Method and system for improving accuracy of customer experience analysis result
By classifying and modeling the customer data of telecom operators, the problem of inaccurate customer experience analysis is solved, more accurate and timely customer experience analysis is achieved, and personalized service recommendations and customer management are supported.
Patent Information
- Application Number
- CN202510886607.2
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-06-30
- Publication Date
- 2025-10-28
Smart Images
Figure CN120851944A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of big data processing technology, specifically to a method and system for improving the accuracy of customer experience analysis results. Background Technology
[0002] In the telecommunications industry, operators focus on improving user experience and service quality, intelligently diagnosing the reasons for low customer satisfaction, providing personalized service recommendations, and building an evaluation system to measure customer experience and service quality, thereby increasing customer satisfaction. Telecommunications operators possess vast amounts of network and customer data; therefore, accurately and promptly understanding customer experience is crucial.
[0003] Operators possess a wide range of data, but lack a mature analytical system to integrate and leverage this source data. Therefore, a design solution is needed to categorize customers, enabling targeted analysis using specific source data. This allows for the creation of different customer experience analysis models for specific scenarios, continuously outputting quantifiable customer experience metrics to provide data support for decision-making in customer marketing, customer retention, and other activities. Summary of the Invention
[0004] The purpose of this invention is to provide a method and system for improving the accuracy of customer experience analysis results, so as to solve the problems mentioned in the background art.
[0005] To achieve the above objectives, the present invention provides the following technical solution: a method for improving the accuracy of customer experience analysis results, comprising the following steps: designing customer data and data sources, dividing the source data into network data and customer data; network data reflects the network quality of the operator, including handover success rate, call drop rate, and wireless resource control establishment success rate; customer data is directly related to customers and is divided into basic customer data, customer behavior data, and customer consumption data; basic customer data includes name, age, and network access time; customer behavior data includes number of calls, internet traffic, subscribed packages, and recharge payments; and customer consumption data includes consumption amount, deposit amount, and installment payment amount.
[0006] Preferably, the method also includes steps to improve the quality of source data: clarifying that network data comes from the operator's performance management system, basic customer data comes from the operator's CRM system, customer behavior data comes from the operator's CRM system and billing system, and customer consumption data comes from the operator's billing system; processing the source data, including filling in missing key customer information fields, selecting data to retain for appropriate periods according to different analysis processes, deduplicating the source data, and expanding more data based on the existing source data, wherein the expanded data includes expanded customer indicator data and expanded customer tags.
[0007] Preferably, the method also includes the step of creating a customer classification model: using an RFM model combined with a binary classification method and a K-Means clustering algorithm to perform an initial classification of customers, resulting in 8 customer classification results; adding new indicators to eliminate customer differences based on the initial classification, and determining the second classification of customers; using a random forest model to complete a churn prediction model to identify customers who are about to churn; determining customer lifecycle based on the customer classification with eliminated differences and customers who are about to churn; based on the above steps, 8 customer categories are obtained according to customer value and 5 customer categories are obtained according to customer lifecycle, and each category can be subclassified according to user tags.
[0008] Preferably, the method also includes the step of creating a customer experience analysis model: determining the analysis model corresponding to different customer categories, with differences in input parameters, input parameter weights, and analysis algorithms / formulas among different analysis models; determining that the input of the analysis model is network data and customer data; and designing the output of the analysis model as a score between 0 and 100 and the probability of that score.
[0009] Preferably, it also includes the step of creating an analysis process template: the analysis process template is used to automate the analysis process; the template includes the template name, steps, the order of execution of steps, the actions performed in the steps, and key attributes of the template's running cycle.
[0010] A system for improving the accuracy of customer experience analysis results includes: a data design module for designing customer data and data sources, dividing the source data into network data and customer data; wherein, network data reflects the quality of the operator's network, covering handover success rate, call drop rate, and wireless resource control establishment success rate; customer data is directly related to customers and is divided into basic customer data, customer behavior data, and customer consumption data. Basic customer data includes name, age, and network access time; customer behavior data includes number of calls, internet traffic, subscribed packages, and recharge payments; and customer consumption data includes consumption amount, deposit amount, and installment payment amount.
