A traffic diversion method and system based on intelligent marketing reach

Abstract

The present invention discloses a diversion method and system based on intelligent marketing contact mode, and relates to the field of marketing contact technology. It discloses that according to the distribution of the contact data used by different types of customer groups, contact marketing content preference, contact mode preference, contact frequency preference and contact time preference are obtained; the revenue contribution and integrity of multiple customers are used to stratify the value of each customer; the contact mode preference is adjusted according to the higher the customer value, the higher the contact cost; it is determined whether the customer is a low-active card-using dormant user to improve the contact timeliness; multiple groups of contact training data are collected to train the contact training model, and the contact marketing content is pushed to multiple customers to be tested. The present invention can improve the marketing contact success rate and reduce the loss of labor and time costs.

Description

Technical Field

[0001] The present invention relates to the field of marketing reach technology, and in particular to a diversion method and system based on intelligent marketing reach mode. Background Art

[0002] The contact method is a key link in marketing activities. In actual business scenarios, more attention is paid to the marketing form and rights-related content of customer groups, while message contact adopts a common process, which leads to the loss of corresponding costs in the contact effect, especially in the contact methods of manual customer service, intelligent customer service, and text messages. There is a large loss of labor and reply time costs. Therefore, there is currently a lack of a diversion method and system based on intelligent marketing contact methods, which can realize the design of personalized contact methods, thereby diverting customer groups, improving the success rate of marketing contact, and reducing the loss of labor and time costs. Summary of the invention

[0003] The technical problem to be solved by the present invention is how to implement a method for designing personalized reach, so as to divert customer groups, improve the success rate of marketing reach, and reduce the loss in labor and time costs. The purpose is to provide a diversion method and system based on intelligent marketing reach to solve the above technical problems.

[0004] The present invention is achieved through the following technical solutions: A traffic diversion method based on intelligent marketing contact mode, including: Obtain historical transaction behavior data, reach data, active operation time and customer basic information of multiple customers; Different types of customer groups are obtained based on the above customer basic information, and the marketing content preference, contact method preference, contact frequency preference and contact time preference of each type of customer are obtained based on the distribution of the above contact data used by different types of customer groups; Obtain the revenue contribution and integrity of each customer based on the above historical transaction behavior data of multiple customers; stratify the value of each customer using the above revenue contribution and integrity of multiple customers; adjust the above contact method preference based on the fact that the higher the customer value, the higher the contact cost; Based on the above historical transaction behavior data of each customer, determine whether they are low-activity card dormant users; if they are, use any one or more of the following methods to improve the contact time efficiency: using the above contact method preference with high time efficiency, extending the above contact time preference of the customer, prioritizing the above contact time preference of the customer, and increasing the above contact frequency preference of the customer; Collect multiple groups of reach training data to train the reach training model; each group of the reach training data includes the historical transaction behavior data, the reach data, the active operation time and the basic customer information of each customer in the input layer, the customer value of each customer in the middle layer and the result of determining whether the customer belongs to a low-activity card-using dormant user; and the reach marketing content preference, the reach method preference, the reach frequency preference and the reach time preference of each customer in the output layer; obtain the reach marketing content preference, the reach method preference, the reach frequency preference and the reach time preference of multiple customers to be tested through the reach training model; According to the above-mentioned contact marketing content preference, the above-mentioned contact method preference, the above-mentioned contact frequency and the above-mentioned contact time preference of the multiple customers to be tested, the contact marketing content is pushed to the multiple customers to be tested.

[0005] According to the above customer basic information, different types of customer groups are obtained. According to the distribution of different types of customer groups using the above contact data, the contact marketing content preference, contact method preference, contact frequency preference and contact time preference of each type of customer are obtained, including: The above-mentioned reach data includes multiple reach marketing contents and browsing time, multiple reach methods and reach frequency, and reach active time period; Divide customer types based on any one or more of the above-mentioned basic customer information; calculate the proportion of browsing time of the same type of customers for each of the above-mentioned contact marketing contents, and select the above-mentioned contact marketing content preference of each customer based on the proportion of browsing time; calculate the contact frequency of the same type of customers for each of the above-mentioned contact methods, and obtain the contact frequency preference of each customer for each contact method; calculate the proportion of the frequency of each of the above-mentioned contact methods for the same type of customers, and select the above-mentioned contact method preference based on the proportion of the contact frequency; calculate the proportion of the above-mentioned contact active time periods for the same type of customers, and obtain the contact time preference of each customer for each contact method.

