A Customer Classification Evaluation Method Based on RFM Model

Abstract

The present invention discloses a customer classification evaluation method based on RFM model, which relates to the technical field of data analysis and classification; specifically includes the following steps: S1: customer data collection, collecting customer information through big data or past transaction records; S2: primary classification, using RFM model combined with customer information for primary classification to form multiple category sets; S3: secondary classification, combining customer information, extracting features for customers of each category set in step S2, and then performing secondary classification according to the extraction results; S4: customer stability evaluation, regularly updating steps S1-S3, and extracting customers whose classification labels have changed. After the present invention performs primary classification of customer potential value based on the RFM model, it then forms a degree of preference based on the range of goods sold by the enterprise and the multi-dimensional action frequency of the customer, and then performs secondary classification based on the degree of preference, so as to be more targeted.

Description

Technical Field

[0001] The present invention relates to the technical field of data analysis and classification, and in particular to a customer classification and evaluation method based on an RFM model. Background Art

[0002] In a highly competitive market environment, companies need to accurately understand customer information and manage customer value in order to improve their competitiveness, so as to formulate effective marketing strategies. As a classic customer classification tool, the RFM model evaluates customer value through three key indicators: the last purchase time (recency), purchase frequency (frequency) and purchase amount (monetary).

[0003] After searching, the Chinese patent application number CN116797253B discloses a classification management method based on customer resources, including: obtaining the financial information of each customer and the corresponding customer component information, obtaining the customer category and the corresponding customer density according to all the component information of each customer, obtaining the deviation degree of each component information of each customer category, obtaining the deviation degree according to the deviation degree of all the component information of each customer category and the corresponding customer density, and then screening out the customer category to be detected. Obtain the retention degree of each customer category to be detected, and then screen out the retained customer category and assign a special category label.

[0004] The above patent has the following deficiencies: it is unable to classify customers in relation to the business commodities of the enterprise, which means that even after the customers are classified by RFM, the targeted recommendation marketing for potential customers, the professional explanation for potential customers, the after-sales management for successful customers, etc. lack the specificity relative to the commodities. Summary of the invention

[0005] The purpose of the present invention is to solve the shortcomings of the prior art and to propose a customer classification evaluation method based on the RFM model.

[0006] In order to achieve the above object, the present invention adopts the following technical solutions:

[0007] A customer classification evaluation method based on an RFM model includes the following steps:

[0008] S1: Customer data collection, collecting customer information through big data or past transaction records;

[0009] S2: One-time classification, using the RFM model combined with customer information to perform one-time classification to form multiple category sets;

[0010] S3: Secondary classification: Combine customer information to extract features of customers in each category set in step S2, and then perform secondary classification based on the extraction results;

[0011] S4: Customer stability assessment: regularly update steps S1-S3, extract customers whose classification labels have changed, and then analyze and assess customer stability based on the frequency of changes.

[0012] As a preferred embodiment of the present invention: the step S2 comprises the following steps:

[0013] S21: According to the R, F, and M dimensions, one or more segmentation thresholds are set. The thresholds divide the entire data range into intervals. Each interval corresponds to a type of customer. The intervals are ;

[0014] S22: Obtain the R, F, and M values ​​corresponding to the customer, and classify the customer into corresponding intervals according to the inclusion relationship of the intervals.

[0015] As a preferred embodiment of the present invention: in the step S21, R is the length of time from the most recent consumption to the current time, F is the consumption frequency in a statistical period, and M is the total amount of consumption in a statistical period. A statistical period can be one month, one quarter, or one year.

[0016] As a preferred embodiment of the present invention: in step S22, one customer may be included in multiple intervals.

[0017] As a preferred embodiment of the present invention: in the step S2, the customer category set includes:

[0018] High-value customers;

[0019] Focus on customer retention;

[0020] Focus on developing customers;

[0021] Focus on retaining customers;

[0022] General value customers;

[0023] generally retain customers;

[0024] General development clients;

[0025] Potential customers.

