A method and system for calculating operator customer life cycle
Through the use of K-mean clustering algorithm and key indicators, the operator's customer life cycle is segmented, the problem of the RFM model cannot be segmented is solved, more accurate customer segmentation and resource allocation is achieved, and customer business value and life are improved. cycle.
Patent Information
- Application Number
- CN202110113184.2
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2021-01-27
- Publication Date
- 2024-11-26
- Estimated Expiration
- 2041-01-27
AI Technical Summary
The existing RFM model cannot effectively segment the life cycle of telecom operator customers, cannot distinguish between growth, maturity, and sleepy periods, resulting in the inability to make differentiated marketing and decision-making, and the customer group is complex and is not suitable for statistical analysis based on the number of consumption times. .
The operator customer life cycle calculation method based on the K-mean clustering algorithm is adopted. By determining the number of on-network products, average on-network time, arpu value and off-grid products, the K-mean clustering algorithm is used to divide customers into mature period, dormant period and growth period, distinguish between public and corporate clients, and adjust the critical value according to data similarity.
It achieves more accurate customer segmentation, reasonably allocates resources, improves customer business value, and extends customer life cycle, solving the problem of inapplicable indicators of traditional RFM model and data distortion caused by customer characteristics differences.
Smart Images

Figure CN112819520B_ABST
Abstract
Description
Technical Field
[0001] The present invention belongs to the technical field of lifecycle calculation, and relates to a method and system for calculating the lifecycle of an operator's customer, and specifically to a method and system for calculating the lifecycle of an operator's customer based on a K-means clustering algorithm. Background Art
[0002] The customer life cycle refers to the period from when a customer begins to understand a company or when the company wants to develop a customer until the business relationship between the customer and the company is completely terminated and all related matters are completely handled. The customer life cycle describes the overall characteristics of the customer relationship moving from one state (one stage) to another state (another stage).
[0003] At present, the RFM model is widely used in the Internet customer life cycle. The RFM model is an important tool and means to measure customer value and customer profitability. The model describes the customer's life cycle and value status through three indicators: a customer's recent purchase behavior, overall purchase frequency, and how much money he spent:
[0004] R (Recency) is the time from the last consumption to the present. The more recent the consumption time, the greater the customer value. A user who consumed 1 year ago is definitely not as valuable as a user who consumed 1 week ago.
[0005] F (Frequency) Consumption frequency: Consumption frequency refers to the number of times a user purchases goods within a statistical period; users who purchase frequently are called regular customers, and their value is definitely greater than that of customers who come only occasionally.
[0006] M (Monetary) Consumption amount: The consumption amount refers to the total amount of consumption by the user during the statistical period, reflecting how much profit the consumer creates for the company; naturally, the more the user consumes, the greater the value.
[0007] The following is an example of an APP using the RFM model:
[0008] stage definition RFM Algorithm Marketing tools potential Contact but no purchase, retain personal information R=0,M=0,F=0 Traffic generation experience conversion first purchase newborn First purchase within 30 days 0<R<=30,F=1 Transform to effective efficient Users who purchase again within three months of their first purchase 0<R<=90,F>=2 active Consecutive consumption for more than 3 times within 4 months 0<R<=120,F>=3 loyalty Consumed more than 4 times in a row within 6 months 0<R<=180,F>=4 Sleep No further consumption within six months R>180, F=0 wake die No re-consumption within one year R>360, F=0 Give up, stop investing
[0009] RFM is a data-driven customer segmentation technique that helps marketers make smarter strategic decisions; it enables marketers to quickly identify and segment users into similar groups and develop differentiated and personalized marketing strategies for them.
