A Customer Relationship Management Method and System Based on Big Data Analytics

Abstract

This invention relates to the field of business system technology, and more particularly to a customer relationship management method and system based on big data analysis. A customer relationship management method based on big data analysis includes the following steps: S1: Acquiring public customer data and customer data from sales personnel's private pools, and preprocessing them; S2: Based on customers' historical periodic purchase data, a public customer recommendation pool is established, comprising a first recommendation pool matching unassigned public customers with public customers matching historical short-term periodic purchase data, and a second recommendation pool matching unassigned public customers with public customers matching historical long-term periodic purchase data and meeting the early contact time requirement. This invention obtains the purchase cycle of public customers, makes recommendations based on the public customer cycle, sets a waiting time for customers not yet replaced by sales personnel, and adjusts the recommendations according to the interest weight of each replaced customer.

Description

Technical Field

[0001] This invention relates to the field of business system technology, and in particular to a customer relationship management method and system based on big data analysis. Background Technology

[0002] Customer Relationship Management (CRM) is a strategy that uses systematic management methods to help businesses establish, maintain, and optimize relationships with their customers. With increasingly fierce market competition, businesses are placing greater emphasis on the management and maintenance of customer resources. Traditional CRM systems primarily store and manage customer information for the convenience of sales personnel. However, with the development of new technologies such as big data, artificial intelligence, and cloud computing, the functions of modern CRM systems are no longer limited to basic data management but are gradually expanding into more intelligent and personalized areas. In reality, when sales personnel use CRM systems, some may constantly replace customers in their private databases to secure better customer resources, and may engage in private exchanges between sales personnel. For example, salesperson A might immediately inform salesperson B when releasing customer resources. This phenomenon prevents customer resources from being allocated appropriately. Summary of the Invention

[0003] To overcome the shortcomings of sales personnel not being able to allocate customer resources reasonably, this invention provides a customer relationship management method and system based on big data analysis.

[0004] The technical solution of this invention is: a customer relationship management method based on big data analysis, comprising the following steps:

[0005] S1: Obtain customer data from the public market and customer data from the sales staff's private market, and preprocess them;

[0006] S2: Based on customers' historical periodic purchase data, the public sea customer recommendation public sea pool includes a first recommendation public sea pool that matches unallocated public sea customers with public sea customers that meet the historical short-term periodic purchase data, and a second recommendation public sea pool that matches unallocated public sea customers with public sea customers that meet the historical long-term periodic purchase data and meet the early contact time.

[0007] S3: Based on the correlation between customers in the target salesperson's private pool and customers in the public pool, the public pool recommends a third recommended public pool that matches the correlation between assigned private pool customers and unassigned public pool customers, and a fourth recommended public pool that replaces and reclaims customers in the salesperson's private pool.

[0008] S4: Based on the public sea customer recommendation pool, the public sea pool recommends customers to target sales personnel.

[0009] Preferably, the process of acquiring public customer data and sales personnel's private customer data, and preprocessing them, includes: performing data cleaning processing on the customer data to remove duplicate values, fill in missing values, and remove outliers.

[0010] Preferably, before the public sea customer recommendation public sea pool based on the customer's historical periodic purchase data includes a first recommended public sea pool matching unassigned public sea customers with public sea customers that meet the historical short-term periodic purchase data, and a second recommended public sea pool matching unassigned public sea customers with public sea customers that meet the historical long-term periodic purchase data and the early contact time, the process includes: obtaining the historical periodic purchase data of public sea customers; using fuzzy inference to obtain periodic threshold judgment rules based on the historical periodic purchase data of public sea customers; and classifying the purchase periodicity of public sea customers according to the periodic threshold judgment rules to obtain short-term periodicity or long-term periodicity.

[0011] Preferably, the public customer recommendation pool based on customers' historical periodic purchase data includes a first recommendation pool of unassigned public customers matched with public customers who meet the historical short-term periodic purchase data, and a second recommendation pool of unassigned public customers matched with public customers who meet the historical long-term periodic purchase data and the early contact time requirement. This includes: obtaining the conversion time of each public user based on a Bayesian algorithm, where the early contact time is the conversion time of the public user; and obtaining public users whose conversion times intersect with those of each public user within a first preset time period of the current time as the second recommendation pool.