[0006] The above basic customer information includes age, gender and industry.

[0007] The above historical transaction behavior data includes AUM value and customer loan repayment status; The revenue contribution and integrity of each customer obtained based on the historical transaction behavior data of multiple customers include: Calculate the income contribution based on the AUM value of each customer; calculate the integrity based on the loan repayment situation of the above customers; The above-mentioned use of the above-mentioned revenue contribution and the above-mentioned integrity of multiple customers to stratify the value of each customer includes: The above-mentioned income contribution and the above-mentioned integrity are combined with the same weight to obtain different levels of customer value.

[0008] Based on the fact that the higher the customer value, the higher the reach cost, the above reach method preferences are adjusted, including: The reaching methods include high-cost reaching and low-cost reaching. The reaching methods with corresponding high and low costs are set according to the above customer value; or If the customer value is high, the types of contact methods will be increased; if the customer value is low, the types of contact methods will be reduced.

[0009] The above historical transaction behavior data include the average monthly transaction amount, the average monthly number of transactions and the time of a single transaction.

[0010] The above analysis of each customer's recent transaction situation based on the above historical transaction behavior data of multiple customers includes: Calculate the average monthly transaction amount based on the average monthly transaction amount of each customer; calculate the average monthly transaction number based on the average monthly transaction number of each customer; obtain the most recent transaction time based on the time of a single transaction; The above historical transaction behavior data of each customer is used to determine whether they are low-activity dormant card users, including: When each customer's average monthly transaction amount, average monthly transaction number, or the time of the most recent transaction meets the preset threshold range of the minimum transaction activity level on a month-on-month basis, the customer will be determined to be a dormant card user.

[0011] A design system based on intelligent marketing contact methods, including: Data collection module, which obtains historical transaction behavior data, reach data, operation active time and customer basic information of multiple customers; The preference analysis module obtains different types of customer groups based on the above customer basic information, and obtains the marketing content preference, contact method preference, contact frequency preference and contact time preference of each type of customer based on the distribution of the above contact data used by different types of customer groups; A value stratification module, which obtains the revenue contribution and integrity of each customer based on the above historical transaction behavior data of multiple customers; stratifies the value of each customer using the above revenue contribution and integrity of multiple customers; and adjusts the above contact method preference based on the fact that the higher the customer value, the higher the contact cost; The dormant contact module determines whether a customer is a dormant card user with low activity based on the historical transaction behavior data of each customer. If the customer is a dormant card user, the dormant contact module improves the contact time efficiency by using any one or more of the above contact method preferences with high time efficiency, extending the contact time preference of the customer, prioritizing the contact time preference of the customer, and increasing the contact frequency preference of the customer; A reach training module collects multiple groups of reach training data to train a reach training model; each group of the reach training data includes the historical transaction behavior data, the reach data, the active operation time and the basic customer information of each customer in the input layer, the customer value of each customer in the middle layer and the result of determining whether the customer is a low-active card-using dormant user; and the reach marketing content preference, the reach method preference, the reach frequency preference and the reach time preference of each customer in the output layer; the reach marketing content preference, the reach method preference, the reach frequency preference and the reach time preference of multiple customers to be tested are obtained through the reach training model; The reach diversion module pushes reach marketing content to multiple customers to be tested based on the reach marketing content preference, reach method preference, reach frequency and reach time preference of the multiple customers to be tested.

[0012] An electronic device includes a memory, a processor, and a computer program running on the processor. When the processor executes the computer program, the steps of the diversion method based on intelligent marketing reach are implemented.

