JPH08314893A - Complaint/inquiry analyzing system - Google Patents

Abstract

PURPOSE: To impart an information which is useful for improving customer service and for making improving plans for present job contents of an article planning/development, etc., by performing a simple data summing up processing of an analysis object file and visually outputting the result. CONSTITUTION: This system is provided with means 201 and 202 performing filtering of collected data and further adopting or rejecting the data for which the filtering is performed, a means 206 storing the adopted or rejected data in an analysis object file based on a preliminarily set code classification system and a means 208 performing the simple summing up processing 207 of the data of the analysis object file and visually outputting the result. The collected complaint/inquiry contents and an existing customer information data base are merged with a related key, the customer attribute information which is not collected in the conversation with a customer is included and the analysis object file is prepared. Based on the data of this analysis object file, the analysis of complaint/inquiry data is done. The analysis result is outputted by the frequency distribution of complaint/inquiry contents and the expression forms of correlation interested persons.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention analyzes complaints / inquiries received from customers by adding information from a database having a plurality of items to extract attributes of customers who make complaints / inquiries. To support customers who have complaints / inquiries, to respond to customers who may receive complaints / inquiries in the future, and to support the formulation of improvement measures for current business content such as product planning / development. Applicable complaints /
It relates to a query analysis system.

[0002]

2. Description of the Related Art Conventionally, regarding database marketing, which analyzes data stored in a database and performs marketing, 116 to 16 of "Arakawa: Database Marketing Strategy and Tactics: Diamond Company".
In Section 4, a method of analyzing data from various angles by using a decision making system capable of displaying tables and graphs on a relational database as a method of developing database marketing (hereinafter, referred to as the prior art 1
It is proposed). In addition, a system that supports Q & A response to complaints / inquiries from customers and a system that collects the history of complaints / inquiries and reflects them in product improvement (hereinafter referred to as conventional technology 2) have been in actual operation. ing.

[0003]

In the above-mentioned prior art, the information of the database having a plurality of items is taken into consideration, the attribute of the customer who makes the complaint / inquiry is extracted, and the customer's unique follow-up or potential existence exists. In order to proactively respond to customer complaints, there are the following problems.
First, in the above-mentioned conventional technique 1, the procedure of what angle to analyze and which item to combine and analyze is not taken into consideration. Especially when performing analysis using new items, the analyst It is necessary to analyze all the combinations, and when a large number of items are to be analyzed, a combination explosion occurs and there is a problem for the first problem that it cannot be analyzed manually. For example, when a two-dimensional graph is created, various combinations of items occur, so if the number of items is n, there will be 2 ⁿ combinations, and if a three-dimensional combination graph is considered, it will be 3 ⁿ . Then, the number of combinations is huge. Further, in the above-mentioned conventional technique 2, although it is possible to reflect Q & A to the customer who has made a complaint / inquiry and to improve the product based on the content of the complaint / inquiry, although no complaint / inquiry is currently sent, Responses to customers who have complaints / inquiries are not considered,
There is a problem for the second problem. The object of the present invention is to solve the above-mentioned problems, analyze the distribution of complaints / inquiries from customers, and, together with the characteristics of the customer making the complaints / inquiries, do not make any complaints / inquiries, but are potentially problematic. It is possible to identify customers who are supposed to have consciousness, and to provide information that is useful for improving customer service and formulating improvement measures for current business content such as product planning / development. To provide a query analysis system.

[0004]

In order to achieve the above object, a complaint / query analysis system of the present invention collects complaint or inquiry information and filters the collected data in a system for analyzing the information. ,
A means (201, 202) for further selecting filtered data, a means (206) for storing the selected data in an analysis target file based on a preset code classification system, and the analysis target file And a means (208) for visually outputting the result. Pre-analysis processing means (201 to 201) that merges the collected data with customer information that has been accumulated in advance (205) and generates an analysis target file that is the result of the merge processing.
206), a means (602) for analyzing a causal relationship between data from a database having a plurality of items when performing the information analysis, and a means for displaying rules in order from the information that best describes the characteristics of the customer. It is also characterized in that (801) is included. A means for visually outputting the information of the potential customer, using the information that best represents the characteristics of the customer as a key, and a reporting means or a data transmission means (21) for providing information to the relevant departments as an analysis result including the information of the potential customer.
0) and are also featured.