[0012] Preferably, obtaining the conversion time of each high seas user based on the Bayesian algorithm includes:

[0013] S201: Organize the historical records of each customer to form a sample set of conversion times;

[0014] S202: Select a probability distribution model, using the gamma distribution, where the probability density function of the gamma distribution is:

[0015] Where x is the conversion time, α is the shape parameter, β is the scale parameter, and Γ(α) is the gamma function;

[0016] S203: Determine the prior distribution, and determine the initial probability distribution parameters as the prior distribution based on historical data;

[0017] S204: Update the rules using Bayesian methods; the posterior distribution is:

[0018] α n =α0+n;

[0019]

[0020] Where n is the number of observed samples, It is the sum of all observed transformation times; α0 is the initial value of the shape parameter; β0 is the initial value of the scale parameter;

[0021] S205: Use the updated posterior distribution for prediction. The predicted purchase cycle is obtained by calculating the expected value of the posterior distribution. The expected value of the gamma distribution is:

[0022]

[0023] Where E[x] is the predicted value of the customer purchase cycle, α n and β n These are the updated shape and scale parameters.

[0024] Preferably, the public customer recommendation pool based on the correlation between customers in the target salesperson's private pool and public pool customers includes a third recommendation pool for matching the correlation between assigned private pool customers and unassigned public pool customers, and a fourth recommendation pool for replacing and recycling customers in the salesperson's private pool. This includes: obtaining the target salesperson's private pool, analyzing the target salesperson's customers based on a correlation analysis model trained on big data, obtaining the correlation value between private pool customers and public pool customers, and placing public pool customers with correlation values ​​greater than a preset correlation threshold into the third recommendation pool.

[0025] Preferably, based on the correlation between customers in the target salesperson's private pool and customers in the public pool, the public pool customer recommendation pool includes a third recommendation pool matching the correlation between assigned private pool customers and unassigned public pool customers, and a fourth recommendation pool for replacement and recycling within the salesperson's private pool. This includes: obtaining the waiting time for a private pool customer to become a public pool customer when the salesperson replaces a private pool customer according to a replacement time calculation formula, and sorting the pools according to the waiting time to obtain the fourth recommendation pool, wherein the replacement time calculation formula is:

[0026]

[0027] in It is the replacement duration. ω is the preset standard replacement duration, T0 is the customer's waiting time in the private sea, and T is the adjustment factor. 标 It is the preset standard waiting time, T max This is the maximum waiting time for customers in the private sea.

[0028] Preferably, the step of sorting according to the waiting time to obtain the fourth recommended public pool includes: taking private customers with a waiting time less than or equal to a preset waiting time as a first preset addition sequence, taking private customers with a waiting time greater than the preset waiting time as a pending sequence, obtaining a first pre-processed private customer based on the pending sequence, wherein the first pre-processed private customer is a private customer who purchased the same type of sales item within a preset time period, obtaining interest-specific parameters for each private customer based on the proportion of sales item types purchased by the first pre-processed private customer within the preset time period, obtaining an adjustment weight based on the interest-specific parameters and the purchase frequency of each private customer, and placing private customers with a weight greater than a weight threshold into the first preset addition sequence to obtain the fourth recommended public pool.

[0029] Preferably, the step of recommending customers to target sales personnel based on the public customer recommendation pool includes: obtaining the intersection of the first recommendation pool, the second recommendation pool, and the third recommendation pool in the public customer recommendation pool as the fifth recommendation pool; removing customers who have been replaced by the target sales personnel from the fourth recommendation pool; and using the remaining customers as the sixth recommendation pool; and recommending customers to target sales personnel based on the fifth recommendation pool and the sixth recommendation pool.

[0030] Preferably, a customer relationship management system based on big data analytics includes:

[0031] The data preprocessing module is used to clean data from both public and private sea clients.

[0032] The cycle determination module is used to obtain the short-term or long-term cycle of customer purchases.

