JP2003114977A - Method and system for calculating customer's lifelong value - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a method and a system for calculating exquisite future earnings from a customer and the customer's lifelong value in terms of contract form of each merchandise. SOLUTION: In the method for calculating the customers' lifelong values as the total of earnings which a banking facility can be obtained from transactions of financial products with the customers in the future, the customers' lifelong values 800 of contracted transactions in the future are calculated by inputting customer information 300 consisting of history of transaction state such as attributes of the respective customers in a certain period, transaction amounts by every financial product, contract period, by inputting bank account information 400 indicating details of the current transaction of the customers, predicting future earnings 700 by every contract period of the contracted transaction in the future for contracting continuation of the transaction to the preliminarily determined future, and by discounting and totaling the future earnings to the current values. Customers' lifelong value of uncontracted transaction in the future is calculated by predicting every period's earnings to be obtained from the uncontracted transaction in the future in which continuation of the transaction to the future is not scheduled on the contract.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a method and system for calculating customer lifetime value in financial institutions.

[0002]

BACKGROUND OF THE INVENTION Customer lifetime value is the sum of a customer's ongoing profit up to a certain point in the future, as well as the profit at a certain point in the future. Therefore, it is necessary to predict customer revenue at each point in time in the future.

[0003] Customer lifetime value is used to target customers who are likely to generate future profits for sales and to give preferential treatment conditions to the customers.

As a conventional technique for predicting the future profit of a customer, there is a technique disclosed in Japanese Patent Laid-Open No. 2000-99497. According to this technology, in order to predict the future revenue of the forecast target customer, the similarity of the past attribute information between the forecast target customer and other customers is calculated, and the actual revenue value of the other customer at the time of the forecast is calculated. The future profit of the forecast target customer is calculated by weighting by the degree of similarity and integrating.

[0005]

[Problems to be Solved by the Invention] When a financial institution is transacting with a customer, depending on the product, a future transaction period such as a fixed deposit or a mortgage is promised, and unless there is an early cancellation, There are transactions where the balance is likely to continue. However, in the conventional technology, in calculating the future profit of the customer, only the actual profit value of the other customer is used, and the current transaction content is used to measure the degree of similarity with the other customer, It is not used for calculating future profits.
For this reason, it is highly likely that the accuracy of forecasting will fall because it does not take into account the profits obtained from transactions such as time deposits and mortgages, which are contracted to continue in the future.

In addition, the profit to be predicted in the prior art is not only dependent on the customer transaction balance, but is also a value that fluctuates due to the influence of interest rate changes accompanying changes in the economic environment. From this, even if the balance remains unchanged, changes in interest rates will cause a difference in earnings. Also, future interest rates are not necessarily the same as past rates. Therefore, if the revenues are directly used to forecast the revenues of other customers, the accuracy is likely to deteriorate.

Further, in the prior art, the customer's profit is used as a target item for prediction and is predicted using one prediction model. However, the actual profit is obtained from a plurality of financial product transactions that the customer is trading. It is the sum of all the profits. As is clear from the fact that the product characteristics of deposits and loans are completely different,
The transaction tendency of financial products differs for each product, and it is expected that the accuracy of prediction will be reduced by using only one prediction model.

Further, since the conventional technology only calculates the lifetime value of the customer, it is considered difficult to utilize how to promote the sales activity to the customer in order to increase the lifetime value of the customer.

The object of the present invention is to distinguish future contracted transactions and future non-contracted transactions from the customer's attributes and transaction data, and further contract the customer transaction amount such as the balance and the fee, which is the basis for calculating the profit, with the product. By predicting by type, the accuracy of customer lifetime value can be improved.

Another object of the present invention is to set, as a recommended product, a product showing a high contribution to future profit improvement by using a highly accurate customer lifetime value and a prediction model used for prediction.
It is to provide a method and system for calculating customer lifetime value that can be used in sales activities.

[0011]

According to the present invention, a financial institution
A method of calculating a lifetime value of a customer, which is the total amount of profits to be earned from future transactions of financial products with a customer,
Enter customer information consisting of the history of transaction status such as the attributes of each customer for a certain period, transaction amount for each financial product, contract period,
If there is a contracted transaction for a financial product that enters the account information indicating the current transaction details of the customer and contracts to continue the transaction registered in advance, future revenue by contract period of each customer's future contracted transaction The future value of future contracted transactions is discounted to the present value, and the customer lifetime value of future contracted transactions is calculated. By predicting future profits from contract transactions up to a certain period and discounting future profits from future uncontracted trades to present value,
Calculate the customer lifetime value of future uncontracted transactions.

[0012]

BEST MODE FOR CARRYING OUT THE INVENTION Embodiments of the present invention will be described below with reference to the drawings.

Note that the example used in the embodiment of the present invention is an example of calculating the customer lifetime value of an individual customer in a bank, especially in a financial institution. Here, hereinafter, a transaction contracted with a future deposit period and a repayment period defined is defined as a future contracted transaction. Then, a transaction that is currently traded but for which a future deposit period is not defined, a transaction for which a repayment period is not specifically defined, or a transaction that is not currently contracted is defined as a future uncontracted transaction.

