) * segment C. The same applies when there are three or more segments.

[0101] According to the third embodiment, by utilizing the information of high-scoring members for each segment in the first embodiment, it is possible to generate profiles of high-scoring members for any combination of multiple segments, and to use these profiles to estimate potential customers.

[0102] The present disclosure is not limited to the above-described embodiments, and the components can be modified and embodied in practice without departing from the spirit of the present disclosure. Furthermore, various inventions can be formed by appropriately combining multiple components disclosed in the above-described embodiments. For example, some components may be omitted from all the components shown in the embodiments. Furthermore, components from different embodiments may be appropriately combined. [Explanation of symbols]

[0103] 1...Life action estimation system, 2...Life action estimation device, 3...Member terminal, 4...Customer terminal, 5...Network, 21...Processor, 22...Memory, 23...Storage, 24...Communication I / F, 25...Input / output I / F, 26...Bus, 31...Processor, 32...Memory, 33...Storage, 34...Camera, 35...Communication I / F, 36...Input / output I / F, 37...Touch screen, 38...Input unit, 39...Display unit, 200...Management unit, 201...Survey result acquisition unit, 202...Member information acquisition unit, 203...Estimation model generation unit, 204...Estimation unit, 205...Profile generation unit, 210...Memory unit, 211...Survey information storage unit, 212...Member information storage unit, 213...Estimation model, 214...Profile information storage unit

Claims

1. a survey result acquisition unit that acquires survey results regarding the consumption behavior of survey target members; an estimation unit that receives member information of the survey target member and estimates a score related to the consumption behavior based on the member information of the survey target member using an estimation model that has been trained to output the survey results; a profile generation unit that generates a profile of the estimation target member based on the score; An information processing device comprising:

2. The information processing device according to claim 1 , wherein the survey results regarding the consumption behavior of the surveyed members include responses regarding whether or not the surveyed members have engaged in consumption behavior, collected through a questionnaire survey of the surveyed members.

3. The information processing device according to claim 1 , wherein the member information includes at least attribute information and purchasing information of the member.

4. The information processing device according to claim 1 , wherein the consumer behavior includes a purchase of a product or service that is the subject of the survey, or an interest in the product or service.

5. The information processing device according to claim 2 , wherein the score indicates a degree of similarity between the consumption behavior of the estimation target member and the consumption behavior of a group of survey target members who responded in the questionnaire survey that they have made consumption actions.

6. The information processing device according to claim 1 , wherein the profile is generated for each product or service segment based on at least member information of members whose scores are equal to or greater than a predetermined threshold.

7. Obtaining survey results regarding the consumption behavior of surveyed members; inputting member information of the survey target member and using an estimation model trained to output the survey results, estimating a score related to the consumption behavior based on the member information of the estimation target member; generating a profile of the estimated member based on the score; An information processing method comprising:

8. A computer program that causes a processor to execute the processing of each unit included in the information processing device according to claim 1 .

Images