[0033] The conversion time acquisition module is used to acquire customer conversion times with long-term cycles;

[0034] The first recommendation acquisition module is used to acquire public customers that match historical short-term cyclical purchase data;

[0035] The second recommendation acquisition module is used to acquire customers among long-term, recurring customers who have the ability to be contacted by sales personnel in advance.

[0036] The third recommendation acquisition module is used to identify customers whose association with the target salesperson's private and public customer pools exceeds a preset association threshold.

[0037] The fourth recommendation module is used to obtain the customers that each salesperson can replace based on the replacement duration and adjustment parameters.

[0038] The recommendation module makes recommendations to users based on all other recommendation modules.

[0039] This invention has the following advantages: By analyzing the purchase cycle of customers in the public pool, this invention recommends customers with short-term purchase cycles and analyzes customers with long-term purchase cycles. Based on the customer conversion time, it recommends sales personnel. This method allows sales personnel to contact customers in advance, increasing the probability of customer purchases. Furthermore, by setting a replacement time for customers replaced in the sales personnel's private pool, it adjusts the time when replaced customers re-enter the public pool. This method reduces the occurrence of collusion and exchange among sales personnel, resulting in a more rational allocation of customer resources. Attached Figure Description

[0040] Figure 1 This is a flowchart of a customer relationship management method based on big data analysis according to the present invention;

[0041] Figure 2 This is a flowchart illustrating the use of the Bayesian algorithm of this invention. Detailed Implementation

[0042] The preferred technical solution of the present invention will be described in detail below with reference to the accompanying drawings.

[0043] Example 1: A customer relationship management method based on big data analytics, such as Figure 1-2 As shown, it includes the following steps:

[0044] S1: Obtain customer data from the public market and customer data from the sales staff's private market, and preprocess them;

[0045] Data cleaning processes are performed on customer data, including removing duplicate values, filling in missing values, and removing outliers.

[0046] S2: Based on customers' historical periodic purchase data, the public sea customer recommendation public sea pool includes a first recommendation public sea pool that matches unallocated public sea customers with public sea customers that meet the historical short-term periodic purchase data, and a second recommendation public sea pool that matches unallocated public sea customers with public sea customers that meet the historical long-term periodic purchase data and meet the early contact time.

[0047] Obtain historical periodic purchase data of public sea customers, use fuzzy reasoning to obtain periodic threshold judgment rules based on the historical periodic purchase data of public sea customers, and divide the purchase periodicity of public sea customers according to the periodic threshold judgment rules to obtain short-term periodicity or long-term periodicity.

[0048] The conversion time of each public sea user is obtained based on the Bayesian algorithm, wherein the advance contact time is the conversion time of the public sea user. The public sea users whose conversion time intersects with that of each public sea user within the first preset time period of the current time period are selected as the second recommended public sea pool.

[0049] S201: Organize the historical records of each customer to form a sample set of conversion times;

[0050] S202: Select a probability distribution model, using the gamma distribution, where the probability density function of the gamma distribution is:

[0051]

[0052] Where x is the conversion time, α is the shape parameter, β is the scale parameter, and Γ(α) is the gamma function;

[0053] S203: Determine the prior distribution, and determine the initial probability distribution parameters as the prior distribution based on historical data;

[0054] S204: Update the rules using Bayesian methods; the posterior distribution is:

[0055] α n =α0+n;

[0056]

[0057] Where n is the number of observed samples, It is the sum of all observed transformation times; α0 is the initial value of the shape parameter; β0 is the initial value of the scale parameter;

[0058] S205: Use the updated posterior distribution for prediction. The predicted purchase cycle is obtained by calculating the expected value of the posterior distribution. The expected value of the gamma distribution is:

[0059]

[0060] Where E[x] is the predicted value of the customer purchase cycle, α n and β n These are the updated shape and scale parameters.

[0061] It's important to note that by using Bayesian methods, businesses can leverage historical data to create predictive models of customer conversion times and update these predictions in real time based on new data, thus more accurately estimating customer purchase cycles. This approach combines the flexibility of the gamma distribution with the adaptability of Bayesian updates, making it well-suited for handling time-based forecasting problems in customer relationship management.

[0062] By using fuzzy reasoning based on historical data, we obtain the periodic threshold judgment rules for each offshore customer, and then classify them into short-term or long-term periodicity based on the periodic threshold judgment rules.