FIG. 1 shows the processing flow of this embodiment.
The processing flow will be described later.

FIG. 2 is a computer 20 for implementing this embodiment.
0 represents the hardware configuration. The computer 200 shown in FIG.
Is a CPU (arithmetic processing device) 201, an input device 202,
The storage device 203 and the output device 204 are provided. The CPU 201 calls and executes a program for calculating the customer lifetime value from the storage device 203. The computer 200 may be a so-called personal computer, but if there is a lot of customer information to be calculated, the CP
U201 has a high-speed processing capability and is a storage device 2
03 is preferably large in capacity to store a large amount of customer information. Specifically, there are large-scale computers and high-performance servers.

FIG. 3 shows an example of the file structure of the customer information 300. The customer information of customers who have already contracted is attribute information such as the customer's age, gender, and annual income, the transaction status such as the number of transactions and the number of transactions of the customer, and the financial product that is the source for calculating customer profit. It is a time-series collection of information on transaction amounts, balances, and fees. The customer information 300 is
A data reference year 301 indicating the time of history data, a customer number 302 for identifying a customer, an age 303 as a customer attribute,
Gender 304, Annual income 305, Annual salary transfer amount 306 which is the transaction status, Public utility withdrawal number 307, Ordinary deposit balance 308 which is the customer transaction amount by financial product, Time deposit balance 309, Education loan balance 310, Safe deposit box It consists of items such as a usage fee 311. The customer information 300 stores customer attributes and transaction history in time series, and has a similar file structure for past data base years.

Here, hereinafter, a customer who has no transaction at present is defined as a new customer. The customer information of a new customer may be information on a company or a group of individuals classified by a specific attribute. In this case, there is no information on the transaction status or transaction amount of the customer, but various information such as age, sex and annual income, which are customer attributes, occupations, hobbies, names of specific groups to which the customer belongs can be considered. .

The customer information items in FIG. 3 are merely examples, and other items may be used. In addition to financial savings, time deposits, and safe deposit boxes, there are various types of financial products such as mortgages, card loans, and direct debits, which are omitted here. explain. In addition, the data reference time point indicates the time point when the customer information is aggregated and recorded at a fixed unit interval, but examples of the time unit include a daily unit, a monthly unit, an annual unit, a 2-year unit, a 3-year unit, etc. There are various possibilities. In the following examples, the unit is year. In addition, the balance shall be the average annual balance, and the usage fee shall be the annual usage fee.

FIG. 4 shows an example of the file structure of the account information 400. The account information 400 shows the transaction details about a certain product with the customer, and the file structure differs depending on the product, but here, an example of the account information of the time deposit is shown. The account information 400 includes a data base year 401 and a customer number 40.
2, account number 403, product course 40 indicating the type of product
4, balance 405, contract period 406, interest rate 407, contract start month 408, and the like. In some cases, one customer may have multiple contracts.

In the case of a mortgage, etc., in addition to the above, information such as monthly repayment amount is added as account information.

Next, the processing flow of FIG. 1 will be described. <Step 101> Input of Customer Information and Account Information Customer information 300 and account information 400 are input from the input device 202. Such information may be directly input by the person in charge of the bank, but normally, the customer information and account information of the bank are primarily acquired and stored in other systems in the bank. Information is previously stored in the storage device 203 in the present system by communication network transmission or a storage medium, and the input device 202 is used.
The customer information used in this processing flow may be designated and selected. <Step 102> Calculation of Future Profit of Future Contracted Transaction Based on the customer information and account information input in Step 101, the system calculates future profit from the present time to the contract deadline of the account. The calculation method will be described later.
Here, in this embodiment, the current data represents the latest data. In the present embodiment, since it is assumed that the data is handled on a yearly basis, for example, the forecast implementation time is 2
In the case of September 001, the current data is the previous 3
This means 2000 data up to the month. <Step 103> Calculation of customer lifetime value of future contracted transaction From the future profit of future contracted transaction calculated in step 102, the customer lifetime value of future contracted transaction is calculated. The formula for calculating the customer lifetime value is as follows.

[0022]

[Equation 1] Customer lifetime value is calculated by discounting future profits for each period to present value and then adding them up. As a discount rate for discounting the present value, an interest rate value is usually used. This is because if you manage your current assets into the future,
Since future assets will have interest rates added to them, the idea is to discount interest rates when converting future values into present values.

First, the customer lifetime value for each customer account is calculated. Next, when one customer has a plurality of accounts for one product, the product-by-product lifetime value is calculated by totaling the products. Then, by summing the lifetime value of each product, the lifetime value regarding the future contracted transaction of the customer is calculated.

FIG. 8 shows a file configuration example of the calculated customer lifetime value 800 of the future contracted transaction. The customer lifetime value of the future contracted transaction is composed of items such as customer number 801, customer lifetime value for each product, time deposit lifetime value 802, education loan lifetime value 803, and product customer lifetime value 804. The customer lifetime value for each product may include other products. <Step 104> Calculation of future profit of future non-contract transaction Based on the customer information 300, the present system calculates future profit obtained from future non-contract transaction for every certain period. How to calculate the profit for each period will be described later. <Step 105> Using the future profit of the future uncontracted transaction calculated in step 104, the customer lifetime value of the future uncontracted transaction is calculated. The customer lifetime value is calculated using the same mathematical formula as in step 103.

