JP7228208B2 - Potential customer discovery support system - Google Patents

Description

The present invention uses information obtained by analyzing the trends of existing customers, statistical information such as public census data, and various information related to sales such as map information to obtain values based on lifestyles ( LSV: Life Style Value) is calculated, and the LSV value is used to discover potential customers who are thought to belong to the same category or similar behavior patterns, and to efficiently conduct business activities of financial institutions and companies. The present invention relates to a potential customer discovery support system having electronic information processing logic that supports well.

There are roughly two patterns for financial institutions, companies, etc. (hereinafter simply referred to as "companies") to conduct business activities. One is "additional sales to existing customers" and the other is "development sales to new customers" who are not existing customers. In the case of "additional sales to existing customers", the salesperson can estimate the customer's needs and propose products and services by analyzing transaction records and customer information (age, gender, etc.) as a reference. However, even with existing business partners, the needs of customers for whom the company is not the main customer (customers whose usage is stagnant, etc.) are unlikely to change in transactions or accumulate new information. is more difficult.

Conventional marketing is based on potential customer selection work in the sale of individual products. For that purpose, for example, a search of a customer DB (Data Base) provided in MCIF (Marketing Customer Information File), CRM (Customer Relationship Management), etc. From the function, the potential customer list is cut out as separate work for each product.

FIG. 1 shows a general procedure of conventional marketing. First, the customer DB of MCIF or CRM is accessed (step S1), and target products are selected (step S2). Then, using search items such as gender, age, liquidity average balance (average balance), payment (salary transfer), etc. (step S3), target customers are listed (step S4). The listed target customers are appended with shop CIF (customer information file of shop: Customer Information File), account name, address, telephone number, e-mail address, and the like.

Next, a method of contacting the listed customers is selected (step S5). The contact methods include DM (Direct Mail), posting, insert advertisement, e-mail, call center (CC), and ATM (Automatic Teller Machine). , external visits, etc. After selecting the optimum contact method for the customer from among these, the result after executing the contact method is confirmed (step S6), and the PDCA after execution (plan-do-check ( (evaluation)-action (improvement)) and marketing is completed (step S7). For example, it is generally said that the hit rate (ratio of actual product purchases to the number of DM orders) indicating the results after DM is about 1% of the number of orders.

Since addresses, names, etc., are registered in a customer DB such as MCIF, the output of this marketing method is the names of individual customers with whom transactions are made. Therefore, various sales activity channels are used for this customer name, and whether or not the search item is an item that appropriately selects the target customer depends on the DM hit rate. is normally around 1%. In addition, since the search items themselves largely depend on the know-how of the salesperson who is the searcher, there are many cases where a plurality of salespersons use the same search items. Therefore, even if the PDCA cycle is repeated after contacting, there is very little room for improving marketing efficiency. Furthermore, since this marketing method is performed for each product, the same process is repeated for each product, resulting in inefficiency.

PDCA is widely known as a continuous improvement method in business management such as quality control, and by repeating the four stages of Plan → Do → Check → Action , is a common technique for continuous improvement of operations.

JP 2011-096209 A Japanese Patent Application Laid-Open No. 2007-310851 JP 2004-118571 A JP 2020-194288 A Japanese Patent Application Laid-Open No. 2020-149706 JP 2019-179498 A JP 2017-49784 A Japanese Patent No. 6729877

LTV (Customer Lifetime Value) is used in conventional sales activities (steps S1 to S4 in FIG. 1), and the basic formula for obtaining LTV is Equation 1 below.
(Number 1)
LTV = annual transaction amount x rate of return x continuous years (length of stay)

However, this number 1 is rarely used. This is because the LTV is different for each customer, and it is nearly impossible to calculate sales and expenses individually. At a practical level, it is often requested based on the total number of customers (N). For example, number of repeat customers (number 2), customer income (number 3), customer retention cost (number 4), single year profit (number 5), initial investment amount (number 6), total LTV (number 7), per person Find the LTV (Equation 8), etc., and find the average LTV by Equation 9.
(Number 2)
Number of repeat customers = Number of new customers x Customer retention rate
(Number 3)
Customer revenue = customer unit price x number of repeat customers
(Number 4)
Customer retention cost = Number of new or repeat customers x Customer retention cost per person
(Number 5)
Single year profit = customer income - customer retention cost
(Number 6)
Initial investment = Expenses invested to acquire customers in the first year
(Number 7)
Total LTV = Cumulative amount of profits for a single year

(Number 8)
LTV per person = total LTV / number of new customers in the first year (number 9)
Average LTV = Added value of LTV per person/N

In order to maximize the value of LTV, the key is to increase the "customer retention rate" in Equation 2, and if the cost is too high, the "customer retention cost" shown in Equation 4 will increase. Profitability declines.

In addition, LTV has many problems and accompanying conditions as described below, and if these problems are solved, LTV calculation is an effective means, but problems including the economic environment must be solved. For example, LSV (Life Style Value) is calculated to indicate value based on lifestyle, and profiling of the LSV value (analyzing a person's personal information and past behavior to infer future behavior, etc.) and Calculating the potential profit by ranking (ranking) (LSV navigation (Life Style Value navigation) (hereinafter simply referred to as "LSV navigation") is effective for sales support for discovering potential customers. Here, “Potential customers” refer to customers who do not yet feel the need for products or services, or who are not aware of their specific needs. Customer.