[0063] Users with short-term purchasing cycles are directly designated as the first recommendation pool. For users with long-term purchasing cycles, the time data from when a salesperson contacts the user to when the user makes a purchase is analyzed. Bayesian methods are used for prediction, and salespersons are recommended based on each user's purchasing cycle and conversion time. For example, if a customer only makes purchases in June each year, and the current time is March, the first preset time period is half a month, and the customer's conversion time is two months, then the salesperson should contact the customer in mid-April. This condition is not met in the current month, so the customer is not included in the second recommendation pool. Users who meet the conditions are included in the second recommendation pool to facilitate early contact by salespeople.

[0064] S3: Based on the correlation between customers in the target salesperson's private pool and customers in the public pool, the public pool recommends a third recommended public pool that matches the correlation between assigned private pool customers and unassigned public pool customers, and a fourth recommended public pool that replaces and reclaims customers in the salesperson's private pool.

[0065] The private pool of the target sales personnel is obtained, and the customers of the target sales personnel are analyzed based on the correlation analysis model trained by big data. The correlation value between the private pool customers and the public pool customers is obtained, and the public pool customers whose correlation value is greater than the preset correlation threshold are put into the third recommended public pool.

[0066] Based on the replacement duration calculation formula, the waiting time for a private client to become a public client when a salesperson replaces a private client is obtained. The public client pool is then sorted according to this waiting time to obtain the fourth recommended pool. The replacement duration calculation formula is as follows:

[0067]

[0068] in It is the replacement duration. ω is the preset standard replacement duration, T0 is the customer's waiting time in the private sea, and T is the adjustment factor. 标 It is the preset standard waiting time, T max This is the maximum waiting time for customers in the private sea.

[0069] Private sea customers with a waiting time less than or equal to a preset time are designated as the first preset addition sequence, while private sea customers with a waiting time greater than the preset time are designated as the pending sequence. Based on the pending sequence, a first pre-processed private sea customer is obtained, wherein the first pre-processed private sea customer is a private sea customer who purchases the same type of sales item within a preset time period. Based on the proportion of sales item types purchased by the first pre-processed private sea customer within the preset time period, interest-specific parameters for each private sea customer are obtained. Based on the interest-specific parameters and the purchase frequency of each private sea customer, an adjustment weight is obtained. Private sea customers with a weight threshold are added to the first preset addition sequence to obtain the fourth recommended public sea pool.

[0070] It should be noted that, based on the frequency and total amount of communication between public pool customers and target sales personnel's private pool customers, the data is obtained and input into a correlation analysis model trained using a deep model to obtain the correlation value between them. Those with correlation values ​​greater than a preset threshold are used as the third recommended public pool.

[0071] Based on the replacement duration calculation formula, the replacement duration of customers removed from the salesperson's private pool is obtained. This replacement duration is the time it takes for the customer to be reinstated into the public pool. The replacement duration formula allows customers with excessively long waiting times in the salesperson queue to enter the public pool earlier, and those with excessively short replacement durations to enter the public pool later. Here, ω can be a random adjustment factor within a certain range to increase the uncertainty when entering the public pool, where T... 标 Less than T max T0 in T 标 With T max You can send notifications to sales staff during the process.

[0072] All customers replaced by sales personnel are obtained. Based on a preset waiting time, customers in the private sea with a waiting time less than or equal to the preset waiting time are included in the first preset addition sequence. The interest-specific parameters of each private sea customer are obtained based on the first pre-processed private sea customers. If the interest-specific parameter is less than or equal to the preset parameter, it is used as the first weight of the private sea customer. For example, if the sales item type is lifestyle and the preset parameter is 0.4, and the purchase type of private sea customer A in the preset time period is lifestyle and fitness, then the interest-specific parameter of private sea customer A is 0.5. The interest-specific parameter is multiplied by the number of purchases to obtain the adjustment weight. Private sea customers with a weight greater than the weight threshold are put into the first preset addition sequence to obtain the fourth recommendation public sea pool.

[0073] S4: Based on the public sea customer recommendation pool, the public sea pool recommends customers to target sales personnel.