Similar to step 103, after calculating the lifetime value of each customer for each product, the lifetime values of future uncontracted transactions of the customer are calculated by summing them. The file configuration example of the calculated customer lifetime value of the future uncontracted transaction is the customer lifetime value of the future contracted transaction shown in FIG.
The description is omitted because it is almost the same as the file configuration example of. <Step 106> Total Customer Lifetime Value The customer lifetime value of future contracted transactions and the customer lifetime value of future uncontracted transactions are totaled for each customer to calculate the customer lifetime value.

FIG. 13 shows a file configuration example 1300 of the calculated customer lifetime value. Customer lifetime value file 130
0 consists of a customer number 1301 and a customer lifetime value 1302 that is a total of future contracted transactions and future non-contracted transactions. <Step 107> Extracting Customers with High Lifetime Value A customer with a high value in the customer lifetime value calculated in step 106 is extracted as a target customer for sales activities.

The customer lifetime value calculated in step 106 is stored in the storage device. Then, the person in charge of the bank or the like inputs the standard for extracting the sales target customer from the input device. There are absolute standards, relative standards, or both absolute and relative standards as standards. The absolute standard method is a method of setting a standard for the customer lifetime value, such as extracting a person who has a customer lifetime value of 5 million yen or more. The method of setting a relative standard is a method of setting a standard according to the number of people, such as extracting 1000 or more of the top customers' lifetime value. The method of taking a standard that combines both of them is a method of setting a standard for both the customer lifetime value and the number of people, such that the customer lifetime value is 5 million yen or more and the top 1000 or more. How to take the standard depends on the cost of sales activities.

These criteria are input from the input device, and the extracted customers are output to the output device. Alternatively, the standard may be stored in a storage device in advance and automatically performed until a customer is extracted. <Step 108> Setting a Product Highly Contributing to Profit Improvement as a Customer-Recommended Product Set which product to sell to the sales target customer extracted in step 107. How to select a product will be described later.

Next, FIG. 5 will be described. FIG. 5 shows a future profit calculation processing flow of a future contracted transaction. That is, the details of step 102. <Step 501> Reading Customer Information From the customer information 300 input in step 101, data representing the current time is extracted. The current data is judged based on the data base year 301.

Furthermore, regarding the extracted current data,
Read one record at a time (for each customer's information). <Step 502> Reading Account Information With respect to the record extracted in step 501, the account information indicating the details of the customer's transaction is read. For example, in step 501, the customer number in FIG. 3 is 123456.
If the 7th record is extracted, the record with the customer number of 1234567 will be extracted from the account information. Furthermore, if there is a transaction that contracts to continue the transaction until the future, that record is extracted. It is determined whether the transaction is a contract to continue the transaction from the future based on the information about the product item and contract period in the account information.

Whether the contract is planned to continue the transaction until the future depends on the characteristics of the product. For example, time deposits are usually contracted for future deposit periods in the form of maturities. In addition, with regards to automobile loans and educational loans, we have contracted to continue transactions in the future in the form of a repayment period. Conversely, ordinary deposits may be canceled at any time due to the characteristics of the products, and they do not contractually guarantee the continuation of future transactions. <Step 503> Calculation of Planned Balance of Future Contracted Transaction Using the record extracted in Step 502, the planned balance for the remaining transaction period is calculated from the current transaction balance 405 of each customer account. Here, the transaction remaining period is
It is the period after subtracting the period from the contract start to the present from the scheduled deposit period.

Since the customer lifetime value is the sum of the continuous profits of the customer up to a certain time point in the future, it is necessary to predict the customer profit at each time point up to a certain time point in the future. . Therefore, the calculation of the planned balance shall be performed every unit time. In other words, if the transaction remaining period is 3 years, the expected balance is calculated for each period of 1 year, 2 years, and 3 years.

Since there are various ways of taking units,
If the unit is 3 years, the expected balance only after 3 years, that is, only the expected balance at the time of transaction completion will be calculated. Therefore, it is possible to calculate the profit only when the transaction is completed.

If the planned balance for each period is a time deposit,
It is the current balance plus interest for the period.
In the case of mortgages, etc., the interest amount is added to the current balance and the repayment amount is reduced to obtain the planned balance.