[0010] However, telecom operators' customer life cycle segmentation cannot meet the complex analysis needs of business development. They can only distinguish between potential, online, and off-network status, but cannot distinguish between growth, mature, and dormant customers, making it impossible to further conduct differentiated marketing and decision-making. At the same time, the customer base of operators is complex and they are continuous consumers, so the RFM model is not applicable; this is mainly reflected in:
[0011] 1. The customer groups of operators are complex. There are public customers who purchase communication products for the convenience of personal or family life and work, and corporate customers who purchase communication products for production, sales, and maintenance of organizational operations. The former are consumers, who are extensive, scattered, and changeable; the latter are producers, who are small in number, large in amount, and infrequent in frequency;
[0012] 2. Operator customers are not suitable for statistical analysis based on the number of consumption. Operator customers generally order products first, and then recharge and are charged on a periodic basis. Neither ordering nor recharging can reflect the customer's actual consumption frequency. Summary of the invention
[0013] In view of the above problems existing in the prior art, the present invention provides an operator customer life cycle calculation method and system based on the K-means clustering algorithm, which can help operators to more accurately segment customers, reasonably allocate enterprise resources, improve customer commercial value, and extend customer life cycle.
[0014] The present invention discloses a method for calculating the life cycle of an operator's customer, comprising:
[0015] Determine the key indicators for calculating the operator's customer life cycle; wherein the key indicators include: the number of online products P, the average online time T, the arpu value A and the number of offline products D;
[0016] Collect four indicator data of each customer: the number of online products P, average online time T, arpu value A and the number of offline products D;
[0017] The average online time T and the arpu value A are mapped to obtain T i and A i ;
[0018] Set the customer index {P, T i , A i}Use K-means clustering algorithm to divide into three clusters, P is higher than the first threshold, T i Above the second threshold, A i The indicator data of customers whose index data is higher than the third threshold is clustered as K1, and P is lower than the fourth threshold, A i The indicator data of customers below the sixth threshold are clustered as K2, P is below the fourth threshold, T iThe indicator data of customers below the fifth threshold are clustered as K3;
[0019] The customer lifecycle belonging to cluster K1 is defined as the mature stage;
[0020] The customer lifecycle belonging to cluster K2 is defined as the dormant period;
[0021] The customer life cycle belonging to cluster K3 is defined as the growth stage.
[0022] As a further improvement of the present invention,
[0023] T i Take constants T1, T2, T3, T4, ..., T n One of T1<T2<T3<T4、···、<T n ;
[0024] A i Take constants A1, A2, A3, A4, ..., A n One of A1<A2<A3<A4、···、<A n .
[0025] As a further improvement of the present invention, the average online time T and the arpu value A are mapped to obtain T i and A i ;include:
[0026] The T and A indicator data with the largest differences are sorted from high to low according to public customers and corporate customers and divided into 5 equal parts, with scores of 5, 4, 3, 2, and 1 respectively; that is, the top 20% are scored 5 points and the bottom 20% are scored 1 point.
[0027] As a further improvement of the present invention, the first threshold is not lower than the fourth threshold, the second threshold is not lower than the fifth threshold, and the third threshold is not lower than the sixth threshold.
[0028] As a further improvement of the present invention, it also includes:
[0029] Classify the collected customer indicator data to obtain public customer indicator data and government and enterprise customer indicator data;
[0030] The public customer indicator data and the government and enterprise customer indicator data are calculated separately.
[0031] As a further improvement of the present invention, it also includes:
[0032] The customer lifecycle with P=0 and D=0 is defined as the introduction period;
[0033] The customer life cycle with P=0 and D>0 is defined as the churn period.
[0034] The present invention also discloses an operator customer life cycle calculation system, comprising:
[0035] A determination module is used to determine the key indicators for calculating the operator's customer life cycle; wherein the key indicators include: the number of online products P, the average online time T, the arpu value A and the number of offline products D;
[0036] The collection module is used to collect four indicator data of each customer: the number of online products P, the average online time T, the arpu value A and the number of offline products D;
[0037] A processing module is used to perform data mapping on the average online time T and the arpu value A to obtain T i and A i ;
[0038] Clustering module is used to cluster customer indicators {P, T i , A i}Use K-means clustering algorithm to divide into three clusters, P is higher than the first threshold, T i Above the second threshold, A i The indicator data of customers whose index data is higher than the third threshold is clustered as K1, and P is lower than the fourth threshold, A i The indicator data of customers below the sixth threshold are clustered as K2, P is below the fourth threshold, T i The indicator data of customers below the fifth threshold are clustered as K3;
[0039] The definition module is used to define the customer life cycle belonging to cluster K1 as a mature period, the customer life cycle belonging to cluster K2 as a dormant period, and the customer life cycle belonging to cluster K3 as a growing period.