[0005]

In the present invention, by providing the information inputting means, the complaint / inquiry information from the customer that causes ambiguity or information loss can be displayed in 5W1H (What, When, Who, Where, Wh).
Since it can be arranged from the viewpoint of (y, How), it is possible to prevent the leakage of input information at the time of analysis. Further, by providing the analysis preprocessing means, it is possible to add the attribute information of the customer who has made a complaint / inquiry and to grasp the attribute of the customer who has not made a complaint / inquiry.
In addition, complaints /
It becomes possible to inform the analyst of the attributes of customers who make inquiries in a simple expression format. At the same time, the distribution status of complaints / customers making inquiries can be displayed in a three-dimensional graph to show complaints made by customers. / It is possible to visually understand the degree of deviation in the overall inquiry. Furthermore, by providing a means for providing the analysis result to the department in charge, the analysis result can be reflected on the spot speedily, and the improvement of customer service can be dealt with promptly. A system that supports the effective allocation of management resources such as sales strategy / product planning by analyzing complaints / inquiries and extracting the characteristics of customers who make complaints / inquiries and complaints / inquiries. Can be realized.

[0006]

Embodiments of the present invention will now be described in detail with reference to the drawings. In this example, an analysis support system was considered as the development of complaint / query analysis work in the insurance industry. FIG. 1 is an operation flowchart showing a complaint / query analysis work according to the embodiment of this invention. In the complaint / inquiry analysis work of this embodiment, first, complaints / inquiries received from customers are collected by telephone or a customer channel of a contact (phone, means such as visiting a company). At this time, the content of the conversation with the customer is coded and stored based on the classification system in which complaints / inquiries from the customer are preset (101). What is encoding?
Of the products, 01 for refrigerator, 02 for TV, 0 for vacuum cleaner
Out of the channels of complaints, the phone number is 1, the contact point is 2, the visitor is 3, and so on. Next, the collected complaint / inquiry content is merged with the existing customer information database by the associated key to create the analysis target file including the customer attribute information not collected in the conversation with the customer (102). . For example, in the case of insurance, merge is
Since the certificate number, address, telephone number, etc. are already stored in the database, you can retrieve other information from the database by searching with your name as a key during analysis. Is to increase. The complaint / query data is analyzed based on the data of this analysis target file (103).
As the analysis result, the frequency distribution of complaints / inquiries (for example, the format of FIG. 7) and the correlation (for example, the format of FIG. 10)
Or the distribution state of customers who have made complaints / inquiries is output in a representational form of a cubic graph (for example, the format of FIG. 9) or IF-THEN rules (for example, the format of FIG. 8). A person in charge of analysis sorts out the analysis results, creates information for providing information on the analysis results to each related department, and provides the analysis results to each related department by data transmission (104). At this time, the information necessary for accumulating long-term data and performing time-series analysis stores target data in a separate file for time-series analysis as part of the data analysis work (105). At this time, the information accumulated for time-series analysis is used again as analysis target information as needed.

FIG. 2 is a block diagram of a data analysis support system for supporting processing of complaints / inquiries according to the present invention. In FIG. 2, 201 and 202 are data collection processing units that assist in collecting information through conversation with customers, and 203 is a collection database that stores complaint / inquiry information according to a predetermined classification system. Is. The merge processing unit 205 supports the process of merging the data accumulated in the collection database 203 and the information in the customer information database 204 in which the existing contract information is accumulated. The output result of the merge processing unit 205 is the analysis target data file 20 as the analysis target data.
6 is stored. Next, the analysis target data is processed / edited by the data analysis processing unit 207 that supports the processing of the complaint / inquiry information, and the analysis result database 209 is created.
Here, since the analysis processing unit 207 is a process performed by a computer and cannot be understood without looking at the analysis result, the person in charge confirms this by outputting a confirmation list. An analysis result confirmation list 208 having a format such as a frequency distribution graph and IF-THEN rules is output to assist the person in charge of analysis until the analysis result data is created. While looking at the output of the analysis result confirmation list 208, the person in charge of analysis selects information to be provided to related departments as analysis results. Here, 210 is a processing unit that provides the analysis result information to the related departments, which is the analysis result data file 209.
Is a process of providing the data of (1) to each related department in the form of data transmission. Further, the analyst refers to the information in the analysis result file 209 to create the time series analysis file 212 using the support of the accumulation processing 211 in order to accumulate the information to be analyzed in time series. For example, data for January,
Data for February, etc. are accumulated. The time series analysis file 212 is used to perform time series analysis at any time.