FIG. 6 shows an example of the file configuration of the calculated future balance contracted planned balance file 600. The planned balance file 600 of future contracted transactions includes a data base year 601 indicating each time until the remaining transaction deadline, a customer number 602,
It includes items such as a product course 603, an account number 604, and a planned balance 605. Here, the transaction remaining term is the time when the contract of a certain transaction expires. The scheduled balance file 600 may include items such as transaction remaining period and transaction remaining period. <Step 504> Calculation of Future Profit of Future Contracted Transaction A planned balance for each fixed unit time during the transaction remaining period calculated in step 503, a transaction continuation probability registered in advance in the storage device 203, and a forecast profit margin. Predict the future revenue of future contracted transactions by taking the product of

As the transaction continuity probability, a value calculated by taking statistics such as the past mid-term churn rate is used. The transaction continuation probability is calculated by the following formula. Probability of continuation of transactions = (1- (number of customers who have canceled the product halfway) /
(Number of contract customers of the product) × 100 It is also possible to finely divide the product categories and the customer categories to obtain the above equation. The margin is the difference in the interest rate that a bank earns, and is defined as the difference between the interest rate on the market and the deposit interest rate paid to customers. As for the future expected margin, the one calculated based on the result of interest rate forecast or the same margin as the present is used. Further, the transaction continuation probability and the expected margin may be input from the input device 202 during the processing of this step.

FIG. 7 shows an example of the future profit calculation result 700 of a future contracted transaction. The future profit calculation result 700 includes items such as a data reference year 701 indicating each time until the transaction remaining deadline, a customer number 702, a product item 703, an account number 704, and a forecast profit 705. Further, items such as the transaction remaining period and the transaction remaining period may be included in the future profit calculation result 700.

Next, FIG. 9 will be described. FIG. 9 shows a future profit forecast flow of a future uncontracted transaction. That is,
Details of step 104. <Step 901> Reading Customer Information The customer information 300 input in step 101 is read.

FIG. 10 shows an example of the parameter setting screen 1000 when the prediction model is created. Parameter setting screen 10
00 is an area 1001 for inputting a prediction period, an area 1002 for selecting an item relating to a customer transaction amount which is a prediction target item, and an area 1 for selecting an item used for prediction.
003. Item selection areas 1002 and 1003
The items displayed in are items 301 to 309 stored in the customer information 300 input in step 101. Further, when the present time is not used as a reference, an area where a certain time in the past to a certain time and a time in the future to a certain time can be selected is added.

Taking the case where the item to be predicted is "balance of ordinary deposit" as an example, using the input device 202, "balance of ordinary deposit" is selected from the item 1002 of the item to be forecast and selected from the item selection area 1003. Select items such as "age" and "annual salary transfer amount", and then click the prediction execution button 1004.
Will be pressed.

The items selected by the above operation are read for the entire customer information record. Here, the customer information to be used may be that of all customers,
Usually, since the processing is difficult, randomly sampled customer information is used.

The forecast period depends on how long the bank staff wants to forecast the future period. You can select all the products handled by banks, but if you know the products that are not used much and do not contribute much to the profit, you can select only the important products. The prediction may be performed. The method of selecting the items to be used for prediction may be all items or any item that the user thinks is related to the prediction target item. Alternatively, a statistical analysis is performed in advance to calculate a quantitative correlation value,
An item having a high correlation may be selected.

Instead of inputting the prediction period, the prediction target item, and the usage item from the input device 202, they may be registered in the storage device 203 in advance and automatically processed. <Step 902> Modeling of Relationship Between Present Customer Transaction Amount and Past Customer Information Based on the read customer information, a prediction model is created as follows.

If the period to be predicted is T years from the present time, the relationship between the current customer transaction amount and the customer attributes up to T years ago is predicted using the customer information 300 from the current time to T years ago. Model as a model. For example, if the current year is 2000 and T = 5, the period to be predicted is from 2001 to 2005, and customer information for creating a prediction model uses information up to 1995, which is five years ago. . In addition, a prediction model is created for each prediction period up to 5 years later. In other words, for the forecast of 2001, a forecast model is created using the customer information of 2000 and 1999, and for the forecast of 2002, 200
Create a forecast model using customer information for years 0 and 1998. Similarly, for the 2005 forecast, 200
Create a prediction model using customer information for years 0 and 1995. Therefore, five prediction models will be created.

Methods used to model the relationship between items include a linear polynomial multiple regression analysis that predicts another variable based on some variables. The following is an example of a prediction model showing the relationship between the ordinary deposit balance and customer information obtained by regression analysis.

[0046]

[Equation 2] On the right

[0047]

[Equation 4] The item such as indicates that it is the item of the customer information of t years ago. Also,

[0048]

[Equation 5] Is a coefficient and c is a constant.

This predictive model expresses the variable relationship between the ordinary deposit balance and the items of each customer information by a linear linear relational expression so that the error of the actual data is the smallest.

The method used for modeling here may be, in addition to the multiple regression analysis, another method such as a neural network in which neuron models are connected to each other and connected to form a network. , The expression format of the prediction model depends on the method used. <Step 903> Calculation of Future Transaction Amount By inputting the current data of the customer information 300 of each forecast target customer into the product-specific forecast model for each forecast period created in step 902, each customer's forecast period Predict the transaction amount. Hereinafter, the ordinary deposit balance will be described as an example.

From Equation 2, the values of Equation 5 and c are determined. Therefore, by inputting the current data into the following formula, the ordinary deposit balance after t years can be obtained.

[0052]

[Equation 3] When T = 5, values of 1, 2, 3, 4, and 5 are applied to t, and five types of ordinary deposit balances are obtained for each prediction period.

