Customer relationship management system, method and medium based on cloud computing evaluation and diagnosis

Abstract

The present application provides a customer relationship management system, method, and medium based on cloud computing evaluation and diagnosis, which relates to the technical field of customer relationship management, including: a stage segmentation module, which connects to a customer management platform and performs customer conversion stage segmentation; a stage matching module, which determines the relationship stage to which the first customer belongs; a stage change prediction module, which collects a first interactive data set and performs stage feature diagnosis and analysis in a cloud center; a missing factor analysis module, which performs missing factor analysis when the first prediction probability is less than a threshold; and an auxiliary reminder module, which performs screening in a shared knowledge database and recommends it to the first service personnel for customer relationship management auxiliary reminders. This application can solve the technical problem in the prior art that it is difficult to dynamically identify changes in customer relationships and cannot provide targeted assistance in maintaining customers, resulting in a low conversion rate. Through stage identification and conversion probability prediction, dynamic management of customer relationships and intelligent intervention are achieved, thereby improving customer conversion rates.

Description

Technical Field

[0001] The present application relates to the technical field of customer relationship management, and in particular to a customer relationship management system, method, and medium based on cloud computing evaluation and diagnosis. Background Art

[0002] Existing customer relationship management systems primarily rely on static customer data and pre-set rules to maintain customer relationships, lacking the ability to dynamically identify changes in customer relationships based on real-time interaction information. This lack of awareness of the real-time state of customer relationships and identification of potential issues makes it difficult to promptly identify key factors influencing conversions and take targeted measures as customers transition from one relationship stage to another. This, in turn, limits conversion rate improvements and, to a certain extent, restricts the effectiveness of customer relationship management systems in improving customer conversion efficiency and optimizing customer lifecycle management.

[0003] In summary, the existing technology has a technical problem of low conversion rate due to the difficulty in dynamically identifying changes in customer relationships and the inability to provide targeted assistance in customer maintenance. Summary of the Invention

[0004] The purpose of this application is to provide a customer relationship management system, method and medium based on cloud computing evaluation and diagnosis, so as to solve the technical problems in the prior art that it is difficult to dynamically identify changes in customer relationships and cannot provide targeted assistance in maintaining customers, resulting in low conversion rates.

[0005] In order to achieve the above objectives, the present application provides a customer relationship management system, method, and medium based on cloud computing evaluation and diagnosis.

[0006] In a first aspect, the present application provides a customer relationship management system based on cloud computing evaluation and diagnosis, wherein the customer relationship management system based on cloud computing evaluation and diagnosis includes: a stage segmentation module, used to connect to a customer management platform, collect historical customer conversion data sets to perform customer conversion stage segmentation, and construct a customer relationship stage sequence; a stage matching module, used to read the first conversion feature data of the first customer currently in progress, and match the real-time relationship stage to which it belongs in the customer relationship stage sequence; a stage change prediction module, used to collect the first interaction data set between the first customer and the corresponding first service personnel in the real-time relationship stage, perform stage feature diagnosis analysis in the cloud center, and predict the first predicted probability of transitioning from the real-time relationship stage to the next relationship stage; a missing factor analysis module, used to perform interaction missing factor analysis based on the first interaction data set to generate a first interaction missing factor when the first predicted probability is less than a preset probability threshold; an auxiliary reminder module, used to use the first interaction missing factor to filter data in the shared knowledge database of the customer management platform, and recommend the first filtered data and the first interaction missing factor to the first service personnel for customer relationship management auxiliary reminder.

[0007] Optionally, a behavior analysis unit is used to perform customer continuous behavior analysis based on the historical customer conversion data set to construct a number of customer continuous behavior sequences; a similar merging unit is used to perform similarity merging of continuous behaviors on the several customer continuous behavior sequences to generate a number of merged behavior sequences; and a stage decomposition unit is used to perform customer relationship stage decomposition based on the several merged behavior sequences to generate the customer relationship stage sequence.

