Intelligent customer follow-up method and system

Abstract

The invention discloses an intelligent customer follow-up method and system, and belongs to the technical field of customer relationship management. The method comprises the following steps: constructing a tag system, defining tag attributes and configuring tag weights; client behaviors and triggered business events are collected, and labels are generated or updated for clients; classifying the clients according to the labels; setting a customer sales life cycle for each type of customers, and configuring an initial follow-up strategy for each stage of the sales life cycle; and performing intelligent follow-up on the customer, including: judging the customer life cycle stage of the customer; a corresponding initial follow-up strategy is matched for follow-up according to the client life cycle stage; and establishing a customer follow-up record in the follow-up process, and performing follow-up after adjusting the initial follow-up strategy according to effective follow-up in the customer follow-up record. In the follow-up process, feedback is analyzed in real time according to customer follow-up records, follow-up effectiveness is judged by means of a language model, strategies are dynamically adjusted, and follow-up efficiency and effect are improved.

Description

Technical Field

[0001] The present invention belongs to the technical field of customer relationship management, and in particular relates to an intelligent customer follow-up method and system. Background Art

[0002] The current mainstream customer relationship management (CRM) system focuses on customer management and has certain automated follow-up functions, including: customer information management, task reminders and follow-up plans, intelligent marketing and automated push, and preliminary exploration of intelligent marketing platforms. Although the existing CRM system has made great progress in customer management and marketing automation, it still has the following deficiencies in terms of intelligence, refinement and multi-channel collaboration: The first is the lack of multi-channel synchronization and information closed loop. Existing CRM systems usually focus on the management of a single channel (such as telephone, email, etc.), and fail to fully open up the information flow between multiple channels (such as WeCom, the institution's own system, etc.). Customer follow-up information is usually concentrated in only one link (such as account managers or marketing platforms), lacking a unified, cross-channel real-time information sharing mechanism, and unable to form an information closed loop. Participation in customer follow-up is usually independent, and the platform lacks the ability to push customer status and tasks in real time.

[0003] The second is the limitation of the automated follow-up strategy. The existing system’s tag trigger mechanism and follow-up strategy are relatively simple and lack dynamic adjustment capabilities. For scenarios where priorities change dynamically or require flexible adjustments, the system is not responsive enough and cannot meet complex needs.

[0004] Prior art Chinese patent application CN202411593613.0 discloses a method, device and electronic device for processing customer follow-up information, the method comprising: obtaining customer follow-up information currently to be processed in a customer management system; performing sentiment analysis and prediction on the customer follow-up information through a preset sentiment analysis model, and outputting sentiment classification labels corresponding to the customer follow-up information; determining the target customer indicator type corresponding to the customer follow-up information based on the sentiment classification labels corresponding to the customer follow-up information; based on the sentiment classification labels and the target customer indicator type, counting the customer follow-up information into existing dashboard data; the dashboard data includes statistical data under a single label dimension and / or a combination of multiple label dimensions, and after monitoring an event that meets the decision trigger condition in the dashboard data, providing a decision reminder to the salesperson.

[0005] In the above prior art, by labeling customers and analyzing the labels, sales staff are reminded to make decisions. However, the follow-up of customers actually still depends on the sales staff, and automatic intelligent follow-up is not truly achieved. Summary of the invention

[0006] The objective of the present invention is to provide an intelligent customer follow-up method and system, which partially solve or alleviate the above deficiencies in the prior art and can achieve intelligent follow-up of customers.

[0007] To solve the above-mentioned technical problems, the present invention specifically adopts the following technical solutions: An intelligent customer follow-up method, including: Construct a label system, define label attributes and configure label weights; Collect customer behaviors and triggered business events, and generate or update labels for customers; Classify customers according to labels; set a customer sales life cycle for each category of customers, and configure an initial follow-up strategy for each stage of the sales life cycle; Conduct intelligent follow-up on customers, including: Judge the stage of the customer life cycle that this category of customers is in; and match the corresponding initial follow-up strategy according to the stage of the customer life cycle for follow-up; Establish a customer follow-up record during the follow-up process, and adjust the initial follow-up strategy according to the effective follow-up in the customer follow-up record and then conduct follow-up.

[0008] As an improvement, the label attributes include basic attribute labels, behavioral feature labels, and recommendation labels.

[0009] As an improvement, when classifying customers, sort the labels according to their weights; and select the labels with a ranking higher than the ranking threshold to participate in the classification.

[0010] As an improvement, the calculation method of the label weight includes using the formula: ω = ω t * ω p Calculate the label weight; where ω is the label weight, ω t is the timeliness weight coefficient, ω p is the business configuration weight; Using the formula: =

[0011] Calculate the timeliness weight coefficient; where ω t is the timeliness weight coefficient, is the initial weight, t is the time, and c is the offset.

[0012] As an improvement, the formulation method of the initial follow-up strategy includes: Retrieve the follow-up records of a certain category of closed customers, including the touch time, touch method, and chat content; Use a language model to analyze the chat content and mark the effective follow-up; Use clustering algorithms to cluster the effective follow-up marks in the follow-up records of closed customers to obtain the effective follow-up contact time range and contact method types; Build an initial follow-up strategy based on the effective follow-up time range and types of contact methods.