FIG. 11 shows an example of the file structure 1100 of the transaction amount which is the prediction result. Forecast transaction amount file 1
Reference numeral 100 denotes a data base year 1101 indicating each time until the transaction remaining deadline, a customer number 1102, a forecast savings balance 1103 as a forecast result, a time deposit forecast balance 1104, an educational loan forecast balance 1105, and a safe deposit forecast usage fee 1106. Consists of. The prediction result at each point until the transaction remaining deadline is accumulated in the predicted transaction amount file 1100.

This processing is carried out for each transaction that is the target of prediction. In the above example, the balance of ordinary deposits was described, but similarly for time deposits, educational loans, and safe deposit boxes, a prediction model is created for the same prediction period, and the transaction amount for each prediction period is calculated. <Step 904> Profit Calculation of Future Uncontracted Transaction The transaction amount predicted in step 903 is multiplied by the profit rate registered in advance to calculate the customer profit in the prediction period.

For interest rate products such as ordinary deposits and fixed deposits, the profit margin is the same as the expected margin used in step 504. Here, the profit rate may be one in which the expense rate for office work and sales of customer transactions is added in advance. As a result, precise profit can be calculated. The same applies to step 504.

The profit calculation result for the future uncontracted transaction is the same as that in the file structure 1100 of the forecast transaction amount, in which the item relating to the forecast balance is replaced with the revenue, and therefore the description thereof will be omitted.

Next, the details of step 108 will be described. Using the prediction model created in step 902, a product closely related to profit improvement is determined. Specifically, since the coefficient of the prediction model indicates the depth of the relationship between the profit and each item, the product of the item having a large coefficient among the items related to the product transaction is selected as the customer recommended product and is output by the output device 204.
To display. As in the case of step 107, there are an absolute reference, a relative reference, or both an absolute reference and a relative reference as a method of obtaining the reference. However, since only people with a high customer lifetime value are targeted here, it is usual to use a relative standard. For example, the criterion is to recommend the top two items in descending order of coefficient. However, some customers have already contracted for the product. This is judged by the presence or absence of the balance in the customer information or the account information. Also in this case, the criteria should be decided such as recommending two items with the next largest coefficient. These criteria are input from the input device 202, and the item of the product corresponding to the customer extracted in step 108 is output to the output device. Alternatively, these criteria may be stored in the storage device in advance and automatically performed until the item of the product corresponding to the customer is extracted.

Next, an example of the module configuration of the customer lifetime value calculation system for implementing the customer lifetime value calculation method of the present invention will be described with reference to FIG.

The customer lifetime value calculation system of the present invention stores in the storage device 203 in the hardware configuration diagram shown in FIG.
It is stored as a program. The module of the present invention is
Corresponding to each step described using FIG. 1, future profit forecasting module 1301 of future contracted transactions, customer lifetime value calculation module 1302 of future contracted transactions, customer lifetime value calculation module 1303 of future uncontracted transactions, It is composed of a future profit forecasting module 1304 for future uncontracted transactions and a customer lifetime value total module 1305.

Finally, FIG. 14 shows an example of the system configuration when the customer lifetime value calculating method of the present invention is realized in a network environment. The customer lifetime value calculation system of this embodiment is a system in which a client 1410 and a server 1420 are connected by a communication network 1430.

The client 1410 is a computer 1411.
To the input device 1412 and the output device 1413. The server 1420 is a computer 1421 connected to a storage device 1422. The module for calculating the lifetime value of the customer operates on the server 1420. Client 141 connected to communication network 1430
0 may be plural. According to this embodiment, it is possible to provide a service of so-called ASP (application service provider) for customer lifetime value calculation processing in a remote environment.

As described above, according to the present embodiment, by predicting future contracted transactions and future non-contracted transactions in which the balance is likely to continue in the calculation of future profits separately, it is possible to obtain a highly accurate customer. Lifetime value can be calculated.

Further, instead of directly using the profit as a target for prediction with a value that fluctuates under the influence of a change in interest rates due to changes in the economic environment, the transaction amount such as the underlying balance or usage fee is used as a target for prediction. By doing so, it becomes possible to calculate the customer lifetime value more precisely.

Furthermore, by using the calculated customer lifetime value, it can be utilized for formulating the sales activity to the customer who is expected to have a high lifetime value. This can take a strategy to improve the total profit by quantitatively adding the future profit as compared with the conventional case where only the high profit clients tend to focus on sales. . According to the present invention, it is possible to set a product that makes a high contribution to future profit improvement as a recommended product and use it for sales activities. When products are sold, the hit rate and contract rate are improved, and when preferential services are used to improve profitability, customer satisfaction is improved, and further expansion of transactions leads to increased profits.

In the above embodiment, the customer lifetime value of an individual customer of a financial institution, particularly a bank, is calculated. However, the financial institution includes an insurance company, a securities company, a credit company, etc. in addition to the bank. There is. Insurance companies and securities companies,
In the case of a credit company, the financial products we handle are insurance and
Stocks, investment trusts, credit cards, etc. are the same as banks, in that a certain percentage of the transaction amount and fees are the profits earned from the transaction. The present invention can be applied to the present invention because it is the same in some products such as insurance, public and corporate bond investment trusts, and revolving payments for credit cards, which are contractually scheduled to continue transactions in the future.