[0013] A computer-readable storage medium storing a computer program, wherein the computer program is executed by a processor to implement the steps of the above-mentioned diversion method based on intelligent marketing contact mode. Compared with the prior art, the present invention has the following advantages and beneficial effects: The present invention provides a diversion method based on intelligent marketing contact mode, which obtains different types of customer groups according to basic customer information, and obtains contact marketing content preference, contact mode preference, contact frequency preference and contact time preference of each customer according to the browsing time distribution of different types of customer groups in contact data; obtains the income contribution and integrity of each customer according to the historical transaction behavior data of multiple customers, and uses the income contribution and integrity of multiple customers to stratify the value of each customer, and selects a high-cost preferred contact mode for high-value customers, thereby optimizing the contact cost of high-value customers in combination with customer contribution and integrity; analyzes the recent transaction activity of each customer according to the historical transaction behavior data of multiple customers, and uses the recent transaction activity of each customer to determine whether the customer is currently a card dormant user, and by extending or advancing the marketing content of the card dormant user Reach time, or increase the frequency of reach, so as to set a more timely reach strategy according to low-activity customers, and effectively promote the improvement of customer activity through efficient reach marketing methods; collect multiple groups of reach training data through the above process, and use the collected customer's historical transaction behavior data, reach data, operation active time and customer basic information as input, analyze the customer value and determine whether it is a low-activity card-using dormant user as the intermediate feature, and obtain and adjust the reach marketing content preference, reach method preference, reach frequency preference and reach time preference as output to obtain a reach training model for designing customer personalized marketing reach strategies; according to the output of each customer's reach marketing content preference, reach method preference, reach frequency preference and reach time preference, classify multiple customers to be tested, so as to conduct marketing diversion reach to multiple customers in batches. This application uses the customer group to conduct overall feature analysis preferences, and selects two marketing strategies of cost priority and time priority based on the customer's transaction content characteristics, active characteristics and multi-dimensional performance of customer value, so as to obtain corresponding reach strategies according to different branches and conduct marketing diversion. The present invention can realize personalized marketing diversion according to the actual needs of customer groups, improve the success rate of marketing contact by adaptively adjusting the contact method and strategy, and reduce the loss of labor and time costs. BRIEF DESCRIPTION OF THE DRAWINGS

[0014] In order to more clearly illustrate the technical solutions of the exemplary embodiments of the present invention, the following briefly introduces the drawings required for use in the embodiments. It should be understood that the following drawings only illustrate certain embodiments of the present invention and should not be regarded as limiting the scope. For those of ordinary skill in the art, other relevant drawings can be obtained based on these drawings without creative work. In the drawings: Figure 1 This is a flow chart of the diversion method based on the intelligent marketing reach method in an embodiment of the present application. DETAILED DESCRIPTION

[0015] In order to make the objectives, technical solutions and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with embodiments and drawings. The exemplary embodiments of the present invention and their description are only used to explain the present invention and are not intended to limit the present invention. Example

[0016] like Figure 1 As shown, the embodiment of the present application provides a traffic diversion method based on an intelligent marketing reach mode, including: Obtain historical transaction behavior data, reach data, active operation time and customer basic information of multiple customers; Different types of customer groups are obtained based on the above customer basic information, and the marketing content preference, contact method preference, contact frequency preference and contact time preference of each type of customer are obtained based on the distribution of the above contact data used by different types of customer groups; Obtain the revenue contribution and integrity of each customer based on the above historical transaction behavior data of multiple customers; stratify the value of each customer using the above revenue contribution and integrity of multiple customers; adjust the above contact method preference based on the fact that the higher the customer value, the higher the contact cost; Based on the above historical transaction behavior data of each customer, determine whether they are low-activity card dormant users; if they are, use any one or more of the following methods to improve the contact time efficiency: using the above contact method preference with high time efficiency, extending the above contact time preference of the customer, prioritizing the above contact time preference of the customer, and increasing the above contact frequency preference of the customer; Collect multiple groups of reach training data to train the reach training model; each group of the reach training data includes the historical transaction behavior data, the reach data, the active operation time and the basic customer information of each customer in the input layer, the customer value of each customer in the middle layer and the result of determining whether the customer belongs to a low-activity card-using dormant user; and the reach marketing content preference, the reach method preference, the reach frequency preference and the reach time preference of each customer in the output layer; obtain the reach marketing content preference, the reach method preference, the reach frequency preference and the reach time preference of multiple customers to be tested through the reach training model; According to the above-mentioned contact marketing content preference, the above-mentioned contact method preference, the above-mentioned contact frequency and the above-mentioned contact time preference of the multiple customers to be tested, the contact marketing content is pushed to the multiple customers to be tested.