[0013] As an improvement, methods to adjust the initial follow-up strategy based on the effective follow-up in the customer follow-up records include: Retrieve a customer's follow-up records, including contact time, contact method, and chat content; Use the language model to analyze the chat content and determine whether the follow-up in the follow-up record is effective; If follow-up is effective, continue with the initial follow-up strategy; If follow-up is ineffective, switch to other times and methods of contact.

[0014] As an improvement, set up a blacklist and add customers who do not want to be disturbed to the blacklist; read the blacklist before following up, and do not follow up with customers on the blacklist; Establish risk control and set a minimum interval between two follow-ups for a single customer.

[0015] As an improvement, methods for determining the customer life cycle stage include: When there is a landmark event at a certain stage in the customer's sales lifecycle, determine whether the customer is at that stage of the sales lifecycle based on the customer's label; When there is no landmark event at a certain stage in the customer sales lifecycle, the time period of the stage is obtained by clustering the follow-up records of successful customers; and whether the customer is in this stage is determined by whether the customer's current time period coincides with the time period.

[0016] As an improvement, when the number of users who have completed transactions of a certain type is less than the quantity threshold, an operation strategy is used for manual follow-up; When the number of users of a certain type who have completed transactions is greater than or equal to the threshold, intelligent follow-up will be adopted for some of these users; When the transaction rate of smart follow-up is higher than that of manual follow-up, such customers who are manually followed up will be switched to smart follow-up.

[0017] The present invention also provides an intelligent customer follow-up system, comprising: The tag building module is used to build a tag system, define tag attributes and configure tag weights; collect customer behaviors and triggered business events, and generate or update tags for customers; The customer classification module is used to classify customers according to tags; set the customer sales life cycle for each type of customer, and configure the initial follow-up strategy for each stage of the sales life cycle; Intelligent follow-up module, used for intelligent follow-up of customers, including: Determine the customer life cycle stage of this type of customer; and match the corresponding initial follow-up strategy according to the customer life cycle stage; During the follow-up process, a customer follow-up record is established, and follow-up is carried out after adjusting the initial follow-up strategy based on the effective follow-up in the customer follow-up record.

[0018] Compared with the prior art, the beneficial effects of the present invention include: 1. Accurately identify customer needs.

[0019] By building a sophisticated tag system, we can comprehensively collect customer behaviors and business events, accurately "profile" customers, and gain in-depth insights into customer preferences, demand stages, and purchase intentions. Instead of relying on broad customer classifications, we can accurately determine the sales life cycle stage of each customer based on their unique tag combination, such as accurately identifying whether the customer is in the budding stage of interest in the product or in the purchase decision stage after careful consideration.

[0020] It can dynamically update labels based on real-time changes in customers, ensure that follow-up strategies always fit the customer's current situation, avoid communication misalignment due to information lags, and achieve accurate matching with customer needs.

[0021] 2. Intelligent optimization follow-up strategy: By using big data analysis and clustering algorithms, we can mine the effective follow-up time range and contact method types from the massive transaction customer records, so as to build a highly adaptive initial follow-up strategy. This data-driven strategy generation method is more scientific and accurate than the traditional manual setting based on experience, and can greatly improve the effectiveness of follow-up.

[0022] During the follow-up process, we analyze feedback in real time based on customer follow-up records, use language models to determine the effectiveness of follow-up, and dynamically adjust strategies. If we find that the current contact method does not arouse the customer's interest, we will immediately switch to a more appropriate time or method to continuously optimize the communication effect, rather than rigidly executing the established plan.

[0023] 3. Improve follow-up efficiency and effectiveness.

[0024] Intelligent algorithms enable automated and batched customer follow-up, greatly saving manpower and time costs. They can process massive amounts of customer data at the same time, quickly match corresponding strategies for different customers, and execute follow-up, breaking through the quantitative limitations of manual follow-up.

[0025] Through precise strategy implementation, we can improve customer response rate and satisfaction to follow-up, and then improve customer conversion rate and loyalty. For example, we can timely push personalized preferential information to potential customers to attract them to place orders; we can provide exclusive care to loyal customers to promote their continued repurchase and word-of-mouth communication.

[0026] 4. Effectively manage and follow up risks.

[0027] Set up a blacklist mechanism to respect customer wishes, avoid forced follow-up on customers who clearly do not want to be disturbed, prevent customer resentment and negative word of mouth, and maintain the corporate image.

[0028] Establish a risk control system, reasonably set the minimum interval between two follow-ups for a single customer, control the follow-up rhythm, prevent excessively frequent disturbances to customers, ensure customer experience, and safeguard long-term and stable customer relationships.

[0029] 5. Achieve flexible strategy transition.