Next, each processing of the data collection units 201 and 202, the merge processing unit 205, the analysis processing unit 207, the analysis result editing / transmission processing 210, and the time series accumulation processing 211 shown in the overall processing block diagram of FIG. The contents will be described in detail. FIG. 3 is a work flow chart of the data collection unit in FIG. The complaint / inquiry receptionist first listens to the complaint contents received by the customer through the telephone or the contact channel and responds to the customer (301). At this time, the receptionist accumulates the conversation content history with the customer by inputting on the screen based on a predetermined input system.
(302). Next, the receptionist performs the information filtering process (tuning process) 2 for the purpose of eliminating the ambiguity of the accumulated data, unentered items, and the data not to be analyzed.
After executing 01 (303), a conversation history is registered as data to be analyzed by screen input (304). At this time, in the information filtering process 201, the conversation history data accumulated from the viewpoint of 5W1H is evaluated, and the person in charge of acceptance selects the data. In particular, for conversation history data for which no mandatory analysis items have been entered, only the conversation history data of interest is displayed, providing a mechanism by which the person in charge of acceptance can tune. For example, the reason why they want it, who the sales staff name is, etc. (5W1H) are indispensable items. Therefore, if the customer does not say these, the person in charge always asks for them and registers.

FIG. 4 is a diagram of a file layout of information collected by the data collection unit of the present invention. In FIG. 4, information is stored by defining the viewpoint of 5W1H as follows. The data 401 that can specify the customer who has made a complaint / inquiry is defined as WHO, WHERE. here,
If it is a person's name, it will be WHO. Also, the complaint / inquiry reception date / time 402 is WHEN, the specific complaint / inquiry target and content 403 (what the inquiry is, what the inquiry is) WHAT, WHY, and customer improvement request 404.
Is defined as HOW. On the other hand, 203 to 20 shown in FIG.
6 is a series of processing blocks for the merge process. In the merge process, the collected customer conversation history information and the data in the customer information database that stores the attribute information of the customers who have already contracted are merged using the security symbol number described in the insurance policy as a key. . FIG. 5 is a diagram showing a file layout image of the analysis target file after the merge processing. In FIG. 5, 501 is the history information data obtained by the conversation with the customer, and 502 is the data stored in the customer information database (that is, the data added by being merged), that is, the address, gender, and product name. , Payment category, and other data.

FIG. 6 is a flow chart of analysis work in complaint / query analysis. The analyst first confirms the distribution status of the data by complaint / inquiry content by using the frequency distribution graph (601). The horizontal axis shows the analysis target items defined in the complaint / inquiry data collection system, and the vertical axis shows the number of complaint / inquiry data actually received (see FIG. 7) to grasp the weight of complaint / inquiry. Figure 7 focuses on the sales staff as a complaint / inquiry, and complains about the sales staff /
It is a figure which shows the example which confirms the distribution situation of the customer according to the inquiry content. In this example, complaints about visiting sales staff /
It turns out that there are a lot of inquiries, and a deeper analysis of the sales staff visits reveals good things. This method is performed for the number of complaint / inquiry content items, and the first stage of complaint / inquiry content is narrowed down. Next, in order to determine the priority of problem solving, the analyst refers to the distribution state of complaints / inquiries received by the customer and grasps the characteristics of the customer in which the deviation of the distribution particularly appears. There is a need
(602). However, in order to understand the characteristics of the customer who is making the complaint / inquiry, all combinations of the attribute information stored in the customer information are implemented. For example, it is assumed that the number of customer attribute fields in the database is 40, and the number of categories in each item is 3. The number of customer attribute fields is each attribute such as name, age, sex, address, work place, ... Stored in the database. Here, the category number 3 means that the attribute "age" is grouped into three units of "20 years old or less", "21 years old or more and 40 years old or less", and "41 years old or more". . At this time, there are 3 ⁴ combinations of customer attributes, and it is necessary to compare many combination conditions and grasp the characteristics of the customer.

In the complaint / inquiry system of the present embodiment, in order to prevent the combination explosion, the analyst can specify a combination condition of customer attributes. Also,
In this system, as a method to grasp the characteristics of the customer,
The IF-THEN rule format shown in FIG. 8 is adopted,
Preventing the combination of customer attributes means narrowing down the number of rules. For example, while there are three axes of insurance premium, complaint / inquiry content, and product, the IF-THEN rule limits to two axes, and the number of combination conditions (number of axes) is arbitrarily limited depending on the expression format. can do. As an index for narrowing down the number of rules, this system defines the concept of concentration. The density is a ratio of the number of data Rn of target data included in the data to the number Tn of data that hit a plurality of conditions. When the concentration is Dn, Dn is shown by the following equation. Dn = Rn / Tn (1) The concentration Dn may be considered as the precision rate of data by various combinations, and can be an index when narrowing down the number of rules. For example, the target of the rule used by the analyst is 30
Is specified, the rules in the upper 30 range of Dn are targeted.