[0026] As a preferred embodiment of the present invention: in the step S3, the secondary classification method comprises the following steps:

[0027] S31: Obtain the customer's browsing dimension L, search dimension S, purchase dimension G, and evaluation behavior dimension P for each product through the collected customer information;

[0028] S32: Taking the current time as the node and Δt as the period, determine n time periods forward ;

[0029] S33: Determine the operation frequency of each dimension in each time period , for Frequency within a time period;

[0030] S34: Use logic based on similar association to obtain the total frequency F of each dimension, and perform weighted average calculation on the total frequency F of each dimension to obtain the customer's preference for the product. ;

[0031] S35: Classify the products according to the categories of the products they sell , which represents the i-th category, each category contains multiple items , which represents the oth product under the i-th category product;

[0032] S36: Calculate the total customer preference for each category of goods. ,in On behalf of customers The total popularity of the category products, Representing clients The degree of preference for the oth item in the category;

[0033] S37: Calculate the total preference of customers for each category of products, and classify the customers into the category with the highest total preference.

[0034] As a preferred embodiment of the present invention: in the step S34, , for The weight of .

[0035] As a preferred embodiment of the present invention: in the step S4, the customer stability evaluation method comprises the following steps:

[0036] S41: Setting the update period T;

[0037] S42: After reaching the update period T, repeat steps S1-S3 to re-divide customers;

[0038] S43: Conduct targeted analysis on the changes in the first-level categories of customers in S2 and the changes in the second-level categories of customers in S3 to obtain information on customer stability.

[0039] As a preferred embodiment of the present invention: in the step S43, the analysis of the change of customers divided into the first-level categories in S2 adopts the formula Calculation, where W is the stability of customer consumption habits, is the number of updates, The number of times the customer category changes before and after the update.

[0040] As a preferred embodiment of the present invention: in the step S43, the customer change analysis of the secondary category classification in S3 adopts the formula ,in To ensure the stability of customer product preferences, is the number of updates, It is the number of times a customer changes within the secondary category and does not change within the primary category before and after the update.

[0041] The beneficial effects of the present invention are:

[0042] 1. The present invention, after primary classification of customer potential value based on the RFM model, then forms a preference level based on the range of products sold by the enterprise and the multi-dimensional action frequency of the customer, and then performs secondary classification based on the preference level, thereby being more targeted.

[0043] 2. The present invention, on the basis of secondary classification, adopts logic based on similar association to perform weight control, so that the weight is distributed in an arithmetic progression, so that when calculating the characteristics of the dimensional operation frequency, it can better fit the behavior close to the current time, so that the calculation is more accurate and reasonable.

[0044] 3. The present invention can analyze customer stability by comparing the number of updates with the number of changes, so as to better carry out targeted maintenance on customers and increase subsequent transaction rates. At the same time, customer stability is evaluated in two aspects: the stability of consumption ability and habits and the stability of product preferences, so that the evaluation of customer stability can be more detailed and the subsequent process can be more targeted. BRIEF DESCRIPTION OF THE DRAWINGS

[0045] Figure 1 A flow chart of a customer classification and evaluation method based on the RFM model proposed by the present invention;

[0046] Figure 2 This is a two-level architecture diagram of customer classification of a customer classification evaluation method based on the RFM model proposed in the present invention. DETAILED DESCRIPTION

[0047] The technical solution of this patent is further described in detail below in conjunction with specific implementation methods.

[0048] The embodiments of the present invention are described in detail below, and examples of the embodiments are shown in the accompanying drawings, wherein the same or similar reference numerals throughout represent the same or similar elements or elements having the same or similar functions. The embodiments described below with reference to the accompanying drawings are exemplary and are only used to explain the present invention, and cannot be understood as limiting the present invention.

[0049] Embodiment 1: A customer classification evaluation method based on the RFM model, comprising the following steps:

[0050] S1: Customer data collection, collecting customer information through big data or past transaction records;

[0051] S2: One-time classification, using the RFM model combined with customer information to perform one-time classification to form multiple category sets;

[0052] S3: Secondary classification: Combine customer information to extract features of customers in each category set in step S2, and then perform secondary classification based on the extraction results;

[0053] S4: Customer stability assessment: regularly update steps S1-S3, extract customers whose classification labels have changed, and then analyze and assess customer stability based on the frequency of changes.