[0030] Considering the reliance of intelligent follow-up on data volume, when the number of users with a certain type of transaction is insufficient in the early stage, an operational strategy of manual follow-up is adopted to accumulate data; after the data reaches the standard, a small-scale pilot of intelligent follow-up is conducted, and the transaction rates of intelligent follow-up and manual follow-up are compared, and the best one is selected. This gradual transition method can fully utilize human experience and steadily introduce intelligent technology to ensure the smooth development of the business. BRIEF DESCRIPTION OF THE DRAWINGS

[0031] In order to more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings required for use in the embodiments or the prior art descriptions are briefly introduced below. In all the drawings, similar elements or parts are generally identified by similar reference numerals. In the drawings, each element or part is not necessarily drawn according to the actual scale. Obviously, the drawings described below are some embodiments of the present invention. For those of ordinary skill in the art, other drawings can also be obtained based on these drawings without paying creative labor.

[0032] Figure 1 It is a schematic diagram of the process of intelligent follow-up in the present invention; Figure 2 This is a schematic diagram of the interface for constructing a label system in the present invention; Figure 3 This is a schematic diagram of the interface for classifying customers in the present invention.

[0033] Figure 4 This is a schematic diagram of the initial follow-up strategy configuration interface in the present invention.

[0034] Figure 5 Schematic diagram of the intelligent follow-up strategy configuration interface in the present invention. DETAILED DESCRIPTION

[0035] To make the objectives, technical solutions and advantages of the embodiments of the present invention clearer, the following will clearly and completely describe the technical solutions in the embodiments of the present invention with reference to the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are some, but not all, of the embodiments of the present invention. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the protection scope of the present invention.

[0036] In this document, suffixes such as "module", "component" or "unit" used to represent elements are only for the convenience of describing the present invention, and have no specific meaning in themselves. Therefore, "module", "component" or "unit" can be used interchangeably.

[0037] In this document, terms such as "upper", "lower", "inner", "outer", "front", "rear", "one end", "the other end", etc. indicate the orientation or positional relationship based on the orientation or positional relationship shown in the drawings. It is only for the convenience of describing the present invention and simplifying the description, rather than indicating or implying that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation, and therefore cannot be construed as a limitation of the present invention. In addition, the terms "first" and "second" are only used for descriptive purposes and cannot be construed as indicating or implying relative importance.

[0038] In this document, unless otherwise clearly defined and limited, terms such as "installation", "provided with", "connection", etc. shall be understood in a broad sense. For example, "connection" can be a fixed connection, a detachable connection, or an integral connection; it can be a mechanical connection, a direct connection, or an indirect connection through an intermediate medium, and can be the communication inside two elements. For those of ordinary skill in the art, the specific meanings of the above terms in the present invention can be understood according to specific situations.

[0039] In this document, "and / or" includes any and all combinations of one or more of the listed related items.

[0040] In this document, "a plurality of" means two or more, that is, it includes two, three, four, five, etc.

[0041] Embodiment 1: As Figure 1 shown, the present invention provides an intelligent customer follow-up method, and its steps specifically include: S1 Build a tag system, define tag attributes and configure tag weights.

[0042] As Figure 2 shown, tags are used to identify the characteristics of customers, which is convenient for customer management. In the present invention, the defined tag attributes are divided into three categories, including: (1) Basic attribute tags, such as age, gender, registration time, last active time, etc. Such tags are usually static and do not change over time.

[0043] (2) Behavioral characteristic tags, such as average daily visit times, last purchase time, interest preferences, etc. Such tags are dynamic and can change with the user's behavior.

[0044] (3) Recommendation tags, which are used to predict the products, services or content that the user may be interested in based on the user's behavior.

[0045] For example, for a certain customer, it can be parsed from the information registered on the platform that his age group is young and his gender is male; according to the browsing records, it can be seen that he frequently browses electronic products; according to his preferences, products such as graphics cards can be recommended for him. Therefore, tags such as "graphics card" and "young male" can be assigned to this customer for later classification and matching follow-up strategies.

[0046] In this embodiment, each tag has its own unique tag weight. The purpose of setting the tag weight is to screen out relatively important tags for classification among many tags. That is, when classifying customers, the tags are sorted according to the weight; and the tags with a ranking higher than the ranking threshold are selected to participate in the classification. For example, during the process of following up users, 10,000 tags are generated. In order to accurately divide the population, the top 200 tags with the highest weights are extracted for clustering calculation.

[0047] More specifically, the calculation method of the tag weight in the present invention includes using the formula: ω = ω t *ω p to calculate the tag weight; where ω is the tag weight, ω t is the timeliness weight coefficient, and ω p is the business configuration weight; Using the formula: =

[0048] to calculate the timeliness weight coefficient; where ω t is the timeliness weight coefficient, is the initial weight, t is the time, and c is the offset.

[0049] In this embodiment, the business configuration weight is specifically configured by the operator according to the business scenario. The purpose of the expiration weight is to prevent some tags from being effective all the time. For example, for the above-mentioned "graphics card" tag, it may be that the customer has no subsequent behavior after browsing, so it can be considered that the customer has a potential purchase possibility for this type of product. Therefore, over time, the weight of this tag will gradually decrease.

[0050] However, for some labels such as "young men", the weight of such labels will not decrease over time. Therefore, they should be treated differently during calculation.