[0017] According to the above customer basic information, different types of customer groups are obtained. According to the distribution of different types of customer groups using the above contact data, the contact marketing content preference, contact method preference, contact frequency preference and contact time preference of each type of customer are obtained, including: The above-mentioned reach data includes multiple reach marketing contents and browsing time, multiple reach methods and reach frequency, and reach active time period; Divide customer types based on any one or more of the above-mentioned basic customer information; calculate the proportion of browsing time of the same type of customers for each of the above-mentioned contact marketing contents, and select the above-mentioned contact marketing content preference of each customer based on the proportion of browsing time; calculate the contact frequency of the same type of customers for each of the above-mentioned contact methods, and obtain the contact frequency preference of each customer for each contact method; calculate the proportion of the frequency of each of the above-mentioned contact methods for the same type of customers, and select the above-mentioned contact method preference based on the proportion of the contact frequency; calculate the proportion of the above-mentioned contact active time periods for the same type of customers, and obtain the contact time preference of each customer for each contact method.

[0018] The above basic customer information includes age, gender and industry.

[0019] The above historical transaction behavior data includes AUM value and customer loan repayment status; The revenue contribution and integrity of each customer obtained based on the historical transaction behavior data of multiple customers include: Calculate the income contribution based on the average monthly transaction amount of each customer; calculate the creditworthiness based on the above-mentioned customer loan repayment situation of each customer; The above-mentioned use of the above-mentioned revenue contribution and the above-mentioned integrity of multiple customers to stratify the value of each customer includes: The above-mentioned income contribution and the above-mentioned integrity are combined with the same weight to obtain different levels of customer value.

[0020] Based on the fact that the higher the customer value, the higher the reach cost, the above reach method preferences are adjusted, including: The reaching methods include high-cost reaching and low-cost reaching. The reaching methods with corresponding high and low costs are set according to the above customer value; or If the customer value is high, the types of contact methods will be increased; if the customer value is low, the types of contact methods will be reduced.

[0021] The above historical transaction behavior data include the average monthly transaction amount, the average monthly number of transactions and the time of a single transaction.

[0022] The above analysis of each customer's recent transaction situation based on the above historical transaction behavior data of multiple customers includes: Calculate the average monthly transaction amount based on the average monthly transaction amount of each customer; calculate the average monthly transaction number based on the average monthly transaction number of each customer; obtain the most recent transaction time based on the time of a single transaction; The above historical transaction behavior data of each customer is used to determine whether they are low-activity dormant card users, including: When each customer's average monthly transaction amount, average monthly transaction number, or the time of the most recent transaction meets the preset threshold range of the minimum transaction activity level on a month-on-month basis, the customer will be determined to be a dormant card user.

[0023] Optionally, set the reach time priority strategy and reach cost priority strategy by building a strategy executor. Use the reach data to analyze the preferences of various types of customers for reach marketing content, reach methods, reach frequency, and reach time. At the same time, combine the classification of basic customer characteristics such as age, gender, and industry to conduct statistics, and use the distribution of various customer groups to obtain the reach marketing content preferences, reach method preferences, reach frequency preferences, and reach time preferences of different types of customers based on the contribution of each customer's preference information. Build a complete customer preference dimension through basic customer information and reach data.