[0054] The step S2 comprises the following steps:

[0055] S21: According to the R, F, and M dimensions, one or more segmentation thresholds are set. The thresholds divide the entire data range into intervals. Each interval corresponds to a type of customer. The intervals are ;

[0056] S22: Obtain the R, F, and M values ​​corresponding to the customer, and classify the customer into corresponding intervals according to the inclusion relationship of the intervals.

[0057] In step S21, R is the length of time from the most recent consumption to the current time, F is the consumption frequency within a statistical period, and M is the total amount of consumption within a statistical period. A statistical period can be one month, one quarter, or one year.

[0058] In step S22, one customer may be included in multiple intervals.

[0059] In the step S2, the customer category set includes:

[0060] High-value customers: recent consumption time is short, consumption frequency is high, and consumption amount is high;

[0061] Focus on retaining customers: those who have spent a long time recently, have a high frequency of consumption, and have a high consumption amount;

[0062] Focus on developing customers: those with a short consumption time, high consumption amount, and low consumption frequency;

[0063] Focus on retaining customers: customers who have spent a long time recently, have a low frequency of consumption, and have a high amount of consumption;

[0064] General value customers: recent consumption time is long, consumption frequency is low, and consumption amount is low;

[0065] Generally retained customers: recent consumption time is short, consumption frequency is low, and consumption amount is low;

[0066] General development customers: recent consumption time is short, consumption frequency is high, and consumption amount is low;

[0067] Potential customers: They have recently consumed for a long time, with low consumption frequency and low consumption amount.

[0068] Embodiment 2: A customer classification evaluation method based on the RFM model, comprising the following steps:

[0069] S1: Customer data collection, collecting customer information through big data or past transaction records;

[0070] S2: One-time classification, using the RFM model combined with customer information to perform one-time classification to form multiple category sets;

[0071] S3: Secondary classification: Combine customer information to extract features of customers in each category set in step S2, and then perform secondary classification based on the extraction results;

[0072] S4: Customer stability assessment: regularly update steps S1-S3, extract customers whose classification labels have changed, and then analyze and assess customer stability based on the frequency of changes.

[0073] The step S2 comprises the following steps:

[0074] S21: According to the R, F, and M dimensions, one or more segmentation thresholds are set. The thresholds divide the entire data range into intervals. Each interval corresponds to a type of customer. The intervals are ;

[0075] S22: Obtain the R, F, and M values ​​corresponding to the customer, and classify the customer into corresponding intervals according to the inclusion relationship of the intervals.

[0076] In step S21, R is the length of time from the most recent consumption to the current time, F is the consumption frequency within a statistical period, and M is the total amount of consumption within a statistical period. A statistical period can be one month, one quarter, or one year.

[0077] In step S22, one customer may be included in multiple intervals.

[0078] In the step S2, the customer category set includes:

[0079] High-value customers: recent consumption time is short, consumption frequency is high, and consumption amount is high;

[0080] Focus on retaining customers: those who have spent a long time recently, have a high frequency of consumption, and have a high consumption amount;

[0081] Focus on developing customers: those with a short consumption time, high consumption amount, and low consumption frequency;

[0082] Focus on retaining customers: customers who have spent a long time recently, have a low frequency of consumption, and have a high amount of consumption;

[0083] General value customers: recent consumption time is long, consumption frequency is low, and consumption amount is low;

[0084] Generally retained customers: recent consumption time is short, consumption frequency is low, and consumption amount is low;

[0085] General development customers: recent consumption time is short, consumption frequency is high, and consumption amount is low;

[0086] Potential customers: They have recently consumed for a long time, with low consumption frequency and low consumption amount.