[0051] More specifically, the method for calculating the label weight in the present invention includes using the formula: ω = ω t * ω p to calculate the label weight; where ω is the label weight, ω t is the timeliness weight coefficient, and ω p is the business configuration weight; When calculating time-sensitive labels, use the formula: =

[0052] to calculate the timeliness weight coefficient; where ω t is the timeliness weight coefficient, is the initial weight, t is the time such as days, and c is the offset. In the initial case, = log(t + c).

[0053] For labels without timeliness, ω t = 1.

[0054] S2 Collect customer behavior and triggered business events, and generate or update labels for the customer.

[0055] Labels are all generated based on customer behavior and triggered business events. The so-called user behavior refers to the behavior of users on the platform, such as logging in, browsing, purchasing, sharing, etc., which reflects the behavior characteristics of users. And business events refer to business-related events automatically generated by the system, such as order generation, payment completion, after-sales application, etc., which are usually related to user transactions and services.

[0056] Specifically, the collection method adopted in the present invention is: (1) Automatically collect based on buried point technology and business logs. After ETL processing, it is stored in the data warehouse. Buried point technology is a technology for automatically collecting user behavior data, usually embedded in applications or web pages. ETL (Extract, Transform, Load) is a three-step method for data processing, namely data extraction, transformation, and loading. It is used to convert raw data into a usable data format and store it in the data warehouse.

[0057] (2) Support the integration of data from multiple subsystems and store them uniformly in the data warehouse.

[0058] The collection of customer behaviors and triggered business events is real-time, while the generation or update of tags needs to be carried out according to a set cycle, such as settling accounts once a day. The production of tags can be manual or automatic, and the update calculation of tags is more efficient in an automatic way.

[0059] S3 classifies customers according to tags; sets a customer sales life cycle for each customer category, and configures an initial follow-up strategy for each stage of the sales life cycle.

[0060] As Figure 3 shown, in order to carry out refined management of customers, corresponding follow-up strategies need to be formulated according to the characteristics of customers. Therefore, in this embodiment, it is first necessary to classify customers according to tags. Classification is the process of dividing a large number of users into different groups according to specific rules or characteristics (tags). In this way, personalized marketing strategies can be formulated and differentiated services can be provided according to the characteristics of different groups, so as to improve the marketing effect and user satisfaction.

[0061] In this embodiment, it supports tag calculation based on condition combinations. That is, it supports calculation based on tag combinations, which means that the system can classify users according to complex business logics. Tag A and tag B can be various tags based on the tag system. For example, condition A may be "young males" (based on basic attribute tags), and condition B may be "browsed graphics card products within the last 5 days" (based on behavioral feature tags). Only users who meet both of these conditions will be classified into a specific group. This flexibility enables operation personnel to accurately define user groups according to different business scenarios and goals.

[0062] More specifically, in this embodiment, spark offline calculation is used for classification. Spark is a fast and general big data processing framework with efficient in-memory computing capabilities and rich data processing operators. It can process large-scale data sets and provides support for multiple programming languages (such as Scala, Java, Python, etc.), and is very suitable for clustering calculations in big data scenarios.

[0063] Performing offline calculations through Spark means that data processing is not real-time but rather batches of data are processed at fixed time intervals. This approach is suitable for scenarios where real-time requirements are not high, but high accuracy and integrity of data processing are required. In the clustering calculation, performing offline calculations every day can ensure that the system collects sufficient user behavior data and label information, enabling comprehensive and accurate conditional combination judgments and label calculations. For example, within a cycle, users may generate various behavior data, such as browsing web pages, purchasing goods, clicking on advertisements, etc. Through offline calculations, this data can be integrated, and all the user's behaviors throughout the cycle can be comprehensively considered to more accurately determine whether the user meets specific conditional combinations and then be classified into the corresponding groups.

[0064] After classifying users, it is also necessary to set the customer sales life cycle for this type of user. The customer sales life cycle refers to the entire process from when a customer first contacts an enterprise to when the business relationship with the enterprise finally ends. It is usually divided into multiple stages, and each stage requires different follow-up strategies for follow-up. For different merchants or products, the stages of the customer sales life cycle will also be different. For example, some customer sales life cycles include five stages: awareness stage, interest stage, consideration stage, purchase stage, and repeat purchase stage. And some customer sales life cycles include four stages: potential customer, first purchase, repeat purchase, and loyal customer. The present invention does not limit the specific stages of the customer sales life cycle.

[0065] After obtaining the customer sales life cycle of a certain type of customer, it is also necessary to match an appropriate initial follow-up strategy for each stage.

[0066] As Figure 4 shown, in this embodiment, the method for formulating the initial follow-up strategy includes: S201 Retrieve the follow-up records of a certain type of closed-won customer, including the contact time, contact method, and chat content.

[0067] Each type of customer requires a set of targeted follow-up strategies. Analyzing past successful cases through big data helps improve the effectiveness of follow-up strategies. In this step, the retrieved follow-up records contain multiple key pieces of information. The contact time records the specific time point when communicating with the customer, which is crucial for analyzing at which time period of the day, week, or month the customer is more receptive to information; the contact method clarifies whether contact with the customer is made through phone calls, text messages, emails, or face-to-face communication, etc.; the chat content details the specific information of the communication between the two parties and is the core basis for analyzing customer needs, preferences, and communication effects.