[0008] Optionally, the first sequence extraction subunit is used to extract the first customer continuous behavior sequence from the several customer continuous behavior sequences; the first merging subunit is used to merge the continuous behavior features in the first customer continuous behavior sequence according to behavior similarity, generate a first merged behavior sequence, and add the several merged behavior sequences.

[0009] Optionally, a first interaction data extraction unit is used to extract the first historical interaction data set of the first relationship stage in the customer relationship stage sequence based on the historical customer conversion data set, wherein each data in the first historical interaction data set carries a binary classification label indicating whether it has entered the second relationship stage; a first prediction channel establishment unit is used to train the Logistic regression model with the first historical interaction data set, establish a first probability prediction channel, and add multiple probability prediction channels; a probability prediction channel matching unit is used to embed the multiple probability prediction channels into the cloud center, match the corresponding probability prediction channels in the multiple probability prediction channels based on the real-time relationship stage and the next relationship stage in the cloud center, analyze the first interaction data set, and output the first prediction probability.

[0010] Optionally, the first relationship stage and the second relationship stage are two adjacent relationship stages, and the first relationship stage is located before the second relationship stage.

[0011] Optionally, an evaluation index construction subunit is used to extract the last relationship stage in the customer relationship stage sequence and construct an adaptation evaluation index for the last stage; a fitness evaluation subunit is used to perform stage interaction fitness evaluation with the last stage adaptation evaluation index if the real-time relationship stage is the last relationship stage, and generate an interaction fitness evaluation result; a prediction probability determination subunit is used to perform normalized mapping based on the interaction fitness evaluation result to generate the first prediction probability.

[0012] Optionally, the data collection unit is used to collect a conversion success data set for any customer relationship stage in the customer relationship stage sequence; the central tendency analysis unit is used to perform central tendency analysis of interaction features on the conversion success data set to generate a calibrated interaction concentration feature for any customer relationship stage; and the missing feature comparison unit is used to extract the calibrated interaction concentration feature corresponding to the real-time relationship stage, perform missing feature comparison on the first interaction data set, and generate the first interaction missing factor.

[0013] Optionally, a database construction unit is used to collect multiple customer interaction reference data that meet predetermined data sharing constraints to construct the shared knowledge database, wherein the multiple customer interaction reference data carry target interaction tags; a data matching and screening unit is used to input the first interaction missing factor into the shared knowledge database, extract customer interaction reference data whose target interaction tags match the first interaction missing factor, and generate the first screening data.

[0014] In a second aspect, the present application also provides a customer relationship management method based on cloud computing evaluation and diagnosis, wherein the customer relationship management method based on cloud computing evaluation and diagnosis includes: connecting to a customer management platform, collecting historical customer conversion data sets to perform customer conversion stage segmentation, and constructing a customer relationship stage sequence; reading the first conversion feature data of the first customer currently in progress, and matching the real-time relationship stage to which it belongs in the customer relationship stage sequence; collecting the first interaction data set between the first customer and the corresponding first service personnel in the real-time relationship stage, performing stage feature diagnosis analysis in the cloud center, and predicting the first prediction probability of transitioning from the real-time relationship stage to the next relationship stage; when the first prediction probability is less than a preset probability threshold, performing interaction missing factor analysis based on the first interaction data set to generate a first interaction missing factor; using the first interaction missing factor to perform data screening in the shared knowledge database of the customer management platform, and recommending the first screening data and the first interaction missing factor to the first service personnel for customer relationship management auxiliary reminders.

[0015] In a third aspect, a computer-readable storage medium stores a computer program, which, when executed, implements the steps of the customer relationship management method based on cloud computing evaluation and diagnosis as described in any one of the first aspects above.