[0011] Preferred options also include:
[0012] The data quality improvement module is used to acquire and process source data.
[0013] The data quality improvement module includes:
[0014] The data source determination unit is used to determine that network data comes from the operator's performance management system, basic customer data comes from the operator's CRM system, customer behavior data comes from the operator's CRM system and billing system, and customer consumption data comes from the operator's billing system.
[0015] The data processing unit is used to process the source data, including filling in missing key customer information fields, selecting data to retain for appropriate periods according to different analysis processes, deduplicating the source data, and expanding the source data with extended customer indicator data and extended customer tags.
[0016] Preferred options also include:
[0017] The customer classification model creation module is used to create customer classification models;
[0018] The customer classification model creation module includes:
[0019] The initial classification unit is used to perform initial classification of customers using the RFM model combined with the binary method and K-Means clustering algorithm, resulting in eight customer classification results: important value customers, recently churned customers, high-frequency loyal customers, important retained customers, potential important customers, important development customers, low value customers, and churned customers.
[0020] The secondary classification unit is used to add new indicators to eliminate customer differences based on the initial classification and determine the customer's second classification.
[0021] The churn prediction unit is used to complete the churn prediction model using the random forest model to identify customers who are about to churn.
[0022] The lifecycle determination unit is used to determine the customer lifecycle based on customer segments that eliminate differences and customers who are about to churn.
[0023] The classification result output unit is used to output 8 customer categories based on customer value and 5 customer categories based on customer lifecycle, based on the above steps, and each category can be subcategorized by user tags.
[0024] Preferred options also include:
[0025] The Customer Experience Analysis Model Creation Module is used to create customer experience analysis models.
[0026] The customer experience analytics model creation module includes:
[0027] The model determination unit is used to determine the analysis model corresponding to different customer categories. Different analysis models differ in input parameters, input parameter weights, and analysis algorithms / formulas.
[0028] The input determination unit is used to determine whether the input to the analysis model is network data and customer data;
[0029] The output design unit is used to design the output of the analysis model as a score between 0 and 100 and the probability of that score.
[0030] Preferred options also include:
[0031] The analysis process template creation module is used to create analysis process templates;
[0032] The analysis process template creation module includes:
[0033] The automation implementation unit is used to automate the analysis process through analysis process templates;
[0034] The attribute setting unit is used to set the key attributes contained in the template. The key attributes include the template name, the steps, the order in which the steps are executed, the actions performed in the steps, and the template running cycle.
[0035] Compared with the prior art, the beneficial effects of the present invention are:
[0036] The method and system proposed in this invention for improving the accuracy of customer experience analysis results clearly define the process of analyzing customer experience and can use different analysis models through customer classification. The analysis process template facilitates continuous analysis of customer experience, thereby improving the accuracy and timeliness of customer experience analysis. Attached Figure Description
[0037] Figure 1 This is a flowchart of the method of the present invention. Detailed Implementation
[0038] To make the objectives, technical solutions, and advantages of the present invention clear and complete, the embodiments of the present invention will be further described in detail below with reference to the accompanying drawings. It should be understood that the specific embodiments described herein are only some, not all, embodiments of the present invention, and are merely illustrative of the embodiments of the present invention. They are not intended to limit the embodiments of the present invention. All other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.
[0039] Example 1, please refer to Figure 1 This invention provides a technical solution: a method for improving the accuracy of customer experience analysis results, comprising the following steps:
[0040] a. Design customer data and data sources.