[0068] S202 Use a language model to analyze the chat content and mark effective follow-ups.

[0069] Once a follow-up is determined to be an effective one, it is marked. This marking provides a clear criterion for subsequent clustering analysis, enabling subsequent analysis to focus on the follow-up records that truly contribute to the conclusion of a deal, and thus more accurately extract effective follow-up patterns. For example, mark the follow-up records in which the customer clearly states in the chat that "this product meets my needs well and I am considering placing an order" for further analysis of the reach time and reach method related to it.

[0070] In this step, the language model such as ChatGPT, Doubao, DeepSeek, etc. can be connected to the system to analyze the chat content to determine whether it is an effective follow-up.

[0071] S203 uses a clustering algorithm to cluster the effective follow-up marks in the follow-up records of customers who have made a deal, and obtain the reach time range and reach method types of effective follow-up.

[0072] The clustering algorithm is an unsupervised learning method that divides the data points in the dataset into different clusters according to similarity. In this process, the algorithm will group the effective follow-up records with similar characteristics according to the distance or similarity measure between data points. For example, for the reach time, the algorithm may find that some effective follow-ups are concentrated between 10 - 12 am, while others are concentrated between 3 - 5 pm, thus forming different time clusters; for the reach method, phone calls, WeChat, etc. may be clustered separately.

[0073] Through clustering analysis, different effective follow-up patterns can be obtained, that is, the reach time range and reach method types of effective follow-up. These results reveal which time periods and reach method combinations are more effective in facilitating transactions. For example, the clustering results may show that for a certain type of customer, following up by phone between 10 - 12 am on weekdays and sending a detailed product introduction by email on Friday afternoon are two common and effective patterns.

[0074] S204 constructs an initial follow-up strategy based on the reach time range and reach method types of effective follow-up.

[0075] Based on the results obtained from the previous clustering analysis, the reach time range and reach method types of effective follow-up are used as the core basis for constructing the initial follow-up strategy. For example, if it is found that there are more effective follow-ups with a certain type of customer through WeChat communication between 7 - 9 pm on weekdays and through phone communication between 10 - 12 am on Saturday, then when formulating the initial follow-up strategy, priority can be given to communicating with customers at these time periods through the corresponding reach methods.

[0076] In addition to the reach time and reach method, other relevant information can also be combined, such as the key issues that customers are concerned about in the chat content, to formulate corresponding communication scripts and content templates, thereby forming a complete initial follow-up strategy. Such a strategy is more targeted and effective, and can increase the probability of closing deals in the subsequent customer follow-up process.

[0077] Generally speaking, the follow-up strategy includes the reach time, reach method, and script template. Table 1 lists the initial follow-up strategies for each stage.

[0078] Table 1 Initial Follow-up Strategy

[0079] Of course, it can be foreseen that the initial follow-up strategy can also be manually configured by the operation staff.

[0080] S4 conducts intelligent follow-up on customers.

[0081] The initial follow-up strategy is a general follow-up strategy for a certain type of customers and is not suitable for all customers in this category. The intelligent follow-up in this step is actually a personalized adjustment for individual customers.

[0082] Such as Figure 5 shown, the specific steps of intelligent follow-up include: S401 determines the customer life cycle stage that this type of customer is in; and follows up according to the corresponding initial follow-up strategy matched with the customer life cycle stage.

[0083] Since different follow-up strategies are configured for each stage of the customer life cycle, it is necessary to first obtain the stage that the customer is in. In this embodiment, there are two methods to obtain the customer stage: One is when there is a landmark event in a certain stage of the customer sales life cycle, to judge whether the customer is in this sales life cycle stage through the customer's tags.

[0084] In each stage of the customer sales life cycle, certain specific events can clearly identify that the customer has entered this stage, and these events are called landmark events. For example, in the "first purchase" stage, the customer's completion of the first product or service purchase behavior is the landmark event.

[0085] The label system records various characteristics and behavioral information of customers. By analyzing the labels a customer has, it is possible to determine whether the customer has experienced a landmark event in a certain stage, and thus determine the life cycle stage the customer is in. For example, if the customer label contains "purchase times = 1", then it can be determined that the customer is in the "first purchase" stage. Another example is that after the customer's first purchase, the label "purchase times = 0" will be updated to "purchase times = 1". When making stage judgments, the system will detect changes in the labels and thus include the customer in the first purchase stage. This method relies on the accurate marking and labeling of customer behaviors and events in the early stage and can relatively intuitively determine the customer stage.

[0086] Second, in the case where there is no landmark event in a certain stage of the customer sales life cycle, by clustering the follow-up records of closed-won customers, obtain the time period of this stage; and judge whether the customer is in this stage according to whether the customer's current time period coincides with the said time period.