Further, in the case of a corporate customer, the customer attribute of the customer information is changed to the number of years of establishment, capital, etc. instead of age or gender, and the financial product to be traded is changed to a bill loan instead of a mortgage. The present invention can be applied.

[0067]

According to the present invention, future contracted transactions and future non-contracted transactions are distinguished from the attributes and transaction data of customers, and the amount of customer transactions such as balances and fees that are the basis of profit calculation are used as products. It is possible to improve the accuracy of the customer lifetime value by making predictions according to the contract type. Alternatively, according to the present invention, it is possible to set, as a recommended product, a product that highly contributes to future profit improvement by using a highly accurate customer lifetime value and the prediction model used for the prediction, and use it as a sales product.

[Brief description of drawings]

FIG. 1 is an overall processing flow of a customer lifetime value calculation method according to the present invention.

FIG. 2 is a hardware configuration diagram of a system for implementing the customer lifetime value calculation method according to the embodiment of the present invention.

FIG. 3 is a file configuration example of customer information according to the embodiment of the present invention.

FIG. 4 is a file configuration example of account information according to the embodiment of the present invention.

FIG. 5 is a future profit calculation flow of a future contracted transaction in the embodiment of the present invention.

FIG. 6 is a file configuration example of a planned balance of future contracted transactions in the embodiment of the present invention.

FIG. 7 is a file configuration example of future profit of a future contracted transaction in the embodiment of the present invention.

FIG. 8 is a file configuration example of a customer lifetime value of a future contracted transaction in the embodiment of the present invention.

FIG. 9 is a future profit calculation flow for future uncontracted transactions in the embodiment of the present invention.

FIG. 10 is an example of a parameter setting screen for constructing a prediction model according to the embodiment of the present invention.

FIG. 11 is a file configuration example of a predicted amount of future uncontracted transaction in the embodiment of the present invention.

FIG. 12 is a file configuration example of customer lifetime value according to the embodiment of the present invention.

FIG. 13 is a module configuration example of a customer lifetime value system in the embodiment of the present invention.

FIG. 14 is a system configuration example when the customer lifetime value calculation method according to the embodiment of the present invention is realized in a network environment.

[Explanation of symbols]

101 ... Customer information and account information input processing, 102 ... Future profit forecasting processing of future contracted transactions, 103 ... Customer lifetime value calculation processing of future contracted transactions, 104 ... Future profit forecasting processing of future uncontracted transactions, 105 ... Customer lifetime value calculation processing for future uncontracted transactions, 106 ... Total processing of customer lifetime value, 10
7 ... Extraction process of high lifetime value customers, 108 ... Customer recommended product setting process of products that highly contribute to profit improvement, 200 ... Calculator, 3
00 ... Customer information, 400 ... Account information, 600 ... Planned balance information, 700 ... Future profit of future contracted transactions, 800 ... Customer lifetime value of future contracted transactions, 1000 ... Parameter setting screen for constructing prediction model, 1100 ... Future Predicted transaction amount of uncontracted transaction, 1200 ... Customer lifetime value, 1410 ... Client, 1420 ... Server, 1430 ... Communication network

─────────────────────────────────────────────────── ─── Continuation of the front page (51) Int.Cl. ⁷ Identification code FI theme code (reference) G06F 19/00 110 G06F 19/00 110 (72) Inventor Masahiro Morita 890 Kashimada, Sachiku, Kawasaki-shi, Kanagawa Stock Company Hitachi, Ltd. Business Solution Division (72) Inventor Yuko Morino 890 Kashimada, Sachi-ku, Kawasaki-shi, Kanagawa Stock Company (72) Seiichi Sugiyama Inventor, Seiichi Sugiyama 890 Kashimada, Sachi-ku, Kawasaki-shi, Kanagawa Hitachi, Ltd. Financial Systems Division

Claims (15)