[0016] One or more technical solutions provided in this application have at least the following technical effects or advantages:

[0017] The stage segmentation module connects to the customer management platform, collects historical customer conversion datasets, performs customer conversion stage segmentation, and constructs a customer relationship stage sequence. The stage matching module reads the first conversion feature data of the current first customer and matches it to the real-time relationship stage in the customer relationship stage sequence. The stage transition prediction module collects the first interaction dataset between the first customer and the corresponding first service personnel in the real-time relationship stage, performs stage feature diagnosis and analysis in the cloud center, and predicts a first predicted probability of transitioning from the real-time relationship stage to the next relationship stage. The missing factor analysis module performs interaction missing factor analysis based on the first interaction dataset to generate a first interaction missing factor when the first predicted probability is less than a preset probability threshold. The auxiliary reminder module uses the first interaction missing factor to filter data in the shared knowledge database of the customer management platform and recommends the first filtered data and the first interaction missing factor to the first service personnel for customer relationship management auxiliary reminders. In other words, by reading the current customer's conversion features, matching their relationship stage, collecting interaction data between the customer and the service personnel, analyzing, and predicting whether the customer will enter the next stage. If the predicted probability is lower than the preset threshold, the interaction missing factors are analyzed and relevant suggestions are generated. Relevant data is filtered through the shared knowledge database, and the missing factors and screening results are recommended to service personnel to provide auxiliary reminders for customer relationship management, better respond to changes in customer status, and thus improve customer conversion rate and customer satisfaction.

[0018] The above description is only an overview of the technical solution of the present application. In order to more clearly understand the technical means of the present application, which can be implemented in accordance with the contents of the description, and to make the above and other purposes, features and advantages of the present application more obvious and easy to understand, the specific implementation methods of the present application are specifically listed below. It should be understood that the content described in this section is not intended to identify the key or important features of the embodiments of the present application, nor is it intended to limit the scope of the present application. Other features of the present application will become easy to understand through the following description. BRIEF DESCRIPTION OF THE DRAWINGS

[0019] In order to more clearly illustrate the technical solutions in this application or the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are merely exemplary, and a person of ordinary skill in the art can obtain other drawings based on the provided drawings without creative work.

[0020] Figure 1 This is a schematic diagram of the structure of the customer relationship management system based on cloud computing evaluation and diagnosis for this application;

[0021] Figure 2This is a flowchart of the customer relationship management method based on cloud computing evaluation and diagnosis for this application.

[0022] Explanation of the accompanying symbols: stage segmentation module 11, stage matching module 12, stage change prediction module 13, missing factor analysis module 14, auxiliary reminder module 15. DETAILED DESCRIPTION

[0023] This application solves the technical problem in the prior art of low conversion rate due to the difficulty in dynamically identifying changes in customer relationships and the inability to provide targeted assistance in maintaining customers by providing a customer relationship management system, method, and medium based on cloud computing evaluation and diagnosis. By reading the conversion characteristics of the current customer and matching the relationship stage they are in, the interaction data between the customer and the service personnel is collected, analyzed, and predicted whether the customer will enter the next stage. If the predicted probability is lower than the preset threshold, the interaction missing factor is analyzed and relevant suggestions are generated. The relevant data is filtered through the shared knowledge database, and the missing factors and screening results are recommended to the service personnel to provide auxiliary reminders for customer relationship management, better respond to changes in customer status, and thus improve customer conversion rate and customer satisfaction.

[0024] Below, the technical solutions in this application will be clearly and completely described with reference to the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of this application, rather than all the embodiments of this application. It should be understood that this application is not limited to the example embodiments described herein. Based on the embodiments of this application, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of this application. It should also be noted that, for the convenience of description, only the parts related to this application, rather than all of them, are shown in the accompanying drawings.

[0025] For example 1, please refer to the attached Figure 1 The present application provides a customer relationship management system based on cloud computing evaluation and diagnosis, wherein the customer relationship management system based on cloud computing evaluation and diagnosis specifically includes the following steps:

[0026] The stage segmentation module 11 is used to connect to the customer management platform, collect historical customer conversion data sets, perform customer conversion stage segmentation, and construct a customer relationship stage sequence.

[0027] Furthermore, the stage segmentation module 11 in the customer relationship management system / device based on cloud computing evaluation and diagnosis is also used to:

[0028] A behavior analysis unit is used to perform customer continuous behavior analysis based on the historical customer conversion data set to construct a number of customer continuous behavior sequences; a similarity merging unit is used to perform similarity merging of continuous behaviors on the several customer continuous behavior sequences to generate a number of merged behavior sequences; a stage decomposition unit is used to perform customer relationship stage decomposition based on the several merged behavior sequences to generate the customer relationship stage sequence.