[0087] In the step S3, the secondary classification method includes the following steps:

[0088] S31: Obtain the customer's browsing dimension L, search dimension S, purchase dimension G, and evaluation behavior dimension P for each product through the collected customer information;

[0089] S32: Taking the current time as the node and Δt as the period, determine n time periods forward ;

[0090] S33: Determine the operation frequency of each dimension in each time period , for Frequency within a time period;

[0091] S34: Use logic based on similar association to obtain the total frequency F of each dimension, and perform weighted average calculation on the total frequency F of each dimension to obtain the customer's preference for the product. ;

[0092] S35: Classify the products according to the categories of the products they sell , which represents the i-th category, each category contains multiple items , which represents the oth product under the i-th category product;

[0093] S36: Calculate the total customer preference for each category of goods. ,in On behalf of customers The total popularity of the category products, Representing clients The degree of preference for the oth item in the category;

[0094] S37: Calculate the total preference of customers for each category of products, and classify the customers into the category with the highest total preference.

[0095] The present invention, after performing primary classification on potential value based on the RFM model, then forms a preference level based on the range of products sold by the enterprise and the multi-dimensional action frequency of the customers, and then performs secondary classification based on the preference level, thereby being more targeted.

[0096] Embodiment 3: A customer classification evaluation method based on the RFM model, comprising the following steps:

[0097] S1: Customer data collection, collecting customer information through big data or past transaction records;

[0098] S2: One-time classification, using the RFM model combined with customer information to perform one-time classification to form multiple category sets;

[0099] S3: Secondary classification: Combine customer information to extract features of customers in each category set in step S2, and then perform secondary classification based on the extraction results;

[0100] S4: Customer stability assessment: regularly update steps S1-S3, extract customers whose classification labels have changed, and then analyze and assess customer stability based on the frequency of changes.

[0101] The step S2 comprises the following steps:

[0102] S21: According to the R, F, and M dimensions, one or more segmentation thresholds are set. The thresholds divide the entire data range into intervals. Each interval corresponds to a type of customer. The intervals are ;

[0103] S22: Obtain the R, F, and M values ​​corresponding to the customer, and classify the customer into corresponding intervals according to the inclusion relationship of the intervals.

[0104] In step S21, R is the length of time from the most recent consumption to the current time, F is the consumption frequency within a statistical period, and M is the total amount of consumption within a statistical period. A statistical period can be one month, one quarter, or one year.

[0105] In step S22, one customer may be included in multiple intervals.

[0106] In the step S2, the customer category set includes:

[0107] High-value customers: recent consumption time is short, consumption frequency is high, and consumption amount is high;

[0108] Focus on retaining customers: those who have spent a long time recently, have a high frequency of consumption, and have a high consumption amount;

[0109] Focus on developing customers: those with a short consumption time, high consumption amount, and low consumption frequency;

[0110] Focus on retaining customers: customers who have spent a long time recently, have a low frequency of consumption, and have a high amount of consumption;

[0111] General value customers: recent consumption time is long, consumption frequency is low, and consumption amount is low;

[0112] Generally retained customers: recent consumption time is short, consumption frequency is low, and consumption amount is low;

[0113] General development customers: recent consumption time is short, consumption frequency is high, and consumption amount is low;

[0114] Potential customers: They have recently consumed for a long time, with low consumption frequency and low consumption amount.

[0115] In the step S3, the secondary classification method includes the following steps:

[0116] S31: Obtain the customer's browsing dimension L, search dimension S, purchase dimension G, and evaluation behavior dimension P for each product through the collected customer information;

[0117] S32: Taking the current time as the node and Δt as the period, determine n time periods forward ;

[0118] S33: Determine the operating frequency of each dimension within each segment , for Frequency within a time period;

[0119] S34: Use logic based on similar association to obtain the total frequency F of each dimension, and perform weighted average calculation on the total frequency F of each dimension to obtain the customer's preference for the product. ;

[0120] S35: Classify the products according to the categories of the products they sell , which represents the i-th category, each category contains multiple items , which represents the oth product under the i-th category product;

[0121] S36: Calculate the total customer preference for each category of goods. ,in On behalf of customers The total popularity of the category products, Representing clients The degree of preference for the oth item in the category;

[0122] S37: Calculate the total preference of customers for each category of products, and classify the customers into the category with the highest total preference.

[0123] In the step S34, , for The weight of .

[0124] The present invention, on the basis of secondary classification, adopts logic based on similar association to control weights, so that the weights are distributed in an arithmetic progression, so that when calculating the characteristics of dimensional operation frequency, it can better fit the behavior close to the current time, so that the calculation is more accurate and reasonable.