[Claims] 1. A method for calculating a customer lifetime value of a financial institution by a computer, which receives customer information, receives account information for a customer of the customer information, and uses the account information until the future. If there is a contracted transaction for a financial instrument that contracts to continue the transaction, the future profit of the future contracted transaction of the customer is predicted, and the future profit of the future contracted transaction is discounted to the present value and totaled. , Calculate the customer lifetime value of future contracted transactions, use the customer information to predict future profits obtained from future uncontracted transactions for each period, and set the future profits of the future uncontracted transactions to the present value. A customer lifetime value calculation method characterized by calculating the customer lifetime value of future uncontracted transactions by discounting and totaling. 2. The transaction remaining period obtained by subtracting from the current balance and the interest rate of the future contracted transaction of the customer the transaction continuation period at the start of the future contracted transaction to the present, the future Calculate the future balance of the future contracted transactions by calculating the planned balance of the contracted transactions and multiplying the calculated planned balance by the previously registered transaction continuity probability and the planned margin. The method for calculating a lifetime value of a customer according to claim 1. 3. A model is created using the customer information as a prediction model for a forecast target product for each forecast period, the relationship between the customer balance and the customer information up to a point before the forecast period. By inputting the current customer information into the created prediction model, the expected balance of the forecast target product for each forecast period of the customer is calculated, and the expected balance is multiplied by the profit rate registered in advance. 2. The method of calculating a lifetime value of a customer according to claim 1, wherein the future profit of the future uncontracted transaction is calculated in a forecast period. 4. The customer lifetime value of the customer is calculated by taking the sum of the customer lifetime value of the future contracted transaction and the customer lifetime value of the future uncontracted transaction for each customer. The customer lifetime value calculation method according to claim 1. 5. A customer with high customer lifetime value is extracted as a sales promotion customer, and transaction items that are closely related to future profits of the future uncontracted transactions of the sales promotion customer are recommended products to the sales promotion customer. The method for calculating customer lifetime value according to claim 1 or 4, characterized in that. 6. A financial institution stores a customer lifetime value in a storage device for each customer, extracts a customer with a high customer lifetime value as a sales promotion target customer, and stores the customer in the storage device. A method of calculating a lifetime value of a customer, which displays recommended products that are financial products that have not been traded based on the customer transaction information of the existing financial products. 7. A customer lifetime value calculation system for a financial institution, comprising means for receiving customer information input, means for receiving account information input for the customer of the customer information, and future use of the account information. For financial instruments that are contracted to continue trading up to, if there is a contracted transaction, the means for predicting the future profit of the future contracted transaction of the customer and the future profit of the future contracted transaction discounted to the present value, and totaled. By using the customer information and means for calculating the customer lifetime value of future contracted transactions, means for predicting future profits obtained from future uncontracted transactions for each period and future profits of the future uncontracted transactions. A customer lifetime value calculation system having means for calculating a customer lifetime value of a future uncontracted transaction by discounting and summing the present value. 8. The future transaction period, which is obtained by subtracting, from the current balance and interest rate of the future contracted transaction of the customer, the scheduled transaction continuation period at the start of the future contracted transaction to the present, the future The future profit of the future contracted transaction is calculated by multiplying the means for calculating the planned balance of the contracted transaction, the calculated balance, the previously registered transaction continuation probability, and the expected margin. The customer lifetime value calculation system according to claim 7, further comprising means. 9. Using the customer information, a model is created for a prediction target product as a prediction model for each prediction period with a relationship between the customer's balance and customer information up to a point before the prediction period. By inputting the current customer information into the means and the created prediction model, means for calculating the expected balance of the forecast target product for each forecast period of the customer and the expected balance, the product of the profit rate registered in advance is calculated. The customer lifetime value calculation system according to claim 7, further comprising means for calculating future profit of the future uncontracted transaction in a forecast period. 10. A customer lifetime value of the future contracted transaction,
8. The customer lifetime value calculation system according to claim 7, further comprising means for calculating the customer lifetime value of the customer by taking the sum of the customer lifetime value of the future uncontracted transaction for each customer. 11. A means for extracting a customer having a high customer lifetime value as a sales promotion customer and a transaction item which is closely related to future profit of the future non-contracted transaction of the sales promotion customer, and a recommended product to the sales promotion customer. The customer lifetime value calculation system according to claim 7 or 10, further comprising: 12. A financial institution stores a customer lifetime value in a storage device for each customer, a means for extracting a customer having a high customer lifetime value as a sales promotion target customer, and a customer in the storage device. A customer lifetime value calculation system comprising means for displaying a recommended product, which is a financial product that has not yet been traded, from the stored customer transaction information of the financial product. 13. A program for a financial institution to calculate a lifetime value of a customer, the means for receiving a customer information by a computer; the means for receiving account information for a customer of the customer information; and the account information. For financial instruments that contract to continue the transaction to the future, if there is a contracted transaction, means for predicting future profits by future contractual transaction by contract period and future profits of the future contracted transaction, A means for calculating the customer lifetime value of a future contracted transaction by discounting to the present value and the customer information, and a means for predicting future profits obtained from future uncontracted transactions for each period and the future uncontracted By discounting the future profit of the transaction to the present value and adding,
By calculating the sum of the customer lifetime value of the future uncontracted transaction and the customer lifetime value of the future contracted transaction, and the customer lifetime value of the future uncontracted transaction for each customer, the customer lifetime value of the customer A customer lifetime value calculation program characterized by causing it to function as a means for calculating. 14. An input device, which is a data reference time indicating a time when data is stored, a product-by-product balance corresponding to the data reference time, and attribute information for each customer corresponding to the data reference time. And a customer information file containing the transaction status for each customer indicating the number of transactions and the number of transactions corresponding to the data reference time, an account information file containing the interest rate for each product corresponding to the data reference time, and the data reference time Corresponding to
