JP2019212126A - Sales support system, sales support method, and sales support program - Google Patents

Abstract

To enable an effective introduction of products of categories customers do not yet purchase.SOLUTION: A sales support system comprises: clustering means that clusters blue-chip customers, who have already purchased more than the prescribed number of products of categories, into a plurality of clusters; purchase order determination means that determines a representative category purchase order as to the plurality of clusters; customer classification means that determines the cluster a target customer belongs to from the plurality of clusters; recommendation category determination means that determines any of the prescribed pieces of categories from the descendent order in the representative category purchase order of the cluster the target customer belongs to of categories the target customer has not yet purchased as a recommendation category; and offer means that offers information about the product to be included in the recommendation category to the target customer.SELECTED DRAWING: Figure 3

Description

  The present invention relates to a sales support system, a sales support method, and a sales support program.

  In recent years, in electronic commerce using the Internet or the like, product information is provided by various methods (Patent Document 1, etc.). For example, there is a method of guiding products that the user is expected to purchase from the past purchase history of the user. There is also a method of guiding products that are often purchased at the same time as the products that the user intends to purchase.

Japanese Patent Laid-Open No. 2006-45901

  The above-described method can be said to be a technique aimed at increasing purchase points and profits from a short-term perspective. Even if such a product guide is adopted, the user does not always continue to purchase at the same site. In addition, the above-described method often introduces products in categories that are the same as or similar to the user's purchased products, and thus has a problem that it does not lead to the purchase of products in other categories.

  Furthermore, considering that there are various online shopping sites, it becomes a long-term profit increase to become a shopping site that can be used regardless of the category of goods purchased.

  In view of the above circumstances, an object of the present invention is to provide a sales support technique for performing effective product guidance so that a new category of product can be purchased.

One aspect of the present invention is a sales support system,
A clustering means for classifying high-quality customers who have purchased products of a predetermined number or more categories into a plurality of clusters,
A purchase order determining means for determining a representative category purchase order for the plurality of clusters;
Customer classification means for determining a cluster to which the target customer belongs from the plurality of clusters;
Recommended category determination means for determining, as a recommended category, one of a predetermined number of categories from the earlier order in the representative category purchase order of the cluster to which the target customer belongs among the categories to which the target customer has not been purchased;
Providing means for providing information related to products included in the recommended category to the target customer;
It is characterized by providing.

  In this aspect, the clustering means may classify the good customers into a plurality of clusters by using a category purchase order that indicates in what order the products of the new category have been purchased.

  In this aspect, the purchase order determining means obtains an average of the purchase order of each category from the category purchase order of excellent customers belonging to each cluster, and determines a representative category purchase order of each cluster based on the average. Also good.

  In this aspect, the apparatus may further include a recommended product determination unit that determines a recommended product that is a product that provides information to the target customer from products included in the recommended category based on a purchase rate of the product. Good. Here, the product purchase rate may be a value calculated for all customers, a value calculated for good customers, or a good customer belonging to the same cluster as the target customer. It may be a value calculated for.

  The present invention can also be understood as a sales support system including at least a part of the above means. The present invention can also be understood as a sales support method executed by a computer. The present invention can also be understood as a computer program for causing a computer to execute this method, or a computer-readable storage medium in which this computer program is stored non-temporarily. Each of the above means and processes can be combined with each other as much as possible to constitute the present invention.

For example, a sales support method according to an aspect of the present invention includes:
A sales support method performed by a computer,
Categorizing good customers who have purchased more than a predetermined number of categories into multiple clusters;
Determining a representative category purchase order for the plurality of clusters;
Determining a cluster to which the target customer belongs from among the plurality of clusters;
Determining, as a recommended category, one of a predetermined number of categories from the earlier order in the representative category purchase order of the cluster to which the target customer belongs among the categories to which the target customer has not been purchased;
Providing the target customer with information related to the product included in the recommended category;
It is characterized by including.

  A program according to another aspect of the present invention is a program for causing a computer to execute each step of the above-described sales support method.