[0029] The first sequence extraction subunit is used to extract the first customer continuous behavior sequence from the multiple customer continuous behavior sequences; the first merging subunit is used to merge the continuous behavior features in the first customer continuous behavior sequence according to behavior similarity to generate a first merged behavior sequence, and add the multiple merged behavior sequences.

[0030] Specifically, a customer management platform is used to store, manage, and analyze customer data, typically including customer information, transaction history, and interaction records. From existing customer management platforms, we can obtain historical customer conversion datasets, which encompass the complete customer journey from initial contact to final transaction or failure, including timelines, behavioral records, communication content, and purchase history.

[0031] Based on historical customer conversion datasets, we conduct continuous customer behavior analysis, analyzing a series of continuous customer behaviors during the conversion process to identify customer behavior patterns and predict future customer behavior. Several continuous customer behavior sequences include all customer behavior data from first contact to completed purchase. Each behavior sequence includes a series of actions related to customer conversion, such as browsing, clicking, consulting, and purchasing.

[0032] A random sequence of customer behavior sequences is extracted from a set of customer data sets to form the first customer behavior sequence. This sequence represents all the continuous actions taken by that customer during the conversion process. For each of these sequences, similarities between customer behaviors are calculated to merge them. For example, a customer may browse the same product multiple times, and each browse can be considered an independent action, but they essentially represent the same action.

[0033] Behavioral similarity calculation is used to merge these similar behaviors to generate a simplified first merged behavior sequence. For example, the behavior of browsing the same product is merged into one action of browsing the product, which reduces redundant steps in the behavior sequence and highlights the core behavior of customer conversion. Continuous behavior characteristics are specific behavioral patterns or characteristics exhibited by customers in a continuous behavior sequence. Based on these behavioral characteristics, cosine similarity is used to measure the similarity between two behaviors. If the similarity value of two behaviors is higher than the set threshold, it means that they are similar and should be merged. Specifically, each behavior in the customer's continuous behavior sequence is converted into a behavior vector, and the cosine similarity formula is used to calculate the similarity between two users. Based on the calculated cosine similarity, a threshold β (for example, 0.8) is set, and user behaviors with similarity greater than β are grouped together.

[0034] Once the similarity calculation is complete, similar behaviors are merged. For example, if a customer browses the same product multiple times or adds the same item to the shopping cart multiple times during a shopping session, these behaviors are merged into a simplified behavior, removing duplicate behaviors. This makes the customer behavior sequence more streamlined and effective, highlighting the customer's main action trajectory.

[0035] To eliminate redundancy and improve analysis efficiency, we perform similar merging on several continuous customer behavior sequences, generating several merged behavior sequences. These sequences represent streamlined versions of customer behavior paths after similar merging. Based on these streamlined merged behavior sequences, we decompose the customer relationship into stages based on behavioral characteristics and conversion nodes (such as first consultation, clarification of intent, payment completion, and first repeat purchase). Each stage represents a key milestone in the development of the customer relationship, resulting in a customer relationship stage sequence, including the potential customer stage, the intended customer stage, the conversion stage, the treatment implementation stage, and the customer retention and loyalty stage.

[0036] Using pre-defined rules and algorithms, each combined behavior sequence is automatically broken down into distinct customer relationship stages. This is based on behavior type (e.g., browsing, adding to cart, purchasing, reviewing), behavior chronology (e.g., browsing may be very frequent during the prospecting phase, while customer feedback is the primary focus during the implementation phase), and specific behavioral markers (e.g., a customer's first purchase marks the transition from the prospecting phase to the conversion phase). For example, if a customer's behavior sequence includes a significant amount of browsing and inquiries but no purchases, they are likely in the prospecting phase. If a customer completes a purchase, they enter the conversion phase.