[0041] Determining which data to use to analyze customer experience is fundamental to the analysis and serves as input for the analytical model. This solution categorizes source data into network data and customer data. Network data reflects the quality of the operator's network, which affects the customer experience when using the operator's services. For example, network data includes handover success rate, call drop rate, and wireless resource control establishment success rate. Network quality affects the experience of multiple customers within a certain area. Customer data is data directly related to the customer and can directly describe the data of a specific customer. It is divided into basic customer data, customer behavior data, and customer consumption data. For example, basic customer data includes name, age, and network access time; customer behavior data includes number of calls, data usage, subscribed packages, and top-up payments; customer consumption data includes purchase amount, deposit amount, and installment payment amount.
[0042]
[0043] b. Improve the quality of source data
[0044] In this solution, network data comes from the operator's performance management system, basic customer data comes from the operator's CRM system, customer behavior data comes from the operator's CRM system and billing system, and customer consumption data comes from the operator's billing system.
[0045] The source data is processed to make it more suitable for the needs of the analysis model.
[0046] Data backfilling refers to filling in missing key customer information fields, such as IMSI, MSISDN, and IMEI data, which may be missing and need to be supplemented.
[0047] Depending on the specific analysis process, select and retain data for appropriate time periods. For example, exclude quiet periods and retain only active data from 8:00 AM to 12:00 AM local time.
[0048] The source data is deduplicated.
[0049] Expand upon existing source data to include more data, such as extended customer metrics and extended customer tags. For example, the "local call duration" metric can be expanded to include "average local call duration over the past 3 days" and "month-on-month change in local call duration." Customer tags include "Voice customers from XX country," "consistently high ARPU customers," and "gaming customers."
[0050] Different analytical models use different source data. More source data is not necessarily more accurate; different source data should be selected for different analytical models.
[0051] c. Create a customer classification model
[0052] After classifying customers multiple times using the RFM model, the first classification results are obtained, which mainly include eight categories: high-value customers, recently churned customers, high-frequency loyal customers, important retained customers, potential high-value customers, important development customers, low-value customers, and churned customers. The RFM model uses a binary search method and the K-Means clustering algorithm.
[0053] After obtaining the initial customer classification, new metrics are added to eliminate customer discrepancies, and a second classification is used to determine the customer's category. For example, customers in the same "high-value customer" category may have significant differences in a certain metric, and the customer classification needs to be redefined based on that metric.
[0054] A random forest model is used to create a churn prediction model to identify customers who are about to churn.
[0055] Determine the customer lifecycle based on the customer classification for eliminating differences in b) and the customers about to churn in c).
[0056] Based on the above steps, several customer classifications can be obtained. According to customer value, there are 8 undifferentiated customer categories. According to the customer lifecycle, there are 5 customer categories: acquisition stage, upgrade stage, maturity stage, decline stage, and churn stage.
[0057] Each category can be further subdivided based on user tags, including upgrade plan users, video users, broadcaster users, game users, home IoT value-added users, and low-quality router users.
[0058] d. Create a customer experience analysis model
[0059] Different analytical models are needed when analyzing the customer experience of customers in different customer categories. Therefore, it is necessary to determine the corresponding analytical models for different customer categories. Different analytical models refer to different input parameters, different input parameter weights, and different analytical algorithms / formulas.
[0060] Determine the inputs to the analytical model, including network data and customer data;
[0061]
[0062]
[0063]
[0064] The output of the analytical model should be designed as a score between 0 and 100, along with the probability of that score. For example, if customer Zhang San's customer satisfaction score is 95, the probability is 89%.
[0065] e. Create an analysis process template
[0066] Automation of the analysis process is achieved through analysis process templates;
[0067] The template includes key attributes such as template name, stages, the order in which stages are executed, the actions performed in each stage, and the template's execution cycle. An example of an analysis process template:
[0068]
[0069] Example 2, based on Example 1, proposes a system for improving the accuracy of customer experience analysis results according to claim 5, comprising: a data design module for designing customer data and data sources, dividing the source data into network data and customer data; wherein, network data reflects the operator's network quality, covering handover success rate, call drop rate, and wireless resource control establishment success rate; customer data is directly related to customers and is divided into basic customer data, customer behavior data, and customer consumption data, wherein basic customer data includes name, age, and network access time, customer behavior data includes number of calls, internet traffic, subscribed packages, and recharge payments, and customer consumption data includes consumption amount, deposit amount, and installment payment amount.