[0074] The intersection of the first, second, and third recommendation pools in the public customer recommendation pool is obtained as the fifth recommendation pool. Customers who have been replaced by the target salesperson are removed from the fourth recommendation pool, and the remaining customers are used as the sixth recommendation pool. The target salesperson is recommended based on the fifth and sixth recommendation pools.

[0075] Example 2: Based on Example 1, a customer relationship management system based on big data analysis includes:

[0076] The data preprocessing module is used to clean data from both public and private sea clients.

[0077] The cycle determination module is used to obtain the short-term or long-term cycle of customer purchases.

[0078] The conversion time acquisition module is used to acquire customer conversion times with long-term cycles;

[0079] The first recommendation acquisition module is used to acquire public customers that match historical short-term cyclical purchase data;

[0080] The second recommendation acquisition module is used to acquire customers among long-term, recurring customers who have the ability to be contacted by sales personnel in advance.

[0081] The third recommendation acquisition module is used to identify customers whose association with the target salesperson's private and public customer pools exceeds a preset association threshold.

[0082] The fourth recommendation module is used to obtain the customers that each salesperson can replace based on the replacement duration and adjustment parameters.

[0083] The recommendation module makes recommendations to users based on all other recommendation modules.

[0084] The present application has been described in detail above. Specific examples have been used to illustrate the principles and implementation methods of the present application. The description of the above embodiments is only for the purpose of helping to understand the method and core ideas of the present application. At the same time, for those skilled in the art, there will be changes in the specific implementation methods and application scope based on the ideas of the present application. Therefore, the content of this specification should not be construed as a limitation of the present application.

Claims

1. A customer relationship management method based on big data analysis, characterized in that, Includes the following steps: S1: Obtain public customer data and customer data in the sales personnel's private accounts, and preprocess them; S2: Based on customers' historical periodic purchase data, the public sea customer recommendation public sea pool includes a first recommendation public sea pool that matches unallocated public sea customers with public sea customers that meet the historical short-term periodic purchase data, and a second recommendation public sea pool that matches unallocated public sea customers with public sea customers that meet the historical long-term periodic purchase data and meet the early contact time. S3: Based on the correlation between customers in the target salesperson's private pool and customers in the public pool, the public pool recommends a third recommended public pool that matches the correlation between assigned private pool customers and unassigned public pool customers, and a fourth recommended public pool that replaces and reclaims customers in the salesperson's private pool. S4: Based on the public sea customer recommendation, the public sea pool recommends customers to target sales personnel; The S2 includes a first recommended public pool that matches unassigned public pool customers with public pool customers that meet the historical short-term periodic purchase data, and a second recommended public pool that matches unassigned public pool customers with public pool customers that meet the historical long-term periodic purchase data and the early contact time requirement. The S2 includes: obtaining the conversion time of each public pool user based on a Bayesian algorithm, wherein the early contact time is the conversion time of the public pool user, and obtaining the public pool users whose conversion time intersects with that of each public pool user within a first preset time period of the current time period as the second recommended public pool. S3 includes a third recommended public pool for matching the association between allocated private pool customers and unassigned public pool customers, and a fourth recommended public pool for sales personnel to replace and reclaim customers within their private pools. This includes: obtaining the waiting time for a private pool customer to become a public pool customer when a sales personnel replaces a private pool customer, based on a replacement time calculation formula; and sorting the customers according to the waiting time to obtain the fourth recommended public pool. The replacement time calculation formula is as follows: in It is the replacement duration. ω is the preset standard replacement duration, T0 is the customer's waiting time in the private sea, and T is the adjustment factor. 标 It is the preset standard waiting time, T max This is the maximum waiting time for customers in the private sea; The process of sorting customers according to their waiting time to obtain the fourth recommended public pool includes: using customers with a waiting time less than or equal to a preset waiting time as a first preset addition sequence, using customers with a waiting time greater than a preset waiting time as a pending sequence, obtaining a first pre-processed private customer based on the pending sequence, wherein the first pre-processed private customer is a private customer who purchased the same type of sales item within a preset time period, obtaining interest-specific parameters for each private customer based on the proportion of sales item types purchased by the first pre-processed private customer within the preset time period, obtaining adjustment weights based on the interest-specific parameters and the number of purchases by each private customer, and placing private customers with a weight threshold into the first preset addition sequence to obtain the fourth recommended public pool. The method for obtaining the conversion time of each high seas user based on the Bayesian algorithm includes: S201: Organize the historical records of each customer to form a sample set of conversion times; S202: Select a probability distribution model, using the gamma distribution, where the probability density function of the gamma distribution is: Where x is the conversion time, α is the shape parameter, β is the scale parameter, and Γ(α) is the gamma function; S203: Determine the prior distribution, and determine the initial probability distribution parameters as the prior distribution based on historical data; S204: Update the rules using Bayesian methods; the posterior distribution is: α n ＝α0+n； Where n is the number of observed samples, It is the sum of all observed transformation times; α0 is the initial value of the shape parameter; β0 is the initial value of the scale parameter; S205: Use the updated posterior distribution for prediction. The predicted purchase cycle is obtained by calculating the expected value of the posterior distribution. The expected value of the gamma distribution is: Where E[x] is the predicted value of the customer purchase cycle, α n and β n These are the updated shape and scale parameters.