In the case of the database shown above, IF A ₁
The THENB rule has 40 customer attribute fields, and since this number is divided into three groups (three axes), the total is 3 × 40 = 120. Further, the number of rules of IF A ₁ and A ₂ THENB is a combination of two rules selected from the number of rules 30, so ₃₀ C
₂ = 30 × 29/2 = 435 ways. That is, A ₁
When A ₂ and A ₂ are the two axes of name and address, it is a combination of selecting two from 30 items. A ₁ A ₂ ... An (N ≦
30) and the number of conditional clauses in the IF section increase, the maximum 30 types of customer attributes initially selected become max, and the combination pattern decreases at a certain fixed value. That is, IF
The number of conditional clauses increases, and when the maximum value is 40, 1
There is only one, and when there are 39, there is one remaining,
There are 40 ways, and as the number of conditional clauses is reduced in this way, the value is successively reduced. In other words, when the number of conditional clauses is increased from one, it increases at first, but it decreases from a certain point. Therefore, 3 ⁴⁰ > 120 + 435 +
----, and it is possible to prevent combination explosion.
Further, the characteristics of the customer to be discovered are also clarified because the combination of the higher density Dn is searched.

The method of confirming the distribution state of complaint / query data using the screen interface realized by the system of the present invention is shown in the IF-THEN rule format output example of FIG. 8 and the three-dimensional graph output example of FIG. It will be specifically described based on the above. When the person in charge of analysis executes the analysis of this system to grasp the distribution state of complaints / inquiries, FIG. 8 and FIG.
Display the analysis result of. From FIG. 8, 80% of customers who have received complaints / inquiries about sales staff and have insurance premiums of 20,000 yen or more and have a fixed lifetime insurance (40%)
It is understood that it occupies (/ 50) (801). This 80
% Indicates the concentration (802), and here, out of the customers (50) who have made complaints / inquiries to the sales staff, the insurance premium is 20,000 yen or more and the fixed lifetime insurance is applied. Existing customers
This shows that the ratio of 40 people is 80%. Further, since the IF to THEN rules are displayed in descending order of density, it is relatively easy to understand the characteristics of the customer who is making a complaint / inquiry.

In FIG. 9, it is possible to visually and easily grasp the distribution state of customers who make a complaint / inquiry shown in FIG. In particular, being able to visually understand the biased state of customers with respect to the distribution of all customers who have made complaints / inquiries improves the work efficiency of the analyst.
In other words, if insurance premiums, products, and complaints / inquiries are used as the three axes, those of customers who have insurance premiums of 20,000 yen or more and whose products are covered by the fixed lifetime insurance are listed at 801. You can see that you are concentrated. The three-dimensional graph display is I
It is a mechanism that can be displayed by operating the screen at the same timing as the screen of the F-THEN rule display. Further, in order to improve the analysis accuracy, it is necessary to reduce the items used in the analysis. When carrying out the analysis, the items in the conditional section (IF section) shown in Fig. 8 that show the same bias in the actual data may have a strong correlation between the data, and are therefore estimated to be reduction items. To be done.

FIG. 10 is a diagram showing an example in which analysis items can be reduced by using the correlation example of the present invention. That is, FIG.
0 is the information held in the customer information database,
It shows the correlation between the segment information of “age” and “contract period”. As shown in the figure, the data of “age” and “contract period” are highly correlated, and one of the items from the symmetry axis of the analysis is shown. It shows that it can be reduced. Next, in the analysis work flow of FIG. 6, the person in charge of analysis needs to immediately transmit the analysis result to the relevant department (603). In this case, the analysis result file 209 and the analysis result compilation transmission processing unit 210 shown in FIG. 2 are processing blocks that provide the analysis result to the related departments. The person in charge of analysis edits / processes the analysis result using the reporting function provided by the system of the present invention, and transmits the data to each related department using batch processing. In editing the analysis result, IF-THEN shown in FIG.
Customers who use the attribute information of customers who make complaints / inquiries grasped in the rule format and do not make any complaints / inquiries at present, but may make complaints / inquiries in the near future (potential customers) The information of the list is also output. IF in FIG.
A specific example is shown based on the THEN rule format example. The contents of the customer information database are searched under the conditions shown in the IF section, "Insurance premium is 20,000 yen" and "I have a fixed life insurance policy", and the information that is output next (security code number, customer Name,
The information obtained by deleting the customer who is currently making a complaint / inquiry from the information that can specify the customer such as the address) becomes the potential customer information.