  ADVANTAGE OF THE INVENTION According to this invention, the object customer can guide effectively the goods of the category which has not been purchased.

FIG. 1 is a schematic configuration diagram of an electronic commerce system according to an embodiment. FIG. 2 is a diagram illustrating a hardware configuration of the recommendation server according to the embodiment. FIG. 3 is a block diagram illustrating a functional configuration of the recommendation server. FIG. 4 is a flowchart showing a flow of clustering processing of excellent customers performed on the recommendation server. FIG. 5 is a flowchart showing the flow of the product guidance process performed by the recommendation server. FIG. 6 is a diagram illustrating an example of purchase history information. FIG. 7 is a diagram illustrating an example of a representative category purchase order for each cluster of excellent customers. FIG. 8 is a diagram schematically illustrating processing for determining a cluster to which a target customer belongs. FIG. 9 is a diagram for explaining processing for determining a category of a product to be guided to the target customer.

  As a result of analyzing purchase information for each user on the online shopping site, the present inventors have found that sales by users who purchase products of many categories occupy most of the total sales. For example, out of about 500 types of medium categories, sales by users who purchase products of about 50 or more categories account for about half of sales, and users who purchase products of about 20 or more and less than 50 categories Sales account for the majority of the rest. Thus, it can be said that users who purchase products in many categories are excellent customers who use one shopping site for almost all online shopping.

  It takes a considerable amount of time from the start of use of the shopping site to becoming a good customer (for example, a customer who has purchased 50 or more types of products in the middle category is defined as a good customer). Therefore, by shortening the time from the start of use to becoming a good customer and promoting the creation of good customers, it is possible to secure good customers and increase sales.

  The present inventors further analyzed the purchase behavior of a good customer and determined in what order the good customer purchased a product of a new category (hereinafter also referred to as a “category purchase order”). We found that the purchase order can be classified into a few patterns. That is, the category purchase order from the start of service use to the excellent customer can be patterned. Therefore, if it becomes clear which pattern a customer other than a good customer corresponds to, it can be considered that a good customer can be promoted by guiding products according to the purchase order of the relevant pattern.

  Based on such a study, the present inventors guide excellent products that are likely to lead to purchase among products in categories that have not been purchased by the user. I thought that I could promote

  Hereinafter, details of the embodiment will be described with reference to the drawings. However, the embodiment described below is merely an example, and the sales support system, the sales support device, the sales support method, and the sales support program according to the present disclosure are not limited to the specific configurations described below. Moreover, in implementation, the specific structure according to embodiment is employ | adopted suitably and various improvement and deformation | transformation may be performed.

<Embodiment>
[System configuration]
FIG. 1 is a schematic configuration diagram of an electronic commerce system according to an embodiment. As shown in FIG. 1, the electronic commerce system 1 includes a shopping server 100, a database 101, and a recommendation server 200. The electronic commerce system 1 is connected via a network N to a terminal 102 that uses the electronic commerce system 1. Although FIG. 1 shows an example in which three terminals 102 are connected, the number of connected terminals 102 is not limited. The network N is, for example, a worldwide public packet communication network such as the Internet, and a WAN (Wide Area Network) or other communication network may be employed.

  Shopping server 100 (hereinafter also simply referred to as server 100) provides a customer with a commercial transaction site for realizing electronic commerce. As basic functions of the commercial transaction site, there are merchandise information provision processing, merchandise search processing, and order processing. The product information providing process includes a form in which a user provides information on a desired product by following a link or searching, and a form in which information on a product determined by the recommendation server 200 is provided.

  The database 101 is connected to the server 100. The database 101 stores various types of information used for electronic commerce executed between the server 100 and the terminal 102 in the present embodiment. The database 101 stores, for example, purchase history information for each customer. The database 101 also stores information on categories for classifying products on the commercial transaction site. Note that the database 101 may be constructed in an auxiliary storage device provided in the server 100 or may be constructed in another computer connected to the server 100. Hereinafter, the database 101 will be described as being built in the auxiliary storage device included in the server 100. The database 101 is an example of a “storage unit”.