[0024] For example, the contact method preference can be determined by the number of times each customer clicks on the app and the number of times the customer's behavior sends and receives text messages in the uplink and downlink of the text message, which determines whether the contact method preference belongs to the app or the text message. Optionally, the contact method preference is determined based on whether the number of multiple contact methods exceeds the preset threshold, or the one with the highest number of times is selected. Therefore, when the proportion of customers of the same type who prefer various contact methods exceeds the threshold or has the largest proportion, the contact method is selected as the contact frequency preference for this type of customer. The threshold can be set to 1 / 2. When there is no contact frequency preference exceeding 1 / 2, the threshold is set to the proportion of the total number of contact method preferences of customers of the same type. For example, if there are a total of 3 contact methods for customers of the same type, the threshold is set to 1 / 3. The contact frequency preference of each customer for various contact methods can be the frequency of use of each contact method, such as the frequency of clicking on the app and the frequency of sending and receiving text messages. The average of the different contact frequencies of the same type of customers for various contact methods is used as the contact frequency preference; or, the proportion of the different contact frequencies of the same type of customers for various contact methods is used to select the contact frequency preference of various contact methods that exceeds the threshold or the maximum value. The contact marketing content preference can be determined by the browsing time ratio of the app behavior habit data and the contact form context data of each customer. When the proportion of each contact marketing content is greater than the preset threshold or exceeds half, it is used as the contact marketing content preference. For the contact marketing content preference of the same type of customers, the proportion of each contact marketing content is counted, and the data with a proportion greater than the preset threshold or a high proportion is used as the contact marketing content; the threshold can be set to 1 / 2, or the proportion of each contact marketing content preference to the total number of all contact marketing content preferences of the same type of customers. The contact time preference can be determined by the proportion of each customer's daily active operation time period. For example, when the number of times the user's daily active operation time period is repeated exceeds half of the days in the month, the customer's contact time preference is obtained. In addition, according to the reach time preferences of multiple customers of the same type, repeated time periods are counted. When multiple partially overlapping time periods are included, they are merged into one time period, and one or more independent and non-overlapping time periods are obtained as the reach time preferences. Among them, when counting the repeated time periods of customers of the same type, the corresponding number of repetitions can be set, such as 2 or 3, so as to collect effective time periods. The elderly and working people have different active times due to sleep, physical strength, free time and other reasons. Therefore, classifying customers according to their basic characteristics can make use of the reach preference rules of the corresponding characteristic groups for statistics, and divert the reach of the groups according to the reach method, reach frequency and reach time, thereby improving the overall reach timeliness. Optionally, the basic characteristics of customers can also specifically include diversified information such as sleep time and health status.

[0025] The higher the customer value, the more expensive the contact method will be, such as by increasing the types of contact methods or setting high-cost contact methods. The cost of the above contact methods can be adjusted by the marketing party based on actual time and labor costs. For example, online paid advertising and manual customer service telephones are high-cost; smart customer service, SMS, apps, and WeChat message push are low-cost.

[0026] According to the aum value, that is, the asset management scale, the customer transaction amount can be obtained, and the proportion of each customer's income contribution to the total income of the institution can be calculated as the income contribution. The aum value refers to the proportion of the total market value of customer assets currently managed by financial institutions. The customer's integrity is calculated through the customer's loan repayment situation, for example, the proportion of the number of times of compliance is used as the integrity level. According to the different intervals of income contribution and integrity from large to small, the corresponding customer value level is obtained. Customers with high contribution value are the focus group. By choosing a higher-cost contact method or increasing the contact method to increase the cost, low-value customers can choose a lower-cost contact method or reduce the number of contact methods to reduce the cost, thereby improving marketing efficiency, maintaining high-value customers, and reducing the consumption of unnecessary costs for customers with low contribution value. When increasing the cost above, the original contact method preference can be judged according to the specified cost high and low rules. If it does not belong to it, it can be changed to high cost. When choosing a higher or lower cost contact method, the cost high and low rules can be set according to the specified contact method or the specified number of contact methods.

[0027] Based on the performance of historical transaction behavior data, it is determined whether the customer has no transactions or few transactions recently, and the customer is judged to be a low-activity card dormant customer. For example, if the average monthly transaction amount and number of transactions decrease by 50% month-on-month, and the time of the most recent transaction is more than 3 months ago, they can be judged as card dormant customers. Based on the customer's active time preference, high-efficiency adjustments are made to low-activity users according to transaction performance. After adjusting the basis of the reach cost according to customer value, low-activity users adopt a more time-efficient reach method, which can prevent the loss of low-activity customers and improve the marketing success rate of low-activity users. The timeliness of the above-mentioned reach methods can be judged based on the historical success rate of various methods. For example, the timeliness of marketing reach through telephone or visit is high, while the timeliness through app or text message is low.