[0040] As a further improvement of the present invention, the processing module is specifically used to: sort the T and A indicator data with the largest differences from high to low according to public customers and corporate customers, and divide them into 5 equal parts, scoring 5, 4, 3, 2, and 1 points respectively; that is, the highest 20% are scored 5 points, and the lowest 20% are scored 1 point.
[0041] As a further improvement of the present invention, it also includes:
[0042] The classification module is used to classify the collected customer indicator data to obtain public customer indicator data and government and enterprise customer indicator data respectively; and to calculate the public customer indicator data and the government and enterprise customer indicator data separately.
[0043] As a further improvement of the present invention, the definition module is further used to: define the customer life cycle of P=0, D=0 as the introduction period, and define the customer life cycle of P=0, D>0 as the churn period.
[0044] Compared with the prior art, the present invention has the following beneficial effects:
[0045] 1. The present invention uses key indicators that are more suitable for the characteristics of operator customers, solving the problem that the current operator customer life cycle division and traditional RFM model indicators are not applicable;
[0046] 2. The present invention calculates public customers and government and enterprise customers separately to solve the problem of data distortion caused by large differences in characteristics between individual and enterprise customers;
[0047] 3. The critical value for distinguishing the customer life cycle stages in the present invention is not fixed, but customers are classified according to the similarity of the data, so that the calculation result is closer to the actual situation; BRIEF DESCRIPTION OF THE DRAWINGS
[0048] Figure 1 A flowchart of a method for calculating a carrier customer lifecycle disclosed in an embodiment of the present invention;
[0049] Figure 2 A framework diagram of a carrier customer lifecycle computing system disclosed in an embodiment of the present invention;
[0050] Figure 3 A schematic diagram of classification results disclosed in an embodiment of the present invention. DETAILED DESCRIPTION
[0051] In order to make the purpose, technical solution and advantages of the embodiments of the present invention clearer, the technical solution in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.
[0052] The present invention is further described in detail below in conjunction with the accompanying drawings:
[0053] The present invention provides a method and system for calculating the operator customer life cycle based on the K-means clustering algorithm. The K-means clustering algorithm is an unsupervised learning algorithm that classifies customers according to the similarity of data. It is very suitable for customer life cycle management and customer classification of operators whose classification standards are not known in advance; it can help operators to more correctly segment customers, reasonably allocate enterprise resources, improve customer commercial value, and extend customer life cycle.
[0054] Specific:
[0055] like Figure 1As shown, the present invention provides a method for calculating the customer life cycle of an operator, comprising:
[0056] S1. Determine the key indicators for calculating the operator's customer life cycle; the key indicators include: the number of online products P, the average online time T, the arpu value A and the number of offline products D;
[0057] The specific determination method is:
[0058] The relevant indicators of the study include: number of online products P, average online time T, arpu value A and number of offline products D, SMS, incoming voice calls, recharge, outgoing voice calls and traffic; as shown in the following table:
[0059] Finally, the number of online products P, average online time T, arpu value A and the number of offline products D were selected as key indicators.
[0060] S2. Collect four indicator data of each customer: the number of online products P, the average online time T, the arpu value A and the number of offline products D.
[0061] S3, perform data mapping on the average online time T and the arpu value A to obtain T i and A i ; Among them, T i Take constants T1, T2, T3, T4, ..., T n One of T1<T2<T3<T4、···、<T n ; A i Take constants A1, A2, A3, A4, ..., A n One of A1<A2<A3<A4、···、<A n .
[0062] Specific:
[0063] Since the data of different indicators vary greatly, it is necessary to transform the data to meet the needs of data mining; therefore, the T and A indicator data with the largest differences are sorted from high to low according to public customers and corporate customers and divided into 5 equal parts, with 5, 4, 3, 2, and 1 points respectively; that is, the highest 20% are scored 5 points, and the lowest 20% are scored 1 point.