LTV refers to the profit that a customer brings to a company throughout his or her lifetime. In general, the higher the customer's attachment to products and services (customer loyalty), the higher the LTV value tends to be. Figure 2 schematically shows general LTV along the time axis, showing the main events (leisure, marriage, home purchase, etc.) from entering the workforce to retirement, and related events It correlates the general relationship with financial needs for The value (transaction amount: value) from entering the working age to old age rises halfway through, and then gradually decreases with age. shown in the flow. Then, as shown in Equation 10, the added value of value and lifetime is LTV, which is the same as Equation 1. The LSV value will be described later, but as shown in FIG. 2, for each event that requires funds (in FIG. 2, a pair of a house purchase and a mortgage loan, etc.), the customer's LSV value at that time is calculate.
(Number 10)
LTV = lifetime + value

The LTV value calculated by Equation 10 is only a pseudo number, and in order to accurately determine the revenue from one customer, it is necessary to accumulate transaction data for decades, but such Transaction data are almost impossible to obtain. Therefore, if it is assumed that people living in the same group follow the same consumer behavior pattern, the accumulated results of transactions with financial institutions of people living in that group area can be considered to be close to the estimated LTV. can. To that extent, the LTV value calculated here is the point-in-time earnings as a cross-section. Usually, when calculating profit, initial investment, calculation of asset value, selling and administrative expenses, etc. are required. After performing these calculations, the financial institution owns the gross interest rate for each product for making a profit forecast. If this gross profit interest rate is presented, then the interest income for the remaining balance can be obtained for each product. However, what we are looking for here is not the interest income of each person, but the sum of all accounts that have accounts in the group area. After that, the characteristics of individual products, for example, the interest rate of a mortgage, varies greatly depending on the loan application period, and this interest rate system is owned by the financial institution, so if you submit the current weighted average interest rate of the mortgage, At that time, the proportion of high-interest housing loans will have a great significance. As it is now, the higher the proportion of low-interest housing loans, the lower the gross interest rate on housing loans, which in turn reduces overall profits.

The present invention has been made in view of the above-mentioned circumstances, and the object of the present invention is to calculate the LSV value that indicates the value based on the lifestyle, and analyze the potential income of each town or chome by profiling, ranking, etc. (LSV Navi), finds potential customers in cooperation with map information, and provides a potential customer discovery support system having electronic information processing logic for supporting business activities of a company. The purpose is to reuse existing customers as new customers for new products, to derive appropriate customers as efficiently as possible, and to use them to contribute to cost reduction in sales activities.

The present invention is a potential customer discovery support system that assists discovery of potential customers by information processing logic. Equipped with an LSV navigation system for inputting lifestyle cluster data, user-owned data of existing customers, and administrative boundary data, a calculation unit and an input/output unit, calculating the LSV value of the existing customer for each town and chome, This is achieved by comparing product contract records for each town and chome, and creating a potential customer list of areas with low product contract records based on the LSV values of areas with high product contract records.

In the present invention, statistical information such as demographic data published in national census etc. is used to group (segment) each town or chome, and each town or chome is based on the lifestyle of each person. The LSV value indicating the value (LSV: Life Style Value) is calculated, and the potential profit including the number of potential customers is calculated using the total number of accounts or the total population of the town or chome. If the LSV value is calculated for each municipality, the target area becomes too wide and it is impossible to calculate the LSV value with high probability. Therefore, in the present invention, the LSV value is calculated for each town or chome.

The potential customer discovery support system using the LSV navigation of the present invention is a technique for finding customers who can trade new products to be offered with low risk and long-term potential. For this reason, the present invention aims to scientifically analyze the components (sources) of the "luck, ability, and personal connections" possessed by the customer on the time axis of the past, present, and future, and to evaluate them.

The components of "luck, ability, and personal connections" are family tree, land, property, family, relatives and friends, land feng shui, mother's genes and luck, ability, and personal connections, assuming a set represented by a finite group. , Father's genes, luck, ability, personal connections, etc. Machi-chome is also one of the components, and limited components such as where you live and family background create the people who live in that machi-chome. It is the idea that a person's "luck, ability, and personal connections" change over time due to the internal components of a person as they grow up.

The items that are the components of "luck, ability, and personal connections" do not have quantities or units, and many of them are expressed in natural language. There is also the problem of whether it is a continuous quantity. The difference between the separated amount and the continuous amount is that the separated amount is to be counted, while the continuous amount is to be measured. "Deposits/savings/real estate/occupation"; Interests", "Purchasing power for ability", and "Age for personal connections" are added and analyzed.

In order to improve conventional inefficient marketing methods, we use public census data and other statistical information to show consumer behavioral characteristics (geographical demographic data) for each area generated by clustering analysis. Style cluster data, a large amount of customer information (CRM/SFA (Sales Force Automation)) owned by companies, administrative boundary data including population and area data, and LSV values based on the lifestyles of existing customers are calculated. The calculation of the LSV value reflects the risk value obtained by analyzing each of the three major risk factors of the existing customers, purchasing power, interest, and age.

By calculating the LSV value considering the risk values based on the three major risk factors of purchasing power, interest, and age, potential customers with high rankings whose LSV values are close to each other are pinpointed sequentially from the top according to the ranking. For selected potential customers, the potential profit (including the number of potential customers) of individual products for each area is calculated, and the sales of individual products with higher probability than before are promoted and supported. It is possible to provide an efficient potential customer discovery support system based on the customer's LSV value to financial institutions and companies that seek to discover potential customers.