[0037] By decomposing the customer relationship stages of several merged behavior sequences, the customer's current relationship stage can be accurately identified, allowing service personnel to provide personalized services based on the customer's different stages, thereby improving conversion rates and customer satisfaction.

[0038] The stage matching module 12 is configured to read the first conversion feature data of the first customer currently in progress and match the corresponding real-time relationship stage in the customer relationship stage sequence.

[0039] Specifically, based on the first customer, the first conversion feature data of a specific customer currently being tracked is obtained. This data, which is relevant to the first customer and represents their current conversion status, can include customer behavior data (e.g., browsing, adding to cart, consulting, ordering), customer feedback, purchase frequency, page dwell time, etc. Within the customer relationship stage sequence, the real-time relationship stage is matched based on the first conversion feature data. A standard customer relationship stage sequence has been pre-established based on historical customer behavior analysis, with each stage corresponding to a specific behavioral feature pattern. After matching, the first customer is labeled to indicate their current customer relationship stage. By reading the first conversion feature data of the current first customer and matching it to the real-time relationship stage within the customer relationship stage sequence, the customer's current relationship status can be dynamically identified. Based on the customer's real-time status, more precise marketing strategies can be delivered, timely intervention in the conversion process, and effectively reducing customer churn.

[0040] The stage transition prediction module 13 is used to collect the first interaction data set between the first customer and the corresponding first service personnel in the real-time relationship stage, perform stage feature diagnosis and analysis in the cloud center, and predict the first prediction probability of transitioning from the real-time relationship stage to the next relationship stage.

[0041] Furthermore, the stage change prediction module 13 in the customer relationship management system / device based on cloud computing evaluation and diagnosis is also used to:

[0042] A first interaction data extraction unit is configured to extract a first historical interaction data set of the first relationship stage in the customer relationship stage sequence based on the historical customer conversion data set, wherein each data item in the first historical interaction data set carries a binary classification label indicating whether the customer has entered the second relationship stage; a first prediction channel establishment unit is configured to train a logistic regression model using the first historical interaction data set, establish a first probability prediction channel, and add multiple probability prediction channels; a probability prediction channel matching unit is configured to embed the multiple probability prediction channels into a cloud center, match corresponding probability prediction channels in the multiple probability prediction channels based on the real-time relationship stage and the next relationship stage in the cloud center, analyze the first interaction data set, and output the first prediction probability.

[0043] The first relationship stage and the second relationship stage are two adjacent relationship stages, and the first relationship stage is located before the second relationship stage.

[0044] Specifically, based on the historical customer conversion dataset, all first historical interaction datasets belonging to the first relationship stage are extracted from the customer relationship stage sequence. The first historical interaction dataset includes historical interaction data of multiple customers in the first relationship stage. Each data item carries a binary classification label indicating whether the customer has entered the second relationship stage. For example, 0 indicates that the customer has not entered the next stage, and 1 indicates that the customer has entered the next stage.

[0045] The first historical interaction dataset is used to train a logistic regression model. A logistic regression model is a statistical model used for binary classification problems. It linearly combines input features (such as interaction behavior, frequency, and content keywords) to generate a score. This score is then mapped to a range between 0 and 1 using a sigmoid function. The model ultimately outputs the probability of an event occurring. If the probability is greater than a set threshold (such as 0.6), the event is predicted to have occurred (e.g., a successful customer conversion); otherwise, it is predicted not to have occurred.

[0046] The first historical interaction dataset, labeled 0 / 1, is used as input for model training. Each data entry contains customer interaction features (such as interaction duration, number of interactions, keywords used) and a conversion result label. A logistic regression model is trained. After training, the model can predict the probability of a customer converting from the current stage to the next stage based on the new interaction feature data. The first historical interaction dataset is divided into 80% of the data as the training set and 20% of the data as the test set. The fully cleaned and standardized training set is used as input, with customer behavioral features as the independent variables and the conversion label (0 or 1) as the dependent variable. The logistic regression algorithm is used to train an optimal set of feature weight coefficients. The goal of logistic regression is to minimize the difference between the predicted probability and the actual label. Maximum likelihood estimation is typically used for parameter optimization. Feature coefficients are adjusted with each iteration to ensure that the model output probability increasingly matches the actual customer conversion rate. Ultimately, after model convergence, each feature is assigned a specific weight value, representing the direction and intensity of its impact on the customer conversion probability. The validation set is used to evaluate model performance, including accuracy (the percentage of correct predictions), AUC (the area under the curve used to measure classification ability), and recall (the percentage of positive examples recognized). If the model performs well on the validation set, it means that the first probability prediction channel is ready for practical application.