  The recommendation server 200 determines a product to be guided to the customer. The recommendation server 200 has a function for classifying excellent customers into several clusters, a function for determining a representative category purchase order for each cluster, a function for determining which of the clusters the target customer belongs to, and a target. A function of determining a recommended category based on a typical category purchase order of a cluster to which the customer belongs and a category not yet purchased by the target customer, and a function of determining a recommended product from the recommended category are provided. The recommendation server 200 is an example of a “sales support system”.

  Although FIG. 1 shows an example in which the shopping server 100 and the recommendation server 200 are implemented as different servers, the functions of these servers may be realized by one server. Conversely, the functions of the shopping server 100 and the recommendation server 200 may be realized by a plurality of servers.

[Constitution]
FIG. 2 is a diagram illustrating a hardware configuration of the recommendation server 200 according to the embodiment. As shown in FIG. 2, the server 200 includes a CPU (Central Processing Unit) 200a.
RAM (Random Access Memory) 200b, HDD (Hard Disk Drive) 200c, N
An IC (Network Interface Card) 200d is provided.

The CPU 200a is a central processing unit, and controls the RAM 200b, the HDD 200c, and the like by processing instructions and data of various programs developed in the RAM 200b and the like. The CPU 200a is not limited to a single processor, and may have a multiprocessor configuration. A single CPU connected by a single socket may have a multi-core configuration. At least a part of the processing of each unit is provided by a DSP (Digital Signal Processor), GPU (Graphics Processing Unit), a numerical processor, a vector processor, a dedicated processor such as an image processor, an ASIC (Application Specific Integrated Circuit), or the like. Also good. In addition, at least a part of each of the above parts is an FPG.
A dedicated LSI (Large Scale Integration) such as A (Field-Programmable Gate Array) or other digital circuits may be used. In addition, an analog circuit may be included in at least a part of each of the above parts.

  The RAM 200b is a main storage device, and is controlled by the CPU 200a to write and read various commands and data. The HDD 200c is a non-volatile auxiliary storage device, and information required to be persistent such as various programs loaded in the RAM 200b is written and read out. The NIC 200d is an interface for connecting to the Internet via a gateway or the like.

FIG. 3 is a block diagram illustrating a functional configuration of the recommendation server 200 according to the embodiment. The CPU 200a of the recommendation server 200 reads out and executes the computer program stored in the HDD 200c, thereby realizing the functions shown in FIG. That is, the recommendation server 200 includes a purchase history storage unit 201, a good customer extraction unit 202, a clustering unit 203, a category purchase order determination unit 204, a cluster information storage unit 205, a customer classification unit 206, a recommended category determination unit 207, and a recommended product determination. Unit 208 and product information providing unit 209 function. Details of these functional units will be described in detail when the processing content is described. Note that some or all of these functional units may be provided by a dedicated logic circuit such as an ASIC or FPGA instead of being implemented by a CPU and a program.

Since the shopping server 100 is a computer having the same configuration as the recommendation server 200, description thereof is omitted. The terminal 102 is assumed to be, for example, a PC (Personal Computer), but various terminal devices such as a mobile terminal such as a smartphone or a tablet or a stationary information terminal may be employed.

[Processing content]
The process performed by the recommendation server 200 includes a process of clustering excellent customers and a process of determining recommended products for non-excellent customers and providing product guidance. Hereinafter, each will be described with reference to the drawings.

(Clustering process)
FIG. 4 is a flowchart of a process for clustering excellent customers. The clustering process will be described in detail according to this flowchart.

  In step S <b> 101, the good customer extraction unit 202 refers to the purchase history storage unit 201 and extracts purchase history information of good customers.

  The purchase history storage unit 201 stores history information related to products purchased by each customer in the past. FIG. 6 is a diagram illustrating an example of purchase history information for a certain customer. As illustrated in FIG. 6, the purchase history storage unit 201 stores, for each customer, the purchase date of the product, the product ID of the purchased product, the large category / medium category / small category of the product category, the product name, and the like. In addition to these pieces of information, the purchase amount and purchase quantity may be stored. Further, information related to a product such as a product category may be stored in the product master DB in association with the product ID and referred to the product master DB.