[0087] When there is no obvious landmark event in a certain stage of the customer sales life cycle, it is necessary to conduct cluster analysis on the follow-up records of closed-won customers to determine the time period of this stage. In this scenario, the follow-up records of closed-won customers at different times are used as data points. Based on various characteristics in the follow-up records (such as follow-up time, customer feedback, interaction frequency, etc.), these data points are divided into different clusters using a clustering algorithm (such as the K-means algorithm). Each cluster represents a group of follow-up records that are similar in certain aspects.

[0088] Through cluster analysis, find the clusters formed by the follow-up records related to a specific stage and determine the time period range corresponding to these clusters. For example, through cluster analysis, it is found that within a certain period of time after the customer establishes contact with the enterprise (such as the 3rd - 7th day), the follow-up records of the customer show a specific pattern (such as frequently consulting product details, comparing different products, etc.), and this pattern is considered to correspond to the "consideration stage", then the 3rd - 7th day can be used as the time period of the "consideration stage".

[0089] After determining the stage the customer is in, just match the follow-up strategy corresponding to this stage for it.

[0090] S402 Establish customer follow-up records during the follow-up process, and adjust the initial follow-up strategy according to the effective follow-up in the customer follow-up records and then conduct the follow-up.

[0091] As mentioned above, the initial follow-up strategy is only a general strategy. In order to serve customers more precisely, it is also necessary to further adjust the follow-up strategy according to the customer's own habits, etc. Specifically, the method of adjusting the initial follow-up strategy specifically includes the following steps: S4021 Retrieve the follow-up records of a certain customer, including the contact time, contact method, and chat content.

[0092] The follow-up records are detailed accounts of each interaction with the customer and are important bases for subsequent strategy adjustments. By analyzing these records, we can deeply understand the customer's reactions and needs, and then optimize the initial general follow-up strategy to better meet the personalized needs of each customer.

[0093] Among them, the contact time records the specific time points of contacting the customer, which helps to analyze which time period of the day, week, or even month the customer is more sensitive or receptive to communication. For example, if it is found that the customer often replies to messages on weekday evenings, then subsequent follow-ups can be considered during this time period.

[0094] The contact method clarifies the channel through which contact is made with the customer, such as text messages, emails, phone calls, WeChat, etc. Different customers may prefer different contact methods. By analyzing this information, we can find the most suitable communication channel for this customer.

[0095] The chat content details the specific information communicated with the customer and is the core basis for judging the effectiveness of the follow-up. By analyzing the chat content, we can understand the customer's interests, concerns, questions, and attitudes towards products or services.

[0096] S4022 Use a language model to analyze the chat content and judge whether the follow-up in the follow-up record is effective.

[0097] With the help of language models such as ChatGPT, Doubao, DeepSeek and other natural language processing tools, deeply analyze the extracted chat content. By analyzing the chat content, judge whether this follow-up has played a positive role in promoting the development of the relationship between the customer and the enterprise, that is, judge whether the follow-up is effective. For example, if the customer expresses strong interest in the product, puts forward specific purchase intentions during the chat, or expresses satisfaction after the previous questions are answered, these can all be regarded as effective follow-ups. On the contrary, if the customer shows coldness, disinterest, or the doubts about the product are not properly resolved, it may mean that the follow-up is ineffective.

[0098] S4023 When the follow-up is effective, continue to use the initial follow-up strategy.

[0099] When it is determined through analysis that a certain follow-up is effective, it means that the current initial follow-up strategy has played a positive role in the interaction with this customer and conforms to the customer's needs and communication habits. Therefore, in subsequent follow-up processes, continue to use this strategy, maintain an effective communication mode with the customer, further consolidate the relationship with the customer, and promote the customer's progress in the sales life cycle.

[0100] When the follow-up of S4024 is ineffective, switch to other reach times and reach methods.

[0101] If it is determined that the follow-up is ineffective, the current follow-up strategy needs to be adjusted. The first thing to consider is to switch the reach time and reach method. This is based on the assumption that different customers have different acceptance levels of different communication channels at different times.

[0102] For example, if the customer was contacted by phone in the morning before but the customer's response was cold, then one can try to communicate with the customer again by text message or WeChat in the afternoon or evening, providing a more convenient and comfortable communication environment for the customer, so as to increase the customer's attention and response rate to the follow-up information, thereby making the subsequent follow-up more effective. This adjustment method aims to re-attract the customer's attention and improve the interaction effect with the customer by changing the communication time and channel.

[0103] Since intelligent follow-up is a customer follow-up method based on data analysis and algorithms, its effectiveness and accuracy largely depend on a large amount of historical data. These historical data cover various behavioral information, communication records, and transaction situations of customers, etc. Through the analysis and learning of these data, the intelligent follow-up system can dig out the customer's behavioral patterns, preferences, and the best follow-up strategies, etc., so as to achieve accurate customer follow-up. Therefore, it is necessary to accumulate a certain amount of data through manual follow-up by operation personnel before conducting intelligent follow-up. The specific process is as follows: (1) When the number of customers in a certain category who have made a deal is less than the quantity threshold, adopt an operation strategy for manual follow-up.