The LSV value is a numerical value that indicates the similarity of lifestyles, and the LSV navigation of the present invention uses the LSV value derived by a predetermined arithmetic expression to perform profiling, ranking, etc. flow business (a style of business in which each transaction sells out and generates profits), and stock business (contracts with customers) that provide customer rankings, calculation of town or chome value (potential income), and profiling It is a tool that can be used for a business style that obtains continuous profits by connecting and securing members.

In addition, by displaying the discovered potential customer information on a residential map or a wide area map, it is an effective tool that leads to visualization, such as obtaining regional characteristics for each branch of a financial institution or company. Regional characteristics include population size, population density, regional divisions, climate, and traffic conditions.

A potential customer discovery support system according to the present invention expresses different multifaceted elements as diversity and maps them onto an intuitively understandable Euclidean plane according to the purpose of use. It can also be widely used by companies in other industries such as sales companies. In addition, using statistical information such as national census conducted every five years, consumer behavior characteristics for each area generated by clustering analysis, and customer transactions accumulated in CRM/SFA owned by companies Target customer information obtained from data can be shared and used in other systems.

It is a flow chart for explaining the procedure of conventional marketing. It is a schematic diagram for demonstrating LTV. 4 is a schematic flow chart showing the principle of LSV navigation used in the present invention; FIG. 10 is a diagram showing a flow of target extraction by LSV navigation; It is a figure which shows an example of an analysis result report. It is a block diagram which shows the structural example of this invention. It is a flow chart which shows an example of the whole operation of the present invention. FIG. 11 is a flow chart showing an example of obtaining the total number of potential customers for products; FIG. 7 is a flowchart showing an operation example of tree analysis; FIG. 4 is a diagram showing an example of a tree; It is a flowchart which shows the operation example which calculates|requires the probability of a goods contract. It is a flowchart which shows the operation example of a potential customer extraction process. FIG. 11 is a chart showing an example of the relationship between life levels and lifestyle groups; FIG. FIG. 10 is a chart showing an example of lifestyle codes and segments; FIG. It is a map which shows an example of the relationship between an address and a living level. It is a schematic diagram of risk value calculation regarding purchasing power. It is a schematic diagram of risk value calculation regarding interest. It is a schematic diagram of risk value calculation regarding age.

The present invention analyzes the similarity of lifestyle cluster data obtained by analyzing the lifestyles of existing customers, potential earnings predicted by becoming customers in the future, and risk values related to risk factors such as age. This is a potential customer discovery support system that calculates a converted LSV value, refers to the LSV value for a product set as a business target, and efficiently pinpoints potential customers in other regions. That is, the present invention expresses different multifaceted elements as diversity, matches them with the purpose of use, maps them onto an intuitively understandable Euclidean plane, and stores them in CRM, SFA, MCIF, etc. within a company. Lifestyle obtained by analyzing lifestyles using existing customer data (data owned by users), public statistical information such as national census, and administrative boundary data including public topography data Cluster data is calculated and the similarity of lifestyle cluster data is obtained. Store VCIF accounts are grouped (segmented) by town or chome (hereinafter simply referred to as "cho-chome"), and potential profits and risk values are calculated for products set (input) as business targets. . Then, based on the similarity of the lifestyle cluster data, the potential profit and the risk value, the LSV value that indicates the value based on the lifestyle for each town-chome is obtained, Calculate the potential customer rate for each town, chome, and from the total number of potential customer products, estimate the potential profit of the number of potential customers using the total number of accounts or the total population, and sell products for individuals at companies such as banks It is an innovative sales support system that can scientifically contribute to the discovery of potential customers in other regions where the product is purchased.

Here, "potential revenue" is the sum of potential customers and prospective customers, and means the revenue expected from customers who will feel the need for products and services in the future and become customers. be done. The "number of potential customers" is the estimated number of customers who do not feel the need for the product or service yet, or who feel the need but have not introduced or purchased the product or service (estimated number of potential customers and prospective customers combined). is. The number of potential customers is calculated by Equation 12 below.
(Number 11)
Potential income = (annual income or annual transaction value of cho-cho-moku using statistical information) / (total population of cho-cho-moku or total number of accounts)
(Number 12)
Number of potential customers = Total number of households or total population of town-chome - Number of existing customer accounts or number of existing customers after name identification

In addition, the "potential customer contract rate" is obtained by obtaining the contract rate of the highest existing customer in the town-chome according to Equation 13 below, subtracting the contract rate of the existing customer in each town-chome, and obtaining the highest existing customer contract rate in the town-chome. Calculated by dividing by the contract rate. That is, the contract rate of potential customers is calculated by the following equation 14.
(Number 13)
Contract rate of existing customers = (number of existing customer accounts or customers after name identification of cho-cho-moku) / (total number of households or total population of cho-chome)
(number 14)
Contract rate of potential customers = (Highest contract rate of existing customers in cho-chome - Contract rate of existing customers in each cho-chome) / (Highest contract rate of existing customers in cho-chome)

In the present invention, in addition to calculating the potential profit for each district, the risk factors (purchasing power, interest, age, etc.), and objectively quantify it as a risk value, and review it on an annual basis (for example, 1 to 5 years). In the present invention, the risk value is obtained from purchasing power that is likely to lead to luck, interest that is likely to lead to ability , and age that is likely to lead to personal connections . In addition, there is an inverse relationship between the creditworthiness of existing customers, which is invisible to the eye (identification characteristics, fulfillment ability, credit history, personal connections, behavior preference, etc.) and the risk value. If the risk value is small and the risk value is small, the creditworthiness will be highly evaluated. The relationship is shown in Equation 15 below. In the present invention, a risk value that is numerically easy to calculate is obtained, and the LSV value is calculated using the risk value.