[0028] When applied, different types of customer groups can be obtained based on the basic customer information, which can be implemented using the K-Means clustering algorithm. According to the distribution of the above-mentioned reach data used by different types of customer groups, when the reach marketing content preferences, reach method preferences, reach frequency preferences and reach time preferences of each type of customer are obtained, the decision tree algorithm can be used to obtain the corresponding probabilities of each customer's reach marketing content preferences, reach method preferences, reach frequency preferences and reach time preferences. According to the above-mentioned reach marketing content preferences, the above-mentioned reach method preferences, the above-mentioned reach frequency and the above-mentioned reach time preferences of multiple customers to be tested, the reach marketing content is pushed to multiple customers to be tested, which can be executed by building an overall strategy executor. Different from the process of traditional methods, through the analysis scenarios reached between customer characteristics and reach strategies, the reach feedback information is automatically collected and a new round of automatic analysis is performed, and the customer characteristics and feedback are continuously adjusted, so as to optimize the reach strategy method that is more suitable for each individual customer.

[0029] When combining multiple characteristic variables of customer groups to obtain each customer's preference for reach marketing content, reach method, reach frequency and reach time, since each characteristic variable has a certain percentage threshold, through the distribution of reach data corresponding to the overall customers, the reach strategy adjusted by cost and time priority rules is collected through the reach training model. The training process can be used to evaluate the comprehensive scores of multiple characteristic variables corresponding to each customer group, so as to obtain the optimal timeliness reach strategy corresponding to the highest score of multiple characteristic variables according to the proportion of each preference.

[0030] When pushing marketing content to multiple customers to be tested based on their preferences for marketing content, methods, frequency and time, you can divert customers based on their preferences for marketing content, methods and time in the same corresponding time period, and reach them based on their frequency, using one or more methods to reach them every half an hour.

[0031] When a customer reaches and clicks, it proves that the reach marketing strategy is successful. Therefore, when adjusting the reach method preference based on the fact that the higher the customer value, the higher the reach cost, the reach method preference can be directly confirmed by this type of customer, or one or more of the reach method preferences can be selected based on the reach click results fed back by this type of customer for the corresponding reach method, or a new reach method preference can be replaced to optimize marketing for customers of different values. When improving the reach timeliness by using any one or more of the following methods, such as using a reach method preference with high timeliness, extending the reach time preference of the customer, prioritizing the reach time preference of the customer, and increasing the reach frequency preference of the customer, continuous optimization can be performed based on the direct confirmation of this type of customer, or the reach click results fed back by this type of customer after the corresponding reach method, reach time or reach frequency are adjusted.

[0032] The embodiment of the present application classifies customer groups according to their historical transaction behavior data and basic customer information, adjusts the reach strategy for each customer based on customer value and active features, and quickly formulates reach strategies after unified training through the reach training model, thereby using different reach strategies for subsequent diversion marketing reach, thereby forming a complete closed-loop strategy processing. Provide guidance on cost- and timeliness-prioritized reach strategy models for intelligent reach marketing, and select the most reasonable reach method according to strategy requirements. By continuously optimizing the reach data model and feature model, and optimizing the training of the reach training model, more accurate marketing strategy processing can be found. Relevant reach strategy settings and marketing diversion reach can be quickly performed, and the operation is simple and convenient.