  The good customer extraction unit 202 refers to the purchase history storage unit 201 to identify a good customer and extract the purchase history information. In the present embodiment, a customer who has purchased a predetermined number or more types of products in the middle category unit is regarded as a good customer. For example, when the total number of medium categories is 500, the “predetermined number” is 50, but the specific value may be appropriately determined according to the system requirements. Although the focus is on the middle category here, the focus may be on the large category or the small category.

  In step S102, the clustering unit 203 determines a category purchase order for a good customer. In the present embodiment, the category purchase order is an order representing the order in which products of a new category (in this case, the middle category) have been purchased since the start of use of the commercial transaction site. For example, if a customer's purchase history is as shown in FIG. 6, the category purchase order of this customer is the order of “copy paper / printer paper”, “marker”, “storage product”, and “detergent”. The clustering unit 203 determines the category purchase order for all excellent customers.

In step S103, the clustering unit 203 clusters good customers based on the category purchase order. The clustering method is not particularly limited, and any method may be employed. The clustering unit 203 performs clustering by, for example, the k-means method. In the k-means method, it is necessary to give the number of cluster divisions k, but this value may be determined as appropriate. In the present embodiment, for example, the number of cluster divisions may be five. Since the k-means method itself is a known method, detailed description thereof is omitted.

  In addition to the k-means method, the clustering unit 203 may employ other clustering algorithms such as the shortest distance method (Nearest neighbor method), the k-NN method, the longest distance method, the Ward method, and the group average method. .

  In step S104, the clustering unit 203 calculates the center of gravity of each cluster obtained in step S103. The cluster centroid is obtained, for example, as the position of the centroid in the feature amount space of the feature points belonging to a certain cluster by representing the category purchase order of good customers by the points in the feature amount space.

  In step S105, the category purchase order determination unit 204 obtains a representative category purchase order for each cluster. For example, the category purchase order determination unit 204 obtains an average value of the purchase order of each middle category for each cluster, and determines a representative cluster purchase order for the clusters according to this average order.

  In step S106, the cluster centroid obtained in step S104 and the representative cluster purchase order obtained in step S105 are stored in the cluster information storage unit 205.

  FIG. 7 shows a typical category purchase order of each cluster obtained as a result of the clustering process. As a result of clustering, high-quality customers having similar purchase order tendencies are collected, and a typical category purchase order is obtained.

(Product information processing)
FIG. 5 is a flowchart of a product guidance process for determining a product recommended for a non-excellent customer and guiding the product. The product guidance process will be described in detail according to this flowchart.

  Prior to the start of the process of FIG. 5, it is assumed that the non-excellent customer to whom product guidance is to be made has already been determined. Further, the timing at which the process of FIG. 5 is executed is not particularly limited. In this embodiment, when the user accesses the shopping server 100, if the user is a non-excellent customer, the recommendation server 200 executes the product guidance process based on an instruction from the shopping server 100.

  In step S <b> 201, the customer classification unit 206 acquires the purchase history of the target non-excellent customer (simply referred to as the target customer) from the purchase history storage unit 201. Here, if the target customer has no purchase record (S202-NO), since the target customer cannot be classified, the product guidance process based on the category purchase order is stopped. In this case, the guide product may be determined by another method. On the other hand, when the target customer has a purchase record (S202-YES), the process proceeds to step S203.

  In step S203, the customer classification unit 206 obtains the category purchase order of the target customer and determines which cluster the target customer belongs to based on the category purchase order. For example, as schematically illustrated in FIG. 8, the customer classification unit 206 sets a cluster 83 having a cluster centroid 82 closest to the position 81 in the feature amount space of the cluster purchase order of the target customer as a cluster to which the target customer belongs. decide. If classification is performed based on the category purchase order of the target customer and the similarity between the category purchase order of the clusters, the cluster to which the target customer belongs may be determined by other methods.