In current markets, such as those related to SDGs (Sustainable Development Goals), the automobile and shipbuilding industries, many methods of rebirth by making new rules are adopted. In other words, rather than seeking a new and separate market (industry creation), it is often made up by reusing existing markets. In order to more effectively propose new products to the existing market, there is a need for a mechanism to calculate customer value closer to the current situation from existing customer lifetime value (LTV). The problem is as described above. Therefore, in the present invention, the lifestyle of an existing customer is analyzed, the person's lifestyle value (LSV value) is calculated, and the LSV value is used as a method of discovering potential customers in areas where sales performance is poor. We propose a new LSV navigation system. For example, the LSV navigation of the present invention uses statistical information such as public census and administrative boundary data (topography data), lifestyle cluster data for each area generated by clustering classification , Good fishing grounds) information, and extract and acquire target customers (potential customers) from customer data accumulated in CRM/SFA, etc. owned by companies such as financial institutions. "Hit") is a mechanism (navigation).

FIG. 3 is a schematic flow chart showing an overview of the LSV navigation used in the present invention. As shown in FIG. As shown in 2(B), the chome town where the target resides (an example of a good fishing ground) is specified. For example, it may be an area in common with an area where the purchase volume of the products targeted for business is high, and an area where many wealthy people live. Then, using the LSV navigation, the target is narrowed down as shown in Figure 3(C), then the list is compared from the analysis of the DB (database) as shown in Figure 3(D), and the sales area Acquisition of targets (potential customers), that is, one-line fishing (hit) is performed to promote sales of products (Fig. 3(E)).

FIG. 4 shows the flow of target (potential customer) discovery by LSV navigation, and the user-held data held by user companies such as financial institutions is customer number, name, gender, as shown in FIG. Address, date of birth, etc. are added and classified, and this user-held data includes lifestyle cluster data LD obtained by lifestyle cluster classification as shown in FIG. Using the administrative boundary data shown in FIG. 4(D), the LSV value shown in FIG. ) to obtain an analyzed potential customer list (ranking) as shown in FIG. The LSV value is obtained by obtaining the similarity of the lifestyle cluster data LD, calculating the potential profit from the number of potential customers for each town, and multiplying the similarity of the lifestyle cluster data LD, the potential profit, and the risk value for each customer. demand. The “similarity” of the lifestyle cluster data LD refers to consumer behavior characteristics (geographical People living in machi-chome areas in groups categorized by demographic data) are expected to have essentially the same consumer behavior. Since there are no regional differences in ordinary deposits and time deposits, which are representative of assets in custody, it is thought that there are similarities in consumer behavior patterns by town-chome unit.

However, if the risk value is multiplied as it is, the higher the risk value, the larger the LSV value. That is, since risk and credit are in a contradictory relationship, when risk value and creditworthiness are represented by the maximum value "100(%)", the risk value is represented by Equation 15 below.
(Number 15)
Risk value (%) = 100 - Creditworthiness → Creditworthiness = 100 - Risk value (%)

Therefore, the LSV value is calculated by Equation 16 below.
(Number 16)
LSV value = similarity of lifestyle cluster data LD × potential profit × (100 - risk value (%))

Details of individual customers (for example, rank 1) in the potential customer list showing the analysis results are shown in FIG. 5, and the potential customer list includes rank, name, group, segment, risk value, and the like. The risk value is a quantification of the risk of a person's characteristic risk factors, such as purchasing power, interest, and age. Lifestyle cluster data LD consisting of a pair of a group (code) and a segment is attached with a lifestyle code, which will be described later. It is obtained by performing clustering classification in a lifestyle cluster data generation unit (210) based on statistical information such as a national census, based on the assumption that they will share lifestyles, behaviors, and attitudes. Details will be described later, but for example, Group A of elites active in big cities (elite with high education and high income who live in high-rise apartments in the city and lead the economy and culture of Japan), and executives in high-end residential areas. Group B (living in a luxury residential area on the outskirts of a big city. Wealthy family households who have achieved social status by succeeding in large companies) are clustered together. In addition, the administrative boundary data includes the population and area for each town-chome unit.

FIG. 6 shows an example of the overall configuration in which the system 100 of the present invention cooperates. The potential customer discovery support system 100 according to the present invention comprises a DB (database) 110 and an LSV navigation system 120 that cooperates with the DB 110. , like a general computer, an input unit 111 such as a keyboard and a mouse, an output unit 112 such as data output, transmission, and printer, a display unit 113 such as a display screen, a CPU, a memory, etc. 114, the LSV navigation system 120 includes an LSV value calculation unit 121 that calculates the LSV value, a ranking unit 122 that ranks potential customers, and provides lifestyle raster data to potential customers. It consists of a profiling unit 123 for estimation and a segmentation unit 124 for grouping into chochomoku.