[0070] It also includes: a data quality improvement module, used to acquire and process source data; the data quality improvement module includes: a data source determination unit, used to determine that network data comes from the operator's performance management system, basic customer data comes from the operator's CRM system, customer behavior data comes from the operator's CRM system and billing system, and customer consumption data comes from the operator's billing system; and a data processing unit, used to process the source data, including filling in missing key customer information fields, selecting data to retain for appropriate periods according to different analysis processes, deduplicating source data, and expanding extended customer indicator data and extended customer tags based on existing source data.
[0071] It also includes: a customer classification model creation module, used to create customer classification models; the customer classification model creation module includes: an initial classification unit, used to use the RFM model combined with the binary classification method and the K-Means clustering algorithm to perform initial classification of customers, obtaining 8 customer classification results: important value customers, recently churned customers, high-frequency loyal customers, important retained customers, potential important customers, important development customers, low-value customers, and churned customers; a secondary classification unit, used to add new indicators to eliminate customer differences based on the initial classification, and determine the second classification of customers; a churn prediction unit, used to use the random forest model to complete the churn prediction model and determine customers who are about to churn; a lifecycle determination unit, used to determine the customer lifecycle based on the customer classification that eliminates differences and customers who are about to churn; and a classification result output unit, used to output 8 customer categories based on customer value and 5 customer categories based on customer lifecycle based on the above steps, and each category can be subclassified by user tags.
[0072] It also includes: a customer experience analysis model creation module, used to create customer experience analysis models; the customer experience analysis model creation module includes: a model determination unit, used to determine the analysis model corresponding to different customer categories, different analysis models have differences in input parameters, input parameter weights, analysis algorithms / formulas; an input determination unit, used to determine that the input of the analysis model is network data and customer data; and an output design unit, used to design the output of the analysis model as a score between 0 and 100 and the probability of that score.
[0073] It also includes: an analysis process template creation module, used to create analysis process templates; the analysis process template creation module includes: an automation implementation unit, used to automate the analysis process through the analysis process template; and an attribute setting unit, used to set the key attributes contained in the template, the key attributes including template name, steps, the order of execution of steps, the actions performed in the steps, and the template running cycle.
[0074] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.
Claims
1. A method for improving the accuracy of customer experience analysis results, characterized in that: The following steps are involved: The design of customer data and data sources divides the source data into network data and customer data. Network data reflects the operator's network quality, including handover success rate, call drop rate, and wireless resource control establishment success rate. Customer data is directly related to customers and is divided into basic customer data, customer behavior data, and customer consumption data. Basic customer data includes name, age, and network access time. Customer behavior data includes number of calls, data usage, subscription packages, and recharge payments. Customer consumption data includes consumption amount, deposit amount, and installment payment amount.
2. The method for improving the accuracy of customer experience analysis results according to claim 1, characterized in that: It also includes steps to improve the quality of source data: clarifying that network data comes from the operator's performance management system, basic customer data comes from the operator's CRM system, customer behavior data comes from the operator's CRM system and billing system, and customer consumption data comes from the operator's billing system; processing the source data, including filling in missing key customer information fields, selecting and retaining data for appropriate periods according to different analysis processes, deduplicating the source data, and expanding more data based on the existing source data, wherein the expanded data includes expanded customer indicator data and expanded customer tags.