[0047] For each stage in the customer relationship sequence, we extract a separate historical interaction dataset (labeled 0 or 1). We train a separate logistic regression model for each stage. Ultimately, we develop multiple stage-specific probabilistic prediction channels, covering the entire customer relationship lifecycle. Given input features, these multiple probabilistic prediction channels output the probability of an event occurring (labeled 1 or 0).

[0048] Multiple probabilistic prediction channels trained for different customer relationship stages are uploaded and deployed to the cloud center. These channels can be considered functional modules within the cloud center, capable of receiving real-time data and outputting prediction results. Leveraging the high-performance processing capabilities of cloud computing, these channels can rapidly process large amounts of data and provide real-time prediction services. The cloud center is a server platform that centrally manages data, models, and computing resources. It receives real-time customer interaction data, invokes corresponding prediction models, performs calculations, and returns results.

[0049] In the cloud center, we match the corresponding probability prediction channel among multiple probability prediction channels based on the customer's real-time relationship stage and the next adjacent relationship stage. In other words, we match the corresponding probability prediction channel based on the customer's current stage and the next stage they are expected to successfully convert from the current stage. For example, if the current stage is the prospective customer stage and the next stage is the conversion stage, we match the probability prediction channel for the prospective customer stage (i.e., a logistic regression model specifically for prospective customers).

[0050] A dataset of first interactions between the first customer and the corresponding first service representative in the real-time relationship phase is obtained. This dataset is a collection of all interaction records between the customer and the service representative responsible for the customer, including call logs, consultation content, browsing behavior, clickthroughs, feedback content, and other information. These records typically include timestamps, behavior types, and content features. Within the cloud center, the interaction features of the first interaction dataset are input into the corresponding probability prediction channel for stage feature diagnostic analysis. The first predicted probability of transitioning from the real-time relationship phase to the next is predicted. This probability represents the likelihood of the customer successfully transitioning from the current phase to the next, or the likelihood of conversion. For example, a dataset of first interactions collected from customer B, currently in the potential customer phase, includes multiple product browsing sessions over the past seven days, three chat consultations averaging 10 minutes each, one preliminary registration for a trial, and no actual order or prepayment history. This dataset is matched to the corresponding potential customer conversion prediction channel. These interaction features are input into a logistic regression model, and the output of the first predicted probability is 0.74. This indicates a 74% probability of customer B transitioning to the potential customer phase. Based on this, the service representative is advised to expedite follow-up efforts, such as by promoting limited-time offers.

[0051] By deploying a probability prediction channel in the cloud center and analyzing customer interaction data in real time, we can quickly and accurately predict the conversion probability of customers at each relationship stage, which helps to timely identify customer conversion opportunities, formulate personalized marketing strategies, and improve conversion rates and customer satisfaction.

[0052] It should be noted that the first relationship stage and the second relationship stage here do not specifically refer to any particular stage, but only indicate that the two stages are adjacent relationship stages, and the first relationship stage is located before the second relationship stage.

[0053] Furthermore, the stage change prediction module 13 in the customer relationship management system / device based on cloud computing evaluation and diagnosis is also used to:

[0054] An evaluation index construction subunit is used to extract the last relationship stage in the customer relationship stage sequence and construct an adaptation evaluation index for the last stage; a fitness evaluation subunit is used to perform stage interaction fitness evaluation with the adaptation evaluation index for the last stage if the real-time relationship stage is the last relationship stage, and generate an interaction fitness evaluation result; a prediction probability determination subunit is used to perform normalized mapping based on the interaction fitness evaluation result to generate the first prediction probability.