  In step S204, the recommended category determination unit 207 determines a medium category of products to be guided to the target customer based on the category purchase order of the target customer and the representative category purchase order of the cluster to which the target customer belongs. For example, the recommended category determination unit 207 may determine, as the recommended category, a category in which the target customer has not been purchased and the category category in which the target customer belongs has the earliest category purchase order.

  For example, as in the example shown in FIG. 9, the typical purchase order of the cluster to which the target customer belongs is as shown in the list 91, and the target customers are the categories “category A”, “category B”, “category C” Consider the situation where you have already purchased the product. In this case, “category D”, which is the category with the earliest order in the representative purchase list 91 for which the target customer has not been purchased, is selected as the recommended category.

  Note that the category with the earliest purchase order does not necessarily have to be the recommended category, and the category selected on the basis of an appropriate criterion from the categories within the predetermined number (for example, three) from the earliest purchase order is the recommended category. Also good. As an example of the selection criterion here, it is conceivable to select a category having the highest purchase price or the number of purchases by excellent customers in the same cluster. Alternatively, it may be determined randomly from a predetermined number of categories.

  In step S205, the recommended product determination unit 208 determines a product to be guided to the target customer from among the products belonging to the determined recommended category (medium category). For example, the recommended product determination unit 208 can randomly select one or more products with a high purchase rate from the recommended categories and determine them as recommended products. The purchase rate here may be a value for good customers in the same cluster, or may be a value for all customers in the site.

  In step S <b> 206, the product information providing unit 209 acquires information related to recommended products from the database 101 and provides the server 100 with information. Thereby, the server 100 can present the guidance regarding the recommended product to the target customer on the commercial transaction site.

[Advantageous effects of this embodiment]
According to this embodiment, good customers are classified into several clusters, and it is determined which cluster the purchase behavior of a non-excellent customer (target customer) is closest to. Then, referring to the category purchase order of the nearest cluster, the target customer recommends (guides) a product in a category that has not been purchased. Since the target customer is expected to perform the same purchase behavior as a good customer of a similar cluster, it can be expected that the target customer has a relatively high probability of purchasing the recommended product. As described above, according to the present embodiment, it is possible to effectively advertise a product in a category that has not been purchased by the target customer and connect it to purchase.

  Further, customers who purchase many categories at the same shopping site are customers with high loyalty to the shopping site, and it can be expected that they will continue to purchase stably at the shopping site in the future. With the recommendation function of the present embodiment, it is possible to shorten the time required for a customer to become an excellent customer, and an increase in sales can be expected when viewed from a shopping site.

<Modification>
[Modified example of clustering process for excellent customers]
In the above embodiment, clustering of excellent customers is performed based on the category purchase order of excellent customers, but other information may be focused on when performing clustering processing. For example, when a new category is purchased, clustering may be performed in consideration of the date and time since the start of site use. Further, the present invention is not limited to the purchase order of new categories, and clustering may be performed based on the purchase order of all categories while allowing duplication.

  Further, clustering may be performed in consideration of information other than the category purchase order. As other information, for example, when the customer is a company, attributes such as the type of business and the number of employees may be considered.

[Variation of determining the cluster to which the target customer belongs]
In the above-described embodiment, the attention is paid to the category purchase order in both the clustering process for the excellent customer and the process for determining the cluster to which the target customer belongs. However, when determining which cluster the target customer belongs to, the processing may be performed based on information different from the information focused on clustering good customers.

  For example, first, good customers are classified into clusters based on the category purchase order. Then, obtain information (features) other than the category purchase order for good customers in the same cluster, find the centroid in the feature space, and find the centroid closest to the same kind of information (feature) on the target customer. The cluster that the customer has may be determined as the cluster to which the target customer belongs. As information that can be used other than the category purchase order, information similar to that described in the above-described modification of the clustering process can be used.