[0064] S4. Set the customer index {P, T i , A i}Use K-means clustering algorithm to divide into three clusters, P is higher than the first threshold, T i Above the second threshold, A i The indicator data of customers whose index data is higher than the third threshold is clustered as K1, and P is lower than the fourth threshold, A iThe indicator data of customers below the sixth threshold are clustered as K2, P is below the fourth threshold, T i The indicator data of customers below the fifth threshold are clustered as K3;
[0065] Among them, the first threshold is not lower than the fourth threshold, the second threshold is not lower than the fifth threshold, and the third threshold is not lower than the sixth threshold; the first threshold to the sixth threshold are selected according to actual needs.
[0066] Specific:
[0067] The k-means clustering algorithm is an iterative clustering analysis algorithm. Its steps are: divide the data into K groups in advance, randomly select K objects as the initial cluster centers, and then calculate the distance between each object and each seed cluster center, and assign each object to the cluster center closest to it. The cluster centers and the objects assigned to them represent a cluster. Each time a sample is assigned, the cluster center of the cluster will be recalculated based on the existing objects in the cluster. This process will be repeated until the cluster center no longer changes.
[0068] S5. Define the life cycle, namely:
[0069] The customer lifecycle with P=0 and D=0 is defined as the introduction period;
[0070] The customer life cycle with P=0 and D>0 is defined as the churn period or off-network period;
[0071] The customer lifecycle belonging to cluster K1 is defined as the mature stage;
[0072] The customer lifecycle belonging to cluster K2 is defined as the dormant period;
[0073] The customer lifecycle belonging to cluster K3 is defined as the growth stage;
[0074] The details are shown in the following table:
[0075]
[0076] Based on this, operators can conduct differentiated marketing based on the calculated customer life cycle, improve customer commercial value, analyze signs of churn, and extend customer life cycle.
[0077] Furthermore, in the above calculation method, since the groups of public customers and corporate customers are very different, their life cycles are calculated separately; that is, the collected customer indicator data are classified to obtain public customer indicator data and government and enterprise customer indicator data respectively; the public customer indicator data and government and enterprise customer indicator data are calculated separately.
[0078] like Figure 2 As shown, the present invention provides an operator customer lifecycle calculation system, comprising:
[0079] Determine a module, used to implement the above S1;
[0080] A collection module, used to implement the above S2;
[0081] A processing module, used to implement the above S3;
[0082] Clustering module, used to implement the above S4;
[0083] Define modules that will implement the above S5.
[0084] Furthermore, in the above-mentioned computing system, since the groups of public customers and corporate customers are very different, their life cycles are calculated separately; that is, the collected customer indicator data are classified to obtain public customer indicator data and government and enterprise customer indicator data respectively; the public customer indicator data and government and enterprise customer indicator data are calculated separately.
[0085] Example 1
[0086] Zhang, a public customer of a certain operator, created customer information on October 1, 2015, but did not apply for any products. On November 1, 2015, the system collected indicator data (P=0, D=0). According to the life cycle calculation rules, this customer belongs to the introduction period.
[0087] Example 2
[0088] Randomly select 100 online public customers of a certain operator, collect the number of online products (P), average online time (T), arpu value (A) indicators of these 100 customers and convert them, and then use the K-means clustering algorithm to classify them. The classification results are as follows: Figure 3 shown.
[0089] Depend on Figure 3 It can be seen that customer type 1 has high T, high P and high A, which belongs to the mature stage; customer type 2 has low P and low A, which belongs to the dormant stage; the remaining customer type 3 belongs to the growth stage.
[0090] The present invention has the following advantages in defining and calculating the operator customer lifecycle:
[0091] 1. The present invention uses key indicators that are more suitable for the characteristics of operator customers, solving the problem that the current operator customer life cycle division and traditional RFM model indicators are not applicable;
[0092] 2. The present invention calculates public customers and government and enterprise customers separately to solve the problem of data distortion caused by large differences in characteristics between individual and enterprise customers;
[0093] 3. The critical value for distinguishing the customer life cycle stages in the present invention is not fixed, but customers are classified according to the similarity of the data, so that the calculation result is closer to the actual situation.
[0094] The above are only preferred embodiments of the present invention and are not intended to limit the present invention. For those skilled in the art, the present invention may have various modifications and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.