[0055] Specifically, we extracted the final relationship stage from the customer relationship stage sequence. This is the final stage of customer relationship development. For example, after a customer completes their initial treatment, the medical aesthetics institution's goal is to maintain their loyalty and encourage them to return for follow-up visits or refer others. At this point, the customer retention and loyalty stage begins, focusing on whether the customer maintains long-term contact, repurchases, or refers new customers. We designed a set of evaluation indicators specifically for this final relationship stage to assess the effectiveness of customer retention during the loyalty stage, including follow-up visit rate, referral rate, feedback positivity, and frequency of after-sales interaction.

[0056] Determine whether the current customer's real-time relationship stage is the final stage (i.e., customer retention and loyalty). If so, proceed to the subsequent loyalty evaluation process, evaluating the interaction fitness based on the final stage fitness evaluation indicators. Otherwise, continue with the previous conversion probability prediction.

[0057] Based on the final stage's adaptation assessment indicators, the customer's current interactions are scored or graded to determine their loyalty and success in retention. For example, the scores for follow-up records, recommendation records, and interaction records are calculated item by item. The interaction adaptation score for each stage is calculated based on the preset weights. For example, the follow-up rate is assessed at 80 points, the recommendation rate at 70 points, the feedback rate at 90 points, and the interaction frequency at 60 points. Based on the institution's preset indicator weights of 0.3 for the follow-up rate, 0.3 for the recommendation rate, 0.2 for the feedback rate, and 0.2 for the interaction frequency, the calculated overall score is 75 points.

[0058] Normalized mapping generates the first predicted probability, normalizing the comprehensive score to a range of 0 to 1. For example, a score of 0 corresponds to 0, and 100 corresponds to 1. Using linear mapping, a score of 75 corresponds to a predicted probability of 0.75. This serves as the customer's first predicted probability in the final stage, representing the likelihood of successfully maintaining customer loyalty. By constructing and normalizing the final stage's adaptive evaluation indicators, we can quantitatively assess customer loyalty and retention effectiveness, helping to improve long-term customer retention efficiency and extend customer lifetime value.

[0059] The missing factor analysis module 14 is configured to perform an interaction missing factor analysis based on the first interaction data set to generate a first interaction missing factor when the first predicted probability is less than a preset probability threshold.

[0060] Furthermore, the missing factor analysis module 14 in the customer relationship management system / device based on cloud computing evaluation and diagnosis is also used to:

[0061] A data collection unit is used to collect a conversion success data set for any customer relationship stage in the customer relationship stage sequence; a central tendency analysis unit is used to perform central tendency analysis of interaction features on the conversion success data set to generate a calibrated interaction concentration feature for any customer relationship stage; a missing feature comparison unit is used to extract the calibrated interaction concentration feature corresponding to the real-time relationship stage, perform missing feature comparison on the first interaction data set, and generate the first interaction missing factor.

[0062] Specifically, a reference line, or pre-set probability threshold, is set based on historical experience. Any value below this threshold is at risk of conversion failure. Interaction loss factor analysis is then performed on the first interaction dataset to identify potential causes of the low conversion probability. If the first predicted probability is less than the pre-set probability threshold, conversion or maintenance is considered risky, and interaction loss factor analysis is performed.

[0063] From any customer relationship stage in the customer relationship sequence, collect a successful conversion dataset, i.e., collect data on customers who successfully transitioned to the next stage. This dataset is collected using data extraction and built-in filtering tools within the customer management platform.

[0064] Central tendency analysis is performed on the interaction features in the successful conversion dataset to determine the central tendency value of each important interaction feature. This includes performing mean, median, and mode analysis on all interaction features to obtain the calibrated interaction central tendency features for any stage of the customer relationship. For example, characteristics such as average browsing time and number of consultations during the prospect stage can be analyzed to identify typical values ​​or ranges for these characteristics. Calibrated interaction central tendency features are representative interaction features derived through central tendency analysis of successful case data and serve as a standard template for ideal or excellent states.