The LSV navigation system 120 receives customer data from the user-owned data 200, lifestyle cluster data LD from the lifestyle cluster data generator 210, and administrative boundary data 220. FIG. The administrative boundary data 220 is prepared as GIS data including prefecture names, branch office/promotion bureau names, counties, ordinance-designated city names, municipality names, administrative area codes, and the like for administrative boundaries throughout the country. The lifestyle cluster data LD is generated by the lifestyle cluster data generation unit 210 from statistical information such as a national census. In general, "statistical information" is aggregated and processed results of surveys conducted under certain conditions such as national census and commercial statistics. (including surveys) and information surveyed by local governments and various industry groups. For example, vegetables = statistical information, cabbage = national census, and radish = commercial statistics, and "mesh" and "administrative boundaries" are often used divisions when used in existing area marketing systems. In the present invention, if it corresponds to FIG. 3, the "administrative boundaries" are the sections showing good fishing grounds, and the "national census" is the statistics obtained by examining what kind of fish exist in those areas.

A customer 200 such as a financial institution has a CRM/SFA 205 and has personal customer data 201, corporate customer data 202, transaction data 203, and product data 204 as user-owned data. SFA (Sales Force Automation) is a sales support system, and CRM includes support and management of sales activities. . SFA, which supports sales, is included as a function of CRM, which is used in a wide range of operations from marketing to support services. User-owned data is held by user companies such as financial institutions. In the case of financial institutions, it is stored in host computers and various systems (including CRM and SFA) for each purpose of acquisition. 201, corporate customer data 202, transaction data 203, product data 204, and the like. These user-owned data groups are taken into the LSV navigation system 120, and the obtained LSV values are returned to the CRM/SFA 205. However, if the user already has the CRM/SFA 205 as in this example, the temporary storage is stored in the DB 110 of

The LSV navigation system 120 receives various data (individual customer data 201, corporate customer data 202, transaction data 203, product data 204) held by the user (purchaser or user) from the lifestyle cluster data generation unit 210. Using the lifestyle cluster data LD and the administrative boundary data 220, the LSV value is calculated based on Equation 16, a potential customer list ranked by the ranking unit 122 is created and output from the output unit 112, and the display unit 113. The LSV value calculation unit 121 analyzes various types of acquired data and calculates the LSV value in a collection of statistical methods and tools that support marketing activities, and the segmentation unit 124 calculates approximate characteristics ( For customers who have different ways of thinking and values), the market is subdivided (by town and village) and grouped by needs (products to be sold (e.g. new products, products that the customer does not own, etc.)). In addition, the ranking unit 122 ranks the obtained aggregation based on the LSV value, and the profiling unit 123 assigns lifestyle cluster data to potential customers who may become customers of the company's products and services. to estimate.

FIG. 7 is a flowchart of sales support when it is assumed that Regional Bank A will introduce "demographic data by town-chome unit". is input, and the customer data of the district bank A's customer is introduced from the data held by the user (step S10). Next, the segmentation unit 124 performs grouping (segmentation) of the store CIF (store-customer information file) by district (step S11). is aggregated for each group (step S12). A CIF is a unique management number assigned to each customer when an account is opened, and is associated with a branch number that identifies a branch of a financial institution and is called a branch CIF. Also, "name identification" is to integrate data into a single entity by assigning the same ID to the same person, the same company, and the same household in a plurality of distributed databases.

Next, the LSV value of each customer is calculated by the LSV value calculation unit 121 according to Equation 16 (step S20). (Step S30). As shown in Equation 16, the LSV value is obtained by finding the similarity of the lifestyle cluster data LD, calculating the potential profit from the number of potential customers for each town, and multiplying each customer by the risk value. ), there are also risk factors due to the characteristics of the customer, so for example, a risk value that quantifies the customer's risk with respect to purchasing power, interest, and age is used. The LSV value is calculated for all customers in the town-chome area, and is calculated by age category, by group, and by product (on a net business profit basis), potential profit, and risk value. Calculation of the LSV value and the risk value will be described later.

Then, based on the LSV value calculated from the data possessed by the user, the product contract record for each town-chome unit (for example, area X) is extracted for the set product (step S31). The ranking unit 122 obtains customers with rankings (step S32). Since the LSV value of all the customers is known, the LSV value of the customer with the high product contract record can also be known. The calculation unit 114 obtains the potential customer contract rate based on the difference in the contract rate of existing customers between the area with the highest product contract record and the area with the lowest product contract record (step S33), and calculates the potential customer prospect for each district. ask for a number. The potential customer's contract prospect rate is obtained according to the following equation (17).
(number 17)
Potential customer contract rate = Highest contract rate for existing customers in Cho-Chome - Contract rate for existing customers in areas with low contract performance

After the potential customer's contract probability rate is obtained by Equation 17, the number of potential customers excluding the existing customers in Cho-cho-moku is multiplied by the potential customer's contract probability rate to obtain the maximum value of the potential customer's prospective number.
(number 18)
Prospect number of potential customers = Number of potential customers × Prospect rate of contract prospects for potential customers

Next, by totaling all groups in the same age group, totaling all age groups, and similarly totaling all products, the total number of potential customer prospects is obtained (step S40). The potential profit of the number is calculated according to the above equation 11 (step S50), and sales (ordinary sales operations) are carried out based on the expected number of potential customers in each chome (step S51). By calculating the potential profit and grasping the difference from the profit of the actual business partner, it is possible to determine the market value, such as whether to dig up existing transactions or review it as a new product. The estimation of the potential profit of the expected number of potential customers in Equation 11 is carried out by obtaining the profit by the LSV value by comparing with the current number of product accounts.