2. The customer relationship management method based on big data analysis according to claim 1, characterized in that, The process of acquiring public customer data and sales personnel's private customer data, and preprocessing it, includes: data cleaning processing of customer data to remove duplicate values, fill in missing values, and remove outliers.

3. The customer relationship management method based on big data analysis according to claim 1, characterized in that, Before the public sea customer recommendation public sea pool, which is based on the customer's historical periodic purchase data, includes a first recommendation public sea pool matching unassigned public sea customers with public sea customers that meet the historical short-term periodic purchase data, and a second recommendation public sea pool matching unassigned public sea customers with public sea customers that meet the historical long-term periodic purchase data and the early contact time requirement, the process includes: obtaining the public sea customer's historical periodic purchase data; using fuzzy inference to obtain a periodic threshold judgment rule based on the public sea customer's historical periodic purchase data; and classifying the purchase periodicity of public sea customers according to the periodic threshold judgment rule to obtain short-term or long-term periodicity.

4. The customer relationship management method based on big data analysis according to claim 1, characterized in that, The public pool for recommending customers is based on the correlation between customers in the target salesperson's private pool and customers in the public pool. It includes a third recommended public pool for matching the correlation between assigned private pool customers and unassigned public pool customers, and a fourth recommended public pool for replacing and recycling customers in the salesperson's private pool. The process includes: obtaining the target salesperson's private pool, analyzing the target salesperson's customers based on a correlation analysis model trained on big data, obtaining the correlation value between private pool customers and public pool customers, and placing public pool customers with correlation values ​​greater than a preset correlation threshold into the third recommended public pool.

5. A customer relationship management method based on big data analysis according to claim 1, characterized in that, The method of recommending customers to target sales personnel based on the public customer recommendation pool includes: obtaining the intersection of the first recommendation pool, the second recommendation pool, and the third recommendation pool in the public customer recommendation pool as the fifth recommendation pool; removing customers who have been replaced by the target sales personnel from the fourth recommendation pool; and using the remaining customers as the sixth recommendation pool; and recommending customers to target sales personnel based on the fifth recommendation pool and the sixth recommendation pool.

6. A customer relationship management system based on big data analysis, used to implement the customer relationship management method based on big data analysis as described in any one of claims 1-5, characterized in that, include: The data preprocessing module is used to clean data from both public and private sea clients. The cycle determination module is used to obtain the short-term or long-term cycle of customer purchases. The conversion time acquisition module is used to acquire customer conversion times with long-term cycles; The first recommendation acquisition module is used to acquire public customers that match historical short-term cyclical purchase data; The second recommendation acquisition module is used to acquire customers among long-term, recurring customers who have the ability to be contacted by sales personnel in advance. The third recommendation acquisition module is used to identify customers whose association with the target salesperson's private and public customer pools exceeds a preset association threshold. The fourth recommendation module is used to obtain the customers that each salesperson can replace based on the replacement duration and adjustment parameters. The recommendation module makes recommendations to users based on all other recommendation modules.

Images