[0125] Embodiment 4: A customer classification evaluation method based on the RFM model, comprising the following steps:

[0126] S1: Customer data collection, collecting customer information through big data or past transaction records;

[0127] S2: One-time classification, using the RFM model combined with customer information to perform one-time classification to form multiple category sets;

[0128] S3: Secondary classification: Combine customer information to extract features of customers in each category set in step S2, and then perform secondary classification based on the extraction results;

[0129] S4: Customer stability assessment: regularly update steps S1-S3, extract customers whose classification labels have changed, and then analyze and assess customer stability based on the frequency of changes.

[0130] The step S2 comprises the following steps:

[0131] S21: According to the R, F, and M dimensions, one or more segmentation thresholds are set. The thresholds divide the entire data range into intervals. Each interval corresponds to a type of customer. The intervals are ;

[0132] S22: Obtain the R, F, and M values ​​corresponding to the customer, and classify the customer into corresponding intervals according to the inclusion relationship of the intervals.

[0133] In step S21, R is the length of time from the most recent consumption to the current time, F is the consumption frequency within a statistical period, and M is the total amount of consumption within a statistical period. A statistical period can be one month, one quarter, or one year.

[0134] In step S22, one customer may be included in multiple intervals.

[0135] In the step S2, the customer category set includes:

[0136] High-value customers: recent consumption time is short, consumption frequency is high, and consumption amount is high;

[0137] Focus on retaining customers: those who have spent a long time recently, have a high frequency of consumption, and have a high consumption amount;

[0138] Focus on developing customers: those with a short consumption time, high consumption amount, and low consumption frequency;

[0139] Focus on retaining customers: customers who have spent a long time recently, have a low frequency of consumption, and have a high amount of consumption;

[0140] General value customers: recent consumption time is long, consumption frequency is low, and consumption amount is low;

[0141] Generally retained customers: recent consumption time is short, consumption frequency is low, and consumption amount is low;

[0142] General development customers: recent consumption time is short, consumption frequency is high, and consumption amount is low;

[0143] Potential customers: They have recently consumed for a long time, with low consumption frequency and low consumption amount.

[0144] In the step S3, the secondary classification method includes the following steps:

[0145] S31: Obtain the customer's browsing dimension L, search dimension S, purchase dimension G, and evaluation behavior dimension P for each product through the collected customer information;

[0146] S32: Taking the current time as the node and Δt as the period, determine n time periods forward ;

[0147] S33: Determine the operation frequency of each dimension in each time period , for Frequency within a time period;

[0148] S34: Use logic based on similar association to obtain the total frequency F of each dimension, and perform weighted average calculation on the total frequency F of each dimension to obtain the customer's preference for the product. ;

[0149] S35: Classify the products according to the categories of the products they sell , which represents the i-th category, each category contains multiple items , which represents the oth product under the i-th category product;

[0150] S36: Calculate the total customer preference for each category of goods. ,in On behalf of customers The total popularity of the category products, Representing clients The degree of preference for the oth item in the category;

[0151] S37: Calculate the total preference of customers for each category of products, and classify the customers into the category with the highest total preference.

[0152] In the step S34, , for The weight of .

[0153] In step S4, the customer stability assessment method includes the following steps:

[0154] S41: Setting the update period T;

[0155] S42: After reaching the update period T, repeat steps S1-S3 to re-divide customers;

[0156] S43: Conduct targeted analysis on the changes in the first-level categories of customers in S2 and the changes in the second-level categories of customers in S3 to obtain information on customer stability.

[0157] In the step S43, the formula is used to analyze the change of customers divided into the first-level categories in S2. Calculation, where W is the stability of customer consumption habits, is the number of updates, The number of times the customer category changes before and after the update.

[0158] In the step S43, the customer change analysis of the secondary category classification in S3 adopts the formula ,in To ensure the stability of customer product preferences, is the number of updates, It is the number of times a customer changes within the secondary category and does not change within the primary category before and after the update.