FIG. 8 shows the details of step S40. First, the combination of attribute items and transaction items of the account with the highest contract rate is examined, and the contract rate by product/group is predicted (step S41). The population is specified based on the account holding the product in the town-chome and the account not holding the product in the town-chome (step S42). That is, to specify the population, first, products, age categories, genders, and groups are specified. Since the group is specified, we aggregate the CIF accounts (hereinafter simply referred to as "accounts") of stores that have products in the same town-chome area and Aggregate all accounts that are not The total number of cases of both forms the population, and if this is less than 5000 cases, the conditions for specification are relaxed. For example, if there are no 5000 items in the age division (5-year increments), 10-year-old increments are used, and certain condition items are completely removed if necessary. Attribute information and transaction items of the account are determined in advance. For example, when grouping a store CIF (after name identification), a group is assigned to the store CIF after name identification, and the mosaic group name, name, age, gender, address, and transaction items are assigned to the account, and account attribute information Items and transaction items are determined in advance and are the same for all accounts.

Next, tree analysis is performed (step S43), and the results of tree analysis are applied to each town-chome account (step S44). Details of the tree analysis will be described later. When applying to each cho-cho-moku account, create a ranking list of accounts that have a combination of transaction items obtained from each cho-chome tree analysis and do not currently have the product, arranged in descending order of contract rate. Then, obtain a list of potential customers with a high probability of contracting the product and the number of customers. The potential customer list (see FIG. 4E) includes store CIF, name, address, telephone number, and the like. After application to each town-chome account in step S44, the potential customer list is given an individual rating (step S45). Individual ratings serve as a database for estimating the risks of individual customers, enabling sales promotion and risk control at the same time. At that time, even if the probability of a potential customer contracting a product is high, it will be possible for a customer with a high risk value (for example, sales promotion and risk control to be performed at the same time). risk value of 95 or higher) must be excluded in advance before contacting, and likewise, accounts of contact-prohibited destinations are also excluded in advance. Among loan products, accounts with high contract rates and low personal ratings have the highest contract rates. The contract rate is high for products related to assets in custody, and customers with high personal ratings are accounts with the highest contract rates. In either case, exclude accounts with high risk values in advance.

Details of the tree analysis in step S43 will be described with reference to the tree diagram of FIG. The population (total number of samples "1000", total number of accounts "1500") is first male (total sample number "6000", total number of accounts "1200") and female (total sample number "4000", total number of accounts " 300”), and both men and women are further subdivided according to whether or not they have a credit card, and nodes “001” to “004” are given to each. The numbers in parentheses for each item in FIG. 10 indicate the number of classified samples, the number of accounts with credit cards among them, and their ratios. Since the variables are all AND conditions, node "001" is a male account with a bank credit card, node "002" is a male account without a bank credit card, node "003" ” is a female account with a bank credit card, and node “004” is an account with a female without a bank credit card. In reality, variables (total number of samples and total number of accounts) appear one after another, but two variables are used here, and a combination of variables is called a node, and four nodes (“001” to “004”) are shown here. ing. A node displays a combination of variables and a contract rate (percentage of card holders) of a credit card product indicated by the combination. Almost all financial institutions provide customers with a cash card integrated with a credit card, but some customers do not use credit cards, so even if they do not have a credit card, the card ownership rate is not zero.

In the tree analysis shown in FIG. 10, the contract rate of credit card products is highest for a male at node "001" who has an account with a credit card, and the contract rate for credit card products is 27.5. %, and on the other hand, the credit card contract rate for female accounts with no credit card (no credit card) at node "004" is extremely low at 1.1%. After tree analysis is completed, the nodes are arranged in descending order of contract rate, and the combination of attribute items and transaction items is determined for each node. In the example of FIG. 10, the highest order of contract rate is (1) men with credit cards, (2) women with credit cards, (3) men without credit cards, (4) women with credit cards. become unpossessed.

The details of step S44 are shown in the flow chart of FIG. 9, which is the process of applying the tree analysis result to each town-chome account. First, the probability of contracting a product is calculated (step S44-1), and the risk (or credit) is analyzed by combining attribute items and transaction items (step S44-2). A method for obtaining the probability of a product contract will be described later, but lifestyle courts and user-owned data are used. Among the accounts of each town and chome in the same group, we select and list accounts that have the same attribute items and transaction items as the node with a high contract rate and do not have credit card products. The above-described selection is performed in descending order (step S44-3), and combinations of transaction information are determined at each node (step S44-4). The selected account is a specific account, and is given a name, address, and telephone number, so whether the contact is a DM or a contact person, it becomes a sales activity for a specific individual. Here, a personal rating is further given to the customer list of the listed account. For personal rating, customers with a risk value of "95" or higher, for example, are excluded from the list.

Here, an operation example of a technique for obtaining the probability of a product contract will be described with reference to the flow chart of FIG.

First, the accounts are sorted into groups from the account list based on the town-chome of the account, age, contract rate for each product in the town-chome, etc. (step 44A), and the product is specified (step 44B). Next, the contract probability 1 of the specific product determined among the accounts in the same group is calculated (step 44C), and the contract probability 1 of the specific product is calculated with the accounts having the specific product and the accounts not having the specific product as one population. 2 is calculated (step 44D). Then, using tree analysis (step 44E), the contract probability for a particular product is indicated at the terminal node (step 44F), the terminal node is indicated with the number of samples and the number of contracted accounts contained in that node, and the percentage is issued as the contract probability (step 44G).