[0104] When the number of customers in a certain category who have made a deal is less than the set quantity threshold, due to insufficient data volume, there is not enough information support for intelligent follow-up. At this time, adopt an operation strategy for manual follow-up. The operation strategy manual follow-up is to execute the follow-up task based on the experience of operation personnel and pre-set rules. For example, formulate specific communication scripts for different types of customers, select appropriate reach methods and times, etc. By this means, data is gradually accumulated during the interaction with customers, including information such as customer feedback and communication effects, laying a foundation for subsequent intelligent follow-up.

[0105] (2) When the number of customers in a certain category who have made a deal is greater than or equal to the quantity threshold, adopt intelligent follow-up for some customers in this category.

[0106] When the number of users in a certain category who have completed a transaction is greater than or equal to the quantity threshold, it indicates that a certain scale of data has been accumulated, meeting the conditions for conducting intelligent follow-up. At this time, intelligent follow-up is adopted for some users in this category. On the one hand, this is to test the actual effect of the intelligent follow-up system in this customer group, and on the other hand, it is also to further optimize the intelligent follow-up strategy in actual applications. Through the practice of intelligent follow-up for these users, relevant data is collected, such as the success rate of follow-up, customer feedback, etc., and compared with the effect of manual follow-up.

[0107] (3) In the case where the closing rate of intelligent follow-up is higher than that of manual follow-up, switch the customers in this category of manual follow-up to intelligent follow-up.

[0108] Compare the closing rate of intelligent follow-up with that of manual follow-up. The closing rate is a key indicator for measuring the follow-up effect, which directly reflects the effectiveness of the follow-up method in facilitating transactions. If intelligent follow-up shows a higher closing rate in actual applications, it indicates that the intelligent follow-up strategy has more advantages in this customer group, can better meet customer needs, and promote the conclusion of transactions.

[0109] In addition, in order not to cause excessive harassment to customers, the present invention is provided with a risk prevention mechanism. In the process of customer follow-up, it is crucial to avoid causing excessive harassment to customers. Excessive harassment may lead to customer resentment, thereby damaging the customer relationship, reducing the customer's favorability and loyalty towards the enterprise, and even may result in customer churn. Therefore, establishing an effective risk prevention mechanism is a necessary measure to ensure the smooth progress of customer follow-up work and maintain a good customer relationship at the same time. This mechanism specifically includes: First, set up a blacklist and add customers who do not want to be harassed to the blacklist; read the blacklist before follow-up, and no longer follow up customers in the blacklist.

[0110] Add customers who clearly indicate that they do not want to be harassed to the blacklist, accurately identify and exclude those customers who do not wish to receive follow-up information. In daily customer follow-up work, when a customer conveys the willingness not to be harassed through various means (such as directly feedback to the customer service, clearly indicating in the communication, etc.), the relevant staff will enter the customer's information into the blacklist. Before each customer follow-up, the system will automatically read the blacklist data. Ensure that in the subsequent follow-up process, customers in the blacklist will no longer receive any form of follow-up information, avoid causing unnecessary harassment to these customers, fully respect the personal wishes of customers, protect the privacy and experience of customers, contribute to maintaining a good relationship between the enterprise and customers, and avoid negative emotions and behaviors of customers caused by forced follow-up.

[0111] Secondly, then establish risk control and set the minimum interval time between two follow-ups for a single customer.

[0112] By setting the minimum interval time between two follow - ups for a single customer, the follow - up frequency is restricted to avoid disturbing the customer too frequently. The enterprise determines a reasonable minimum interval time based on factors such as its own business characteristics, customer group characteristics, and industry experience. For example, for some products or services that focus on customer experience and have a long customer decision - making cycle, the minimum interval time may be set to one week; while for some high - frequency consumer products, the interval time may be relatively short, such as 1 - 2 days. After each customer follow - up is completed, the system records the follow - up time and automatically checks whether the time since the last follow - up meets the minimum interval time requirement when planning the next follow - up. Only when the time interval meets the set conditions is the next follow - up of the customer allowed.

[0113] Embodiment 2: The present invention also provides an intelligent customer follow - up system, including: A label construction module, which is used to construct a label system, define label attributes and configure label weights; and collect customer behaviors and triggered business events to generate or update labels for customers. A customer classification module, which is used to classify customers according to labels; set a customer sales life cycle for each type of customer, and configure an initial follow - up strategy for each stage of the sales life cycle. An intelligent follow - up module, which is used to perform intelligent follow - up on customers, including: Judging the stage of the customer life cycle in which this type of customer is located; and matching the corresponding initial follow - up strategy according to the stage of the customer life cycle in which it is located for follow - up. Establishing a customer follow - up record during the follow - up process, and adjusting the initial follow - up strategy according to the effective follow - up in the customer follow - up record for follow - up.

[0114] It should be noted that in this article, the term "including", "comprising" or any other variant thereof is intended to cover non - exclusive inclusion, so that a process, method, article or device including a series of elements not only includes those elements, but also includes other elements not explicitly listed, or further includes elements inherent to this process, method, article or device. Without more limitations, an element defined by the statement "including one..." does not exclude the existence of another identical element in the process, method, article or device including this element.