[0033] In summary, the embodiments of the present application provide a traffic diversion method and system based on intelligent marketing reach: The embodiments of the present application effectively divert the crowd by performing overall feature analysis and strategy selection with the help of the selected customer groups, and select appropriate contact methods according to different features to complete the intelligent processing of marketing contact. Specifically disclosed are different types of customer groups obtained according to basic customer information, and the distribution of browsing time of contact data of different types of customer groups, so as to obtain the contact marketing content preference, contact method preference, contact frequency preference and contact time preference of each customer; the income contribution and integrity of each customer are obtained according to the historical transaction behavior data of multiple customers, and the income contribution and integrity of multiple customers are used to stratify the value of each customer, and a high-cost preferred contact method is selected for high-value customers, so as to optimize the contact cost of high-value customers in combination with customer contribution and integrity; the recent transaction activity of each customer is analyzed according to the historical transaction behavior data of multiple customers, and the recent transaction activity of each customer is used to determine whether the customer is currently a dormant card user, by extending or advancing the contact time of marketing content for dormant card users, or improving Reach frequency, so as to set a more timely reach strategy according to low-activity customers, and effectively promote the improvement of customer activity through efficient reach marketing methods; collect multiple sets of reach training data through the above process, and use the collected customer's historical transaction behavior data, reach data, operation active time and customer basic information as input, analyze the customer value and determine whether it is a low-activity card-using dormant user as the intermediate feature, and obtain and adjust the reach marketing content preference, reach method preference, reach frequency preference and reach time preference as output to obtain a reach training model for designing customer personalized marketing reach strategies; according to the output of each customer's reach marketing content preference, reach method preference, reach frequency preference and reach time preference, classify multiple customers to be tested, so as to conduct marketing diversion reach to multiple customers in batches. This application uses the customer group to conduct overall feature analysis preferences, and selects two marketing strategies of cost priority and time priority based on the customer's transaction content characteristics, active characteristics and multi-dimensional performance of customer value, so as to obtain corresponding reach strategies according to different branches and conduct marketing diversion. The present invention can realize personalized marketing diversion according to the actual needs of customer groups, improve the success rate of marketing contact by adaptively adjusting the contact method and strategy, and reduce the loss of labor and time costs.

[0034] The specific implementation methods described above further illustrate the objectives, technical solutions and beneficial effects of the present invention in detail. It should be understood that the above description is only a specific implementation method of the present invention and is not intended to limit the scope of protection of the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the scope of protection of the present invention.

Claims

1. A traffic diversion method based on intelligent marketing contact mode, characterized in that: include: Obtain historical transaction behavior data, reach data, active operation time and customer basic information of multiple customers; Different types of customer groups are obtained based on the basic customer information, and the marketing content preference, contact method preference, contact frequency preference and contact time preference of each type of customer are obtained based on the distribution of the contact data used by different types of customer groups; Obtaining the revenue contribution and integrity of each customer based on the historical transaction behavior data of multiple customers; and stratifying the value of each customer using the revenue contribution and integrity of multiple customers; According to the fact that the higher the customer value, the higher the reaching cost, adjusting the reaching method preference; According to the historical transaction behavior data of each customer, determine whether they are low-activity card dormant users; if they are, use any one or more of the following methods to improve the contact time efficiency: adopt the contact method preference with high time efficiency, extend the contact time preference of the customer, give priority to the contact time preference of the customer, and increase the contact frequency preference of the customer; Collect multiple sets of reach training data to train the reach training model; each set of the reach training data includes the historical transaction behavior data, the reach data, the active operation time and the basic customer information of each customer in the input layer, the customer value of each customer in the middle layer and the result of determining whether the customer belongs to a low-activity card-using dormant user; and the marketing content preference, the contact method preference, the contact frequency preference and the contact time preference of each customer in the output layer; The contact marketing content preference, the contact method preference, the contact frequency preference and the contact time preference of multiple customers to be tested are respectively obtained through the contact training model; According to the marketing content preference, the contact method preference, the contact frequency and the contact time preference of the multiple customers to be tested, the marketing content is pushed to the multiple customers to be tested.

2. A traffic diversion method based on intelligent marketing contact mode according to claim 1, characterized in that: The method of obtaining different types of customer groups based on the customer basic information and obtaining the marketing content preference, contact method preference, contact frequency preference and contact time preference of each type of customer based on the distribution of the contact data used by different types of customer groups includes: The reach data includes multiple reach marketing contents and browsing time, multiple reach methods and reach frequency, and reach active time period; Divide customer types according to any one or more of the basic customer information; calculate the proportion of browsing time of the same type of customers for each of the contact marketing contents, and select the contact marketing content preference of each customer according to the proportion of browsing time; calculate the contact frequency of the same type of customers for each of the contact methods, and obtain the contact frequency preference of each customer for each contact method; calculate the frequency proportion of each of the contact methods for the same type of customers, and select the contact method preference according to the contact frequency proportion; calculate the proportion of the active contact time period for the same type of customers, and obtain the contact time preference of each customer for each contact method.