A potential customer extraction process by the potential customer discovery support system 100 will be described in detail with reference to the flowchart of FIG. 12 (detailed version of FIG. 7).

If the same customer has multiple accounts, use customer attribute information (name, address, date of birth (individual), date of establishment (corporation), etc.) to confirm attribute matching with existing accounts, If they match, they must be "unifiedly managed" as multiple accounts of the same customer. After the name identification process for centralized management (step S100), the segmentation unit 124 assigns a group according to the town and chome of the account owner's address (step S110), and classifies the accounts according to age, for example, in increments of 5 years old ( step S111). Purchasing accounts classified by age group as external data, similarity of consumer behavior generated by clustering analysis (lifestyle cluster data LD) for each area obtained from lifestyle code and user possession data The classification result is divided into groups (step S112), and the accounts of the groups are divided again by town-chome (step S113). Combining external purchase data with customer transaction data stored in CRM/SFA 110 generates LSV values (albeit pseudo-values) based on key individual products. The generated LSV value is required to be maximized by its definition, and the maximization process is performed. The purpose of the LSV navigation system is to increase "potential customer discovery" and "customer retention rate". If you spend too much, the "customer retention cost" will increase, which will reduce your profitability. It is important to increase the "discovery of potential customers" and "customer retention rate (contract utilization rate)" while keeping costs down by improving work efficiency.

Potential customers are predicted by maximizing the purchase frequency and the possibility of converting existing customers to new customers in order to increase the transaction price for each town and district for targets residing in the specified sales area. In other words, confirm the difference with the current supplier. Based on the LSV values obtained in this process, the scale of target customers for each individual product area is calculated. Then, the sum of the number of accounts and the number of contracts for each cho-chome is calculated (step S114), the population of each cho-chome is calculated (step S115), and the contract rate for each cho-chome product is calculated (step S115). S116). This contract rate is a contract rate per population or per account.

Next, for each product, the contract rate for the town-chome with the highest contract rate is determined (step S117), and the contract rate for each town-chome is subtracted from the highest contract rate to calculate the prospective contract rate for potential customers ( Step S118), the prospective contract rate of the potential customer is multiplied by the population or the number of accounts in that town/chome to determine the prospective number of potential customers (step S119). Even if the number of potential customers is determined, normal sales efforts are required to make it actual. The above steps S110 to S119 are the classification process.

From the account list obtained in step S110, the accounts are classified into groups of potential customers for each individual product area (step S120), and the product is specified (step S121). Next, the contract rate of the specified product determined among the accounts in the same group is calculated (step S122), and the contract rate of the specific product is calculated with accounts having and not having the specified product as one population (step S122). S123), the tree analysis shown in FIG. 10 is performed (step S124). The contract rate of a specific product is indicated as a node at the end, and the node indicates the number of samples and the number of contracted accounts included in the node as shown in FIG. 10, and the rate is displayed as the contract rate.

Next, all nodes are rearranged in order from the node with the highest contract rate (step S125), and the number of accounts that do not have a specific product among the sample nodes with the highest contract rate is displayed (step S126). Next, the accounts that do not have a specific product are classified by town and street and listed (step S127), and the above method is sequentially performed for each node with a high contract rate (step S130). Accounts are listed in descending order of specific approval contract rate for each town/chome of the same group (step S131), the above method is performed for each group (step S132), and the above method is performed for all products (step S132). S133). In this way, all accounts that do not have a specific product may be covered by the contract. However, in the current trading situation, it is essential to designate an account with a high contract rate. For this reason, it is necessary to multiply the accounts listed in descending order of contract rate in each cho-chome by a cap, and the cap is the number of potential customers in each cho-chome shown in step S127. The calculation of the number of potential customers is performed for each age group. The accounts listed up for each cho-chome are reclassified by age group (step S134), and the results are compared with the number of potential customers. The final output is a list of accounts with a high contract rate for purchasing the product for each group, for each product, and for each town/chome (step S140).

Classification of lifestyle including living level is known, for example, as shown in FIG. Wealth levels are also determined for each of "rural area", "suburban area", "city area", and "city center area". For example, the wealth level of the "city center area" is classified into "trendy life," "urban middle class," and "downtown." A list of lifestyle cluster data assigned to each category is shown in FIG. A lifestyle code can be automatically assigned based on

Also, the contents of the lifestyle cluster data LD are, for example, shown in FIG. 14, and the lifestyle cluster data consists of lifestyle codes and segments.

Next, the analysis items that are considered to be owned by companies are, for example, deposits/savings/investments, occupation, real estate, credit history, income, family structure, age, and so on. , including map information indicating companies and personal names (for family tree information).

Figure 16 is a schematic diagram showing an example of risk calculation for purchasing power among the three major risks. , final educational background (vocational school, high school, university, etc.), etc. Income includes age, dividend income, salary, etc. Family structure includes cohabitation of two or more households by age group, nuclear family, and singles. For real estate, inheritance/inheritance (business land, home, farmland, rental condominium, etc.), self-acquisition (business land, home, farmland, rental condominium, etc.), Real estate rentals (business land, farmland, rental apartments, etc.) are included, and credit records include good, fair, poor, and anti-social relations.