[0159] The present invention can analyze customer stability by comparing the number of updates with the number of changes, so as to better carry out targeted maintenance on customers and increase the subsequent transaction rate. At the same time, customer stability is evaluated in two aspects: the stability of consumption ability and habits and the stability of product preferences, so that the evaluation of customer stability can be more detailed and the targeting of the subsequent process can be increased.

[0160] The above description is only a preferred specific implementation manner of the present invention, but the protection scope of the present invention is not limited thereto. Any technician familiar with the technical field can make equivalent replacements or changes according to the technical scheme and inventive concept of the present invention within the technical scope disclosed by the present invention, which should be covered by the protection scope of the present invention.

Claims

1. A customer classification evaluation method based on the RFM model, characterized in that: The following steps are involved: S1: Customer data collection, collecting customer information through big data or past transaction records; S2: One-time classification, using the RFM model combined with customer information to perform one-time classification to form multiple category sets; S3: Secondary classification: Combine customer information to extract features of customers in each category set in step S2, and then perform secondary classification based on the extraction results; S4: Customer stability assessment: regularly update steps S1-S3, extract customers whose classification labels have changed, and analyze and assess customer stability based on the change frequency; In the step S3, the secondary classification method includes the following steps: S31: Obtain the customer's browsing dimension L, search dimension S, purchase dimension G, and evaluation behavior dimension P for each product through the collected customer information; S32: Taking the current time as the node and Δt as the period, determine n time periods forward ; S33: Determine the operation frequency of each dimension in each time period , for Frequency within a time period; S34: Use logic based on similar association to obtain the total frequency F of each dimension, and perform weighted average calculation on the total frequency F of each dimension to obtain the customer's preference for the product. ; S35: Classify the products according to the categories of the products they sell , which represents the i-th category, each category contains multiple items , which represents the oth product under the i-th category product; S36: Calculate the total customer preference for each category of goods. ,in On behalf of customers The total popularity of the category products, Representing clients The degree of preference for the oth item in the category; S37: Calculate the total preference of customers for each category of products, and classify the customers into the category with the highest total preference.

2. According to the customer classification evaluation method based on the RFM model of claim 1, it is characterized in that: The step S2 comprises the following steps: S21: According to the R, F, and M dimensions, one or more segmentation thresholds are set. The thresholds divide the entire data range into intervals. Each interval corresponds to a type of customer. The intervals are ; S22: Obtain the R, F, and M values ​​corresponding to the customer, and classify the customer into the corresponding interval according to the inclusion relationship between the customer and the interval; In step S21, R is the length of time from the most recent consumption to the current time, F is the consumption frequency within a statistical period, and M is the total amount of consumption within a statistical period. A statistical period can be one month, one quarter, or one year.

3. A customer classification evaluation method based on RFM model according to claim 2, characterized in that: In step S22, one customer may be included in multiple intervals.

4. The customer classification evaluation method based on the RFM model according to claim 1 is characterized in that: In the step S2, the customer category set includes: High-value customers; Focus on customer retention; Focus on developing customers; Focus on retaining customers; General value customers; generally retain customers; General development clients; Potential customers.

5. The customer classification evaluation method based on the RFM model according to claim 1 is characterized in that: In the step S34, , for The weight of .

6. The customer classification evaluation method based on the RFM model according to claim 1 is characterized in that: In step S4, the customer stability assessment method includes the following steps: S41: Setting the update period T; S42: After reaching the update period T, repeat steps S1-S3 to re-divide customers; S43: Conduct targeted analysis on the changes in the first-level categories of customers in S2 and the changes in the second-level categories of customers in S3 to obtain information on customer stability.

7. The customer classification evaluation method based on the RFM model according to claim 6 is characterized in that: In the step S43, the formula is used to analyze the change of customers divided into the first-level categories in S2. Calculation, where W is the stability of customer consumption habits, is the number of updates, The number of times the customer category changes before and after the update.

8. The customer classification evaluation method based on the RFM model according to claim 6 is characterized in that: In the step S43, the customer change analysis of the secondary category classification in S3 adopts the formula ,in To ensure the stability of customer product preferences, is the number of updates, It is the number of times a customer changes within the secondary category and does not change within the primary category before and after the update.

Images