Claims
1. A method for calculating the customer life cycle of an operator, characterized in that: include: Determine the key indicators for calculating the operator's customer life cycle; wherein the key indicators include: the number of online products P, the average online time T, the arpu value A and the number of offline products D; Collect four indicator data of each customer, namely, the number of online products P, the average online time T, the arpu value A, and the number of off-network products D; classify the collected customer indicator data to obtain public customer indicator data and government and enterprise customer indicator data; and calculate the public customer indicator data and the government and enterprise customer indicator data separately; The average online time T and the arpu value A are mapped to obtain T i and A i ; Among them, T i Take constants T1, T2, T3, T4, ..., T n One of T1<T2<T3<T4、···、<T n ; A i Take constants A1, A2, A3, A4, ..., A n One of A1<A2<A3<A4、···、<A n ; Set the customer index {P, T i , A i }Use K-means clustering algorithm to divide into three clusters, P is higher than the first threshold, T i Above the second threshold, A i The indicator data of customers whose index data is higher than the third threshold is clustered as K1, and P is lower than the fourth threshold, A i The indicator data of customers below the sixth threshold are clustered as K2, P is below the fourth threshold, T i The indicator data of customers whose values are lower than the fifth threshold are clustered as K3; wherein the first threshold is not lower than the fourth threshold, the second threshold is not lower than the fifth threshold, and the third threshold is not lower than the sixth threshold; The customer lifecycle belonging to cluster K1 is defined as the mature stage; The customer lifecycle belonging to cluster K2 is defined as the dormant period; The customer lifecycle belonging to cluster K3 is defined as the growth stage; The customer lifecycle with P=0 and D=0 is defined as the introduction period; The customer life cycle with P=0 and D>0 is defined as the churn period.
2. The calculation method according to claim 1, characterized in that: The average online time T and the arpu value A are mapped to obtain T i and A i ; include: The T and A indicator data with the largest differences are sorted from high to low according to public customers and corporate customers and divided into 5 equal parts, with scores of 5, 4, 3, 2, and 1 respectively; that is, the top 20% are scored 5 points and the bottom 20% are scored 1 point.
3. An operator customer lifecycle calculation system, characterized in that: include: A determination module is used to determine the key indicators for calculating the operator's customer life cycle; wherein the key indicators include: the number of online products P, the average online time T, the arpu value A and the number of offline products D; The collection module is used to collect four indicator data of each customer: the number of online products P, the average online time T, the arpu value A and the number of offline products D; A processing module is used to perform data mapping on the average online time T and the arpu value A to obtain T i and A i ; Among them, T i Take constants T1, T2, T3, T4, ..., T n One of T1<T2<T3<T4、···、<T n ; A i Take constants A1, A2, A3, A4, ..., A n One of A1<A2<A3<A4、···、<A n ; Clustering module is used to cluster customer indicators {P, T i , A i }Use K-means clustering algorithm to divide into three clusters, P is higher than the first threshold, T i Above the second threshold, A i The indicator data of customers whose index data is higher than the third threshold is clustered as K1, and P is lower than the fourth threshold, A i The indicator data of customers below the sixth threshold are clustered as K2, P is below the fourth threshold, T i The indicator data of customers whose values are lower than the fifth threshold are clustered as K3; wherein the first threshold is not lower than the fourth threshold, the second threshold is not lower than the fifth threshold, and the third threshold is not lower than the sixth threshold; A definition module is used to define the customer life cycle belonging to cluster K1 as a mature period, the customer life cycle belonging to cluster K2 as a dormant period, and the customer life cycle belonging to cluster K3 as a growing period; define the customer life cycle of P=0 and D=0 as an introduction period, and define the customer life cycle of P=0 and D>0 as a churn period; The classification module is used to classify the collected customer indicator data to obtain public customer indicator data and government and enterprise customer indicator data respectively; and to calculate the public customer indicator data and the government and enterprise customer indicator data separately.
4. The computing system of claim 3, wherein: The processing module is specifically used to: sort the T and A indicator data with the largest differences from high to low according to public customers and corporate customers, and divide them into 5 equal parts, scoring 5, 4, 3, 2, and 1 points respectively; that is, the highest 20% is scored 5 points, and the lowest 20% is scored 1 point.