The target products for risk analysis are asset management (investment trusts), life insurance, liquid deposits (foreign currency deposits, time deposits, etc.), housing loans (individuals, corporations), credit cards, and are also applied to merchandise sales. An example of risk calculation for risk factor interest is shown in FIG. 17, and an example of risk calculation for age is shown in FIG. Investment in FIG. 17 includes investment trusts, deposits, and stocks with results, interest, and none. Health in FIG. included.

If "interest = no interest" and "90" is assumed to be a risk value, a value converted to creditworthiness is obtained. It is better to give up. If you are hesitant, the risk value is 40, and if you are interested, the risk value is 15 and converted to creditworthiness. In addition, "purchasing power = no money," which is affected by educational background, occupation, income, credit, family structure, hobbies, and real estate. Resurrection is difficult and there is no after. There are three major risks: no interest (A), no money to buy (B), and no age (C). The combination of the above three factors (A, B, C, AB, AC, BC, ABC) determines creditworthiness. For example, "I don't have money, but I'm interested and young," "I have money, but I'm young, but I'm not interested," and "I'm interested in money, but I'm not ready."

100 Potential customer discovery support system 110 DB (database)
111 Input unit 112 Output unit 113 Display unit 114 Calculation unit 120 LSV navigation system 121 LSV value calculation unit 122 Ranking unit 123 Profiling unit 124 Segmentation unit 200 Customer system 201 Individual customer data 202 Corporate customer data 203 Transaction data 204 Product data 205 CRM/ SFA
210 lifestyle cluster data generator 220 administrative boundary data 250 statistical data

Claims (3)

a database;
Based on statistical information aggregated and processed from national census, commercial statistics, ministries and agencies, local governments, various industry groups, etc., surveyed and published under certain conditions, the cluster of the lifestyle cluster data generation department Lifestyle cluster data for each town or chome generated by ring classification, user-owned data including customer's individual customer data, corporate customer data, transaction data, product data, and nationwide administrative boundaries, Publicly available administrative boundary data organized as GIS (Map Information System) data such as prefecture name, branch office/promotion bureau name, county, ordinance-designated city name, municipality name, administrative area code, etc. is input, and the database is entered. LSV (Life Style Value) navigation system linked to
a calculation unit that performs calculation processing in cooperation with the database and the LSV navigation system;
an input/output unit that inputs/outputs commands and inputs/outputs information and data;
are interconnected and
The LSV navigation system is
a segmentation unit that groups the lifestyle cluster data and the user-owned data into the town-chome based on the administrative boundary data;
LSV (Life Style) that indicates the value based on each person's lifestyle
a LSV value calculation unit that calculates a Value) value;
Based on the LSV value, the potential customers who are customers who do not feel the need for products or services yet or customers who do not recognize specific needs are ranked, a potential customer list is created, and the input/output unit a ranking section output from
a profiling unit for estimating potential revenue, which means revenue expected from becoming a customer, from the potential customer count for each town-chome calculated by the computing unit;
is composed of
The LSV value calculation unit,
Cooperating with the computing unit,
The similarity of the lifestyle cluster data obtained from the consumer behavior patterns of the town-chome unit grouped in the segmentation unit, and the profiling unit “potential income = (annual income of the town-chome using statistical information or annual transaction amount) / (total population or total number of accounts in the town-chome)” and the risk value (%) that quantifies the risk of the customer, the existing The customer's LSV value is calculated by "LSV value = similarity of lifestyle cluster data x potential income x (100 - risk value (%))" for all customers in the town-chome,
The ranking unit
creating the potential customer list by ranking the potential customers who have a high product contract record in each town-chome for the product input from the input/output unit;
The computing unit
Based on the difference in the contract rate of the existing customers between the areas with high and low product contract records, the contract prospect rate of the potential customers is calculated as "potential customer contract prospect rate = highest existing customer contract rate in Cho-chome - "Contracts of existing customers in areas with low contract records", and the expected number of potential customers for each town-chome unit is calculated by "Expected number of potential customers = Number of potential customers x Potential contract rate of potential customers", and the expected number of potential customers Find the total product total,
The profiling unit estimates the potential revenue of the expected number of potential customers using the total number of accounts or the total population of the total product total,
A potential customer discovery support system characterized in that sales can be conducted based on the number of potential customers for each town-chome unit . The risk value is calculated based on purchasing power that is likely to lead to luck, interest that is likely to lead to ability, and age that is likely to lead to personal connections,
Regarding the purchasing power, family composition, hobbies, real estate, educational background, occupation, income and credit performance are characteristic factors, and for the interest, family composition, hobbies, investment, educational background, occupation, income and credit performance are characteristic factors, 2. A potential customer discovery support system according to claim 1, wherein said age is characterized by family structure, hobbies, health, educational background, occupation, income and credit record . In the calculation unit,
The number of potential customers is calculated by "number of potential customers = total number of households or total population in town-chome - number of existing customer accounts or number of existing customers after name identification",
The existing customer contract rate is calculated by "existing customer contract rate = (number of existing customer accounts or number of customers after cho-chome name identification) / (total number of cho-chome households or total population)",
The potential customer contract rate is defined as "potential customer contract rate = (highest contract rate of existing customers in a town-chome - contract rate of existing customers in each town-chome) / (highest contract rate of existing customers in a town-chome) 3. The potential customer discovery support system according to claim 1 or 2, wherein the calculation is performed by " .

Images