[0115] Through the description of the above embodiments, those skilled in the art can clearly understand that the above-described example methods can be implemented by means of software plus a necessary general hardware platform. Of course, they can also be implemented by hardware, but in many cases the former is a better implementation. Based on such an understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art can be embodied in the form of a software product. This computer software product is stored in a storage medium (such as ROM / RAM, magnetic disk, optical disk) and includes several instructions for causing a computer terminal (which can be a mobile phone, computer, server, or network device, etc.) to execute the methods described in various embodiments of the present invention.

[0116] The embodiments of the present invention have been described above in conjunction with the accompanying drawings. However, the present invention is not limited to the above specific embodiments. The above specific embodiments are merely illustrative and not restrictive. Under the inspiration of the present invention, those of ordinary skill in the art can also make many forms without departing from the purpose of the present invention and the scope protected by the claims. All of these fall within the protection scope of the present invention.

Claims

1. An intelligent customer follow-up method, characterized in that include: Build a tag system, define tag attributes and configure tag weights; Collect customer behaviors and triggered business events, and generate or update tags for customers; Categorize customers based on tags; set the customer sales lifecycle for each type of customer, and configure the initial follow-up strategy for each stage of the sales lifecycle; Intelligent follow-up of customers, including: Determine the customer life cycle stage of this type of customer; and match the corresponding initial follow-up strategy according to the customer life cycle stage; During the follow-up process, a customer follow-up record is established, and follow-up is carried out after adjusting the initial follow-up strategy based on the effective follow-up in the customer follow-up record.

2. The intelligent customer follow-up method according to claim 1, characterized in that: The tag attributes include basic attribute tags, behavior feature tags, and recommendation tags.

3. The intelligent customer follow-up method according to claim 1, characterized in that: When classifying customers, sort the tags according to their weights; and select tags with rankings higher than the ranking threshold to participate in the classification.

4. The intelligent customer follow-up method according to claim 3, characterized in that The label weight calculation method includes using the formula: ω=ω t *oh p Calculate the label weight; where ω is the label weight, ω t is the time-effectiveness weight coefficient, ω p Configure weights for services; For time-limited labels, use the formula: = Calculate the time-effectiveness weight coefficient; where ω t is the time weight coefficient, is the initial weight, t is the time, and c is the offset; For labels without expiration, ω t =1.

5. The intelligent customer follow-up method according to claim 1, characterized in that The method for formulating the initial follow-up strategy includes: Retrieve the follow-up records of a certain type of successful customers, including contact time, contact method, and chat content; Use language models to analyze chat content and mark effective follow-ups; Use clustering algorithms to cluster the effective follow-up marks in the follow-up records of closed customers to obtain the effective follow-up contact time range and contact method types; Build an initial follow-up strategy based on the effective follow-up time range and types of contact methods.

6. The intelligent customer follow-up method according to claim 1, characterized in that Ways to adjust your initial follow-up strategy based on effective follow-up in your customer follow-up records include: Retrieve a customer's follow-up records, including contact time, contact method, and chat content; Use the language model to analyze the chat content and determine whether the follow-up in the follow-up record is effective; If follow-up is effective, continue with the initial follow-up strategy; If follow-up is ineffective, switch to other times and methods of contact.

7. The intelligent customer follow-up method according to claim 1, characterized in that: Set up a blacklist and add customers who do not want to be disturbed to the blacklist; read the blacklist before following up, and do not follow up with customers on the blacklist; Establish risk control and set a minimum interval between two follow-ups for a single customer.

8. The intelligent customer follow-up method according to claim 1, characterized in that Methods for determining the customer life cycle stage a customer is in include: When there is a landmark event at a certain stage in the customer's sales lifecycle, determine whether the customer is at that stage of the sales lifecycle based on the customer's label; When there is no landmark event at a certain stage in the customer sales lifecycle, the time period of the stage is obtained by clustering the follow-up records of successful customers; and whether the customer is in this stage is determined by whether the customer's current time period coincides with the time period.

9. The intelligent customer follow-up method according to claim 1, characterized in that: When the number of users who have completed transactions of a certain type is less than the threshold, an operational strategy is used for manual follow-up; When the number of users of a certain type who have completed transactions is greater than or equal to the threshold, intelligent follow-up will be adopted for some of these users; When the transaction rate of smart follow-up is higher than that of manual follow-up, such customers who are manually followed up will be switched to smart follow-up.

10. An intelligent customer follow-up system, characterized in that include: The tag building module is used to build a tag system, define tag attributes and configure tag weights; It also collects customer behaviors and triggered business events, and generates or updates tags for customers; The customer classification module is used to classify customers according to tags; set the customer sales life cycle for each type of customer, and configure the initial follow-up strategy for each stage of the sales life cycle; Intelligent follow-up module, used for intelligent follow-up of customers, including: Determine the customer life cycle stage of this type of customer; and match the corresponding initial follow-up strategy according to the customer life cycle stage; During the follow-up process, a customer follow-up record is established, and follow-up is carried out after adjusting the initial follow-up strategy based on the effective follow-up in the customer follow-up record.

Images