A storage device that receives the margin of each product that a financial institution obtains from a customer, the target customer, the target product that is a target of uncontracted future transactions that are not contracted to continue in the future, and the forecast period Stores the customer information file, the account information file, the margin for each product, the prediction target customer, the prediction target product, and the prediction period, and the arithmetic processing device stores the customer information file from the customer information file. , By referring to the data reference time, to extract the product-specific balance of the customer corresponding to the specific prediction target customer, the arithmetic processing unit, out of the extracted product-based balance of the future Depending on the presence or absence of the balance of the product subject to the future contracted transaction contracted to continue, determine the presence or absence of the future contracted transaction, the arithmetic processing device, if there is the future contracted transaction, Previous By calculating the extracted balance for each product and the interest rate for each product, the expected balance for each product related to the future contracted transaction of the forecast target customer at the time of the prediction period is calculated, and the calculation process is performed. The device calculates the profit for each product related to the future contracted transaction, which is obtained from the prediction target customer at the time when the prediction period elapses, by accumulating the planned balance for each product and the margin for each product. Then, the arithmetic processing unit converts the profit for each product related to the future contracted transaction into a present value by using the interest rate for each product, and the arithmetic processing device is a product for the future contracted transaction. The customer lifetime value of the prediction target customer is calculated by summing the different present values, and the storage device stores the prediction target customer for the future contracted transaction. Store the customer lifetime value, the arithmetic processing device, from the customer information file, for the prediction target product, extracting the attribute information and transaction status of a plurality of customers, the arithmetic processing device, of the plurality of customers extracted Using the attribute information and the transaction status, create a prediction model, the arithmetic processing device extracts specific attribute information and the transaction status of the prediction target customer from the customer information file, and the arithmetic processing device, By applying the extracted attribute information and the transaction status to the prediction model, the expected balance of the prediction target product that is the target of the future uncontracted transaction at the time when the prediction period elapses is calculated, and the arithmetic processing device, By accumulating the expected balance of the forecast target product and the margin for each product, the future uncontracted transaction at the time when the forecast target period elapses Calculating the profit obtained from the measurement target customer, the processor obtains the profit obtained from the prediction target customer with respect to the future uncontracted transaction, by using the interest rate for each product, it is converted to the present value, The processing device calculates the customer lifetime value of the prediction target customer by summing the present value of each product of the future non-contract transaction, and the storage device calculates the prediction target customer regarding the future non-contract transaction. Stores the customer lifetime value of, the arithmetic processing device, from the storage device, the customer lifetime value of the future contracted transaction, and the customer lifetime value of the future uncontracted transaction, the arithmetic processing device, The customer lifetime value of the forecast target customer is calculated by summing the customer lifetime value of the future contracted transaction and the customer lifetime value of the future uncontracted transaction. Lifetime value calculation method. 15. A data reference time representing the time when data is stored, a product-specific balance corresponding to the data reference time for each customer, attribute information for each customer corresponding to the data reference time, and the data. A customer information file containing the transaction status of each customer representing the number of transactions and the number of transactions corresponding to the reference time, an account information file containing the interest rate of each product corresponding to the data reference time, and corresponding to the data reference time, An input device that receives an input of the profit margin of each product that a financial institution obtains from a customer, a target customer, a target product that is a target of a future uncontracted transaction that is not contracted to continue in the future, and a forecast period A storage device for storing the customer information file, the account information file, the margin for each product, the prediction target customer, the prediction target product, and the prediction period, and the customer information. By referring to the data reference time from the file, the balance by product of the customer corresponding to the specific target customer is extracted, and among the extracted balance by product, it is contracted to continue the transaction until the future. According to the presence or absence of the balance of the product subject to the future contracted transaction, the presence or absence of the future contracted transaction is determined, and if there is the future contracted transaction, the extracted balance for each product and the By calculating the interest rate for each product, the expected balance for each product regarding the future contracted transaction of the forecast target customer is calculated at the time when the prediction period elapses, and the expected balance for each product and the profit margin for each product are calculated. By calculating the profit for each product related to the future contracted transaction obtained from the forecast target customer at the time when the prediction period elapses, By calculating the profit for each product by using the interest rate for each product, it is converted to the present value, and by summing the present value for each product of the future contracted transaction, the customer lifetime value of the forecast target customer is calculated. Then, for the prediction target product from the customer information file,
Extracting attribute information and transaction status of a plurality of customers, for the prediction target product, create a prediction model using the extracted attribute information and transaction status of a plurality of customers, from the customer information file, of the prediction target customer By extracting specific attribute information and transaction status, and applying the extracted specific attribute information and transaction status to the prediction model, the prediction target product that is the target of the future uncontracted transaction at the time when the prediction period elapses. The expected balance of the forecast target product and the profit margin of each product, by accumulating the forecast balance, the profit obtained from the forecast target customer for the future uncontracted transaction at the time point of the forecast target period And the profit obtained from the forecast target customer for the future uncontracted transaction is calculated by using the interest rate for each product to calculate the present value. Converted to a value and calculating the customer lifetime value of the forecast target customer by summing the present value for each product of the future uncontracted transaction, the customer lifetime value of the future contracted transaction and the future uncontracted transaction A customer lifetime value calculation system including an arithmetic processing unit that calculates the customer lifetime value of the prediction target customer by adding up the customer lifetime values.