11 and 12 are diagrams showing information images provided to specific related departments. FIG. 11 shows the provision data 1100 regarding the visited list, and FIG. 12 shows the provision data 1110 regarding the product improvement proposal. Data 1100 related to Mr. A 110
Reference numeral 1 is correspondence information for a customer who is currently making a complaint / inquiry, and 1102 regarding Mr. B is correspondence information for a potential customer. Regarding the product A of the data 1110, regarding the complaint contents that the dividend is not good, the improvement method includes a review of the dividend method, and it is shown that the number of cases related to this was 100. In FIG. 6, finally, the analyst accumulates the analysis data used this time in the time series analysis file 212 in order to carry out the medium-term analysis (604). The analysis target file 206 shown in FIG.
Analysis result file 209, time series analysis accumulation processing unit 21
1 and the time series analysis minute file 212 are the time series analysis file creation processing block. The analyst selects the analysis data used this time by using the analysis result information of this time, and accumulates the analysis target information.

[0017]

As described above, according to the present invention,
By providing the data collection processing unit, it is possible to filter the ambiguity of complaints / inquiries received from customers for data analysis, improve the accuracy of the analysis target data, and perform one-dimensional analysis (frequency distribution), Multidimensional analysis (IF
..- THEN rule format), it is possible to improve the accuracy and efficiency of the analysis by gradually narrowing down the data, and to deal with potential customers. Furthermore, by providing a means for transmitting analysis results, it is possible to smoothly and speedily provide information to related departments, and by providing a means for accumulating data for time-series data analysis, medium- to long-term analysis is possible. Is.

[Brief description of drawings]

FIG. 1 is an overall flowchart of a complaint / query analysis work using an analysis support system according to an embodiment of the present invention.

FIG. 2 is an overall processing block diagram of an analysis support system showing an embodiment of the present invention.

FIG. 3 is a processing flowchart of a data collection unit in FIG.

FIG. 4 is a diagram showing an example of collected data of a data collecting unit in FIG.

5 is a diagram showing an example of analysis target data in FIG.

FIG. 6 is an operation flowchart of analysis work according to the present invention.

FIG. 7 is a frequency distribution chart used in the present invention.

FIG. 8 is a diagram showing an example of IF-THEN rule format output used in the present invention.

FIG. 9 is a diagram showing an example of a three-dimensional graph output used in the present invention.

FIG. 10 is a correlation diagram showing items to be reduced in the present invention.

FIG. 11 is an image diagram of information provision data (visit list) to related departments in the present invention.

FIG. 12 is an image diagram of information provision data (product improvement proposal) to related departments.

[Explanation of symbols]

201 ... Information filtering processing unit, 202 ... Collected data storage processing unit, 203 ... Collected data file, 204 ...
Customer information file, 205 ... Merge processing section, 206 ...
Analysis target file, 207 ... Analysis processing unit, 208 ... Analysis result confirmation list, 209 ... Analysis result file, 210
... analysis result edit transmission processing unit, 211 ... time series analysis storage processing unit, 212 ... time series analysis file.

 ─────────────────────────────────────────────────── ─── Continuation of the front page (72) Inventor Tatsuyuki Arao 12 Hitachi, Ltd. Information Systems Division, 890 Kashimada, Sachi-ku, Kawasaki-shi, Kanagawa

Claims (3)

[Claims] 1. In a system for collecting complaint or inquiry information and analyzing the information, a means for filtering the collected data and further selecting the filtered data, the selected data in advance A means for accumulating in the analysis target file based on the set code classification system, and performing a simple totalization process of the data of the analysis target file,
A complaint / query analysis system comprising means for visually outputting a result. 2. The complaint / query analysis system according to claim 1, wherein the collected data is merged with customer information accumulated in advance, and a pre-analysis process for generating an analysis target file as a result of the merge process. Means, a means for analyzing a causal relationship between data from a database having a plurality of items when carrying out the information analysis, and a means for displaying rules in order from the information that best describes the characteristics of the customer. Complaint / query analysis support system. 3. The information analysis processing according to claim 2, wherein means for visually outputting the information of the potential customer is used as a key, with the information that best represents the characteristics of the customer as a key, and as an analysis result including the information of the potential customer. A complaint / query analysis support system having a reporting means or a data transmission means for providing information to related departments.