3. A traffic diversion method based on intelligent marketing contact mode according to claim 2, characterized in that: The basic customer information includes age, gender and industry.

4. The method for diverting traffic based on intelligent marketing contact mode according to claim 1, characterized in that: The historical transaction behavior data includes the aum value and the customer's loan repayment status; The obtaining of the revenue contribution and integrity of each customer based on the historical transaction behavior data of multiple customers includes: Calculate the income contribution based on the AUM value of each customer; calculate the integrity based on the loan repayment situation of each customer; The method of stratifying the value of each customer by using the revenue contribution and integrity of multiple customers includes: The income contribution and the integrity are combined with the same weight to obtain different levels of customer value.

5. The method for diverting traffic based on intelligent marketing contact mode according to claim 4, characterized in that: The adjusting the contact method preference according to the fact that the higher the customer value, the higher the contact cost, includes: The reaching methods include high-cost reaching and low-cost reaching; setting corresponding high-cost and low-cost reaching methods according to the value of the customer; or If the customer value is high, the types of contact methods are increased; if the customer value is low, the types of contact methods are reduced.

6. The method for diverting traffic based on intelligent marketing contact mode according to claim 1, characterized in that: The historical transaction behavior data includes the average monthly transaction amount, the average monthly number of transactions and the time of a single transaction.

7. The method for diverting traffic based on intelligent marketing contact mode according to claim 6, characterized in that: Analyzing the recent transaction status of each customer based on the historical transaction behavior data of multiple customers includes: Calculate the average monthly transaction amount based on the average monthly transaction amount of each customer; calculate the average monthly transaction number based on the average monthly transaction number of each customer; obtain the most recent transaction time based on the time of a single transaction; Determining whether a customer is a low-activity dormant card user based on the historical transaction behavior data of each customer includes: When each customer's average monthly transaction amount, average monthly transaction number, or the time of the most recent transaction meets the preset threshold range of the minimum transaction activity level on a month-on-month basis, the customer will be determined to be a dormant card user.

8. A design system based on intelligent marketing contact mode, characterized in that: include: Data collection module, which obtains historical transaction behavior data, reach data, operation active time and customer basic information of multiple customers; A preference analysis module, which obtains different types of customer groups based on the basic customer information, and obtains the marketing content preference, contact method preference, contact frequency preference and contact time preference of each type of customer based on the distribution of the contact data used by different types of customer groups; A value stratification module, which obtains the income contribution and integrity of each customer based on the historical transaction behavior data of multiple customers; and stratifies the value of each customer using the income contribution and integrity of multiple customers; According to the fact that the higher the customer value, the higher the reaching cost, adjusting the reaching method preference; The contact sleep module determines whether the customer is a low-activity card sleep user based on the historical transaction behavior data of each customer; if the customer is a low-activity card sleep user, the contact time efficiency is improved by using any one or more of the contact method preferences with high time efficiency, extending the contact time preference of the customer, prioritizing the contact time preference of the customer, and increasing the contact frequency preference of the customer; A reach training module collects multiple sets of reach training data to train a reach training model; each set of the reach training data includes the historical transaction behavior data, the reach data, the active operation time and the basic customer information of each customer in the input layer, the customer value of each customer in the middle layer and the result of determining whether the customer belongs to a low-activity card dormant user; and the marketing content preference, the contact method preference, the contact frequency preference and the contact time preference of each customer in the output layer; The contact marketing content preference, the contact method preference, the contact frequency preference and the contact time preference of multiple customers to be tested are respectively obtained through the contact training model; The reach diversion module pushes reach marketing content to multiple customers to be tested according to the reach marketing content preference, the reach method preference, the reach frequency and the reach time preference of the multiple customers to be tested.

9. An electronic device comprising a memory, a processor, and a computer program running on the processor, characterized in that: When the processor executes the computer program, it implements the steps of a diversion method based on intelligent marketing reach as described in any one of claims 1 to 7.

10. A computer-readable storage medium storing a computer program, characterized in that: When the computer program is executed by the processor, the steps of a diversion method based on intelligent marketing reach as described in any one of claims 1 to 7 are implemented.

Images