3. The method for improving the accuracy of customer experience analysis results according to claim 2, characterized in that: The process also includes the steps of creating a customer classification model: using the RFM model combined with the binary classification method and the K-Means clustering algorithm to initially classify customers, resulting in 8 customer classification results; adding new indicators to eliminate customer differences based on the initial classification, and determining the second classification of customers; using the random forest model to complete the churn prediction model and identify customers who are about to churn; determining the customer lifecycle based on the customer classification with eliminated differences and the customers who are about to churn; based on the above steps, 8 customer categories are obtained according to customer value and 5 customer categories are obtained according to customer lifecycle, and each category can be subclassified according to user tags.
4. The method for improving the accuracy of customer experience analysis results according to claim 3, characterized in that: It also includes the steps of creating a customer experience analysis model: determining the analysis model corresponding to different customer categories, with differences in input parameters, input parameter weights, and analysis algorithms / formulas between different analysis models; determining that the input of the analysis model is network data and customer data; and designing the output of the analysis model as a score between 0 and 100 and the probability of that score.
5. The method for improving the accuracy of customer experience analysis results according to claim 4, characterized in that: It also includes the step of creating an analysis process template: the analysis process template automates the analysis process; the template includes the template name, steps, the order of steps, the actions performed in each step, and key attributes of the template's running cycle.
6. A system for using the method for improving the accuracy of customer experience analysis results according to claim 5, characterized in that: include: The data design module is used to design customer data and data sources, dividing the source data into network data and customer data. Among them, network data reflects the operator's network quality, covering handover success rate, call drop rate, and wireless resource control establishment success rate. Customer data is directly related to customers and is divided into basic customer data, customer behavior data, and customer consumption data. Basic customer data includes name, age, and network access time. Customer behavior data includes number of calls, data usage, subscription packages, and recharge payments. Customer consumption data includes consumption amount, deposit amount, and installment payment amount.
7. The system according to claim 6, characterized in that: Also includes: The data quality improvement module is used to acquire and process source data. The data quality improvement module includes: The data source determination unit is used to determine that network data comes from the operator's performance management system, basic customer data comes from the operator's CRM system, customer behavior data comes from the operator's CRM system and billing system, and customer consumption data comes from the operator's billing system. The data processing unit is used to process the source data, including filling in missing key customer information fields, selecting data to retain for appropriate periods according to different analysis processes, deduplicating the source data, and expanding the source data with extended customer indicator data and extended customer tags.
8. The system according to claim 7, characterized in that: Also includes: The customer classification model creation module is used to create customer classification models; The customer classification model creation module includes: The initial classification unit is used to perform initial classification of customers using the RFM model combined with the binary method and K-Means clustering algorithm, resulting in eight customer classification results: important value customers, recently churned customers, high-frequency loyal customers, important retained customers, potential important customers, important development customers, low value customers, and churned customers. The secondary classification unit is used to add new indicators to eliminate customer differences based on the initial classification and determine the customer's second classification. The churn prediction unit is used to complete the churn prediction model using the random forest model to identify customers who are about to churn. The lifecycle determination unit is used to determine the customer lifecycle based on customer segments that eliminate differences and customers who are about to churn. The classification result output unit is used to output 8 customer categories based on customer value and 5 customer categories based on customer lifecycle, based on the above steps, and each category can be subcategorized by user tags.
9. A system according to claim 8, characterized in that: Also includes: The Customer Experience Analysis Model Creation Module is used to create customer experience analysis models. The customer experience analytics model creation module includes: The model determination unit is used to determine the analysis model corresponding to different customer categories. Different analysis models differ in input parameters, input parameter weights, and analysis algorithms / formulas. The input determination unit is used to determine whether the input to the analysis model is network data and customer data; The output design unit is used to design the output of the analysis model as a score between 0 and 100 and the probability of that score.
10. A system according to claim 9, characterized in that: Also includes: The analysis process template creation module is used to create analysis process templates; The analysis process template creation module includes: The automation implementation unit is used to automate the analysis process through analysis process templates; The attribute setting unit is used to set the key attributes contained in the template. The key attributes include the template name, the steps, the order in which the steps are executed, the actions performed in the steps, and the template running cycle.