[Modification of Product Guidance Process Execution Timing and Guidance Method]
In the above-described embodiment, an example has been described in which when a non-excellent customer accesses a shopping site, the product guidance process shown in FIG. 5 is executed, and the product is guided on the shopping site. However, among the product guidance processing shown in FIG. 5, the recommended product determination processing may be executed before the user accesses the shopping site. For example, the recommended products may be periodically determined for all non-excellent customers and stored in the storage unit.

  Also, presentation of information regarding recommended products is not necessarily performed on a shopping site. For example, you may perform by other methods, such as a guidance advertisement by electronic mail.

[Target customers]
A condition other than the condition of being a non-excellent customer may be imposed on a customer who provides product information according to the above embodiment. For example, a condition that the number of purchased middle categories is a threshold value (for example, 3) or more may be imposed. This is because, when the number of purchased categories is large, the determination accuracy of which cluster the target customer belongs to becomes high, and the recommendation of products based on the representative purchase order of the clusters becomes more effective.

  In addition, even if the number of purchased categories of non-excellent customers is greater than or equal to the threshold, if the non-excellent customer has a low degree of similarity in any cluster, it must be a target customer for providing product information by the above method Also good. For example, the case where the purchased category of a non-excellent customer is not a category that is positioned higher in the representative purchase order for any cluster is applicable. In such a case, the determination accuracy of the cluster to which the customer belongs may decrease, and may be excluded.

[Computer-readable recording medium]
A management tool for setting the server in a computer or other machine or device (hereinafter referred to as a computer or the like), a program for realizing an OS or the like can be recorded on a computer-readable recording medium. The function can be provided by causing a computer or the like to read and execute the program of the recording medium.

Here, a computer-readable recording medium is a recording medium that stores information such as data and programs by electrical, magnetic, optical, mechanical, or chemical action and can be read from a computer or the like. Say. Examples of such a recording medium that can be removed from a computer or the like include a flexible disk, a magneto-optical disk, a CD-ROM, a CD-R / W, a DVD, a Blu-ray disc, a DAT, an 8 mm tape, a flash memory, and the like. There are cards. Moreover, there are a hard disk, a ROM, and the like as a recording medium fixed to a computer or the like.

100: Shopping server 101: Database 200: Recommendation server 201: Purchase history storage unit 202: Excellent customer extraction unit 203: Clustering unit 204: Category purchase order determination unit 205: Cluster information storage unit 206: Customer classification unit 207: Recommended category determination Part 208: Recommended Product Determination Unit 209: Product Information Providing Unit

Claims (6)

A clustering means for classifying high-quality customers who have purchased products of a predetermined number or more categories into a plurality of clusters,
A purchase order determining means for determining a representative category purchase order for the plurality of clusters;
Customer classification means for determining a cluster to which the target customer belongs from the plurality of clusters;
Recommended category determination means for determining, as a recommended category, one of a predetermined number of categories from the earlier order in the representative category purchase order of the cluster to which the target customer belongs among the categories to which the target customer has not been purchased;
Providing means for providing information related to products included in the recommended category to the target customer;
A sales support system. The clustering means classifies the good customers into a plurality of clusters by using a category purchase order that indicates in what order the products of the new category are purchased for the good customers.
The sales support system according to claim 1. The purchase order determining means obtains the average purchase order of each category from the category purchase order of excellent customers belonging to each cluster, and determines a representative category purchase order of each cluster based on the average.
The sales support system according to claim 1 or 2. A recommended product determination means for determining a recommended product that is a product that provides information to the target customer from among products included in the recommended category, based on a purchase rate of the product;
The sales support system according to any one of claims 1 to 3. A sales support method performed by a computer,
Categorizing good customers who have purchased more than a predetermined number of categories into multiple clusters;
Determining a representative category purchase order for the plurality of clusters;
Determining a cluster to which the target customer belongs from among the plurality of clusters;
Determining, as a recommended category, one of a predetermined number of categories from the earlier order in the representative category purchase order of the cluster to which the target customer belongs among the categories that the target customer has not purchased;
Providing the target customer with information related to the product included in the recommended category;
Including sales support methods.   The computer program for making a computer perform each step of the method of Claim 5.

Images