[0065] Based on the current real-time relationship stage, the corresponding calibrated interaction concentration features are matched. The first interaction dataset of the current customer (first customer) is then compared against the calibrated interaction features one by one to identify missing or deviated key features, thereby generating an interaction loss factor. For example, if the current customer is in the prospective customer stage, the standard interaction features formed during this stage are extracted. If a key feature is missing (such as a missing key consultation question) or significantly deviates from the central trend (such as an excessively long response time), it is recorded as an interaction loss factor. Ultimately, the first interaction loss factor is generated, which serves as a list of key factors contributing to low conversion predictions. For example, after comparison, it may be found that the current customer has insufficient consultations, has excessively long response times, or has a single recommended item. This can assist service personnel in making targeted remedial efforts.

[0066] Through interaction missing factor analysis, we can identify the key factors that influence customer conversions and take targeted measures to improve conversion rates. For example, if we find that customers' browsing time at the prospect stage is lower than average, we can increase customer engagement by optimizing website content or providing personalized recommendations.

[0067] The auxiliary reminder module 15 is used to screen data in the shared knowledge database of the customer management platform using the first interaction loss factor, and recommend the first screened data and the first interaction loss factor to the first service personnel for customer relationship management auxiliary reminder.

[0068] Furthermore, the auxiliary reminder module 15 in the customer relationship management system / device based on cloud computing evaluation and diagnosis is also used to:

[0069] A database construction unit is used to collect multiple customer interaction reference data that meet predetermined data sharing constraints to construct the shared knowledge database, wherein the multiple customer interaction reference data carry target interaction tags; a data matching and screening unit is used to input the first interaction missing factor into the shared knowledge database, extract customer interaction reference data whose target interaction tags match the first interaction missing factor, and generate the first screening data.

[0070] Specifically, predetermined data sharing constraints are determined—restrictions that must be adhered to when sharing data. These include the need for data anonymization (removing sensitive information such as names, facial features, and contact information), client authorization, de-identification, and the sharing of specific user-specific dialogue, case studies, and treatment feedback. Multiple pieces of client interaction reference data that meet these constraints are collected to construct the shared knowledge database. This means that client privacy must be protected during data collection, such as through anonymization and de-identification. This data is then stored in the shared knowledge database for subsequent analysis and query.

[0071] After filtering and processing, interaction data is labeled with target interaction tags based on different interaction purposes and then stored uniformly in a shared knowledge database. Multiple customer interaction reference data items are labeled with target interaction tags, facilitating rapid retrieval and matching.

[0072] Input the first missing interaction factor into the shared knowledge database, and search the shared knowledge database for reference data corresponding to the missing factor and with the same or similar target interaction tags. For example, if the missing factor is a single recommended item, search for historical examples of excellent interactions with the target interaction tag "optimized item recommendation." If the missing factor is slow customer response, search for script templates with the tag "effective communication examples."

[0073] A set of customer interaction reference data that meets the missing factor requirements is extracted to form the first screening data. This first screening data is selected from the shared knowledge database based on the current missing factor and can be used to guide remedial interactions. This first screening data and the first interaction missing factor are pushed in real time to the user interface of the primary service representative responsible for the customer. This includes pop-up notifications, work order task reminders, and suggestions for supplementary interactions. This allows the service representative to quickly understand potential customer issues before or during the next round of communication and adjust communication strategies based on the recommended content.

[0074] By building a shared knowledge database and utilizing the reference data therein, personalized recommendations and services are generated based on the customer's interaction missing factors, which helps service personnel provide targeted services, thereby improving customer satisfaction and conversion rates.

[0075] In summary, the customer relationship management system based on cloud computing evaluation and diagnosis provided by this application has the following technical effects:

[0076] The stage segmentation module connects to the customer management platform, collects historical customer conversion datasets, performs customer conversion stage segmentation, and constructs a customer relationship stage sequence. The stage matching module reads the first conversion feature data of the current first customer and matches it to the real-time relationship stage in the customer relationship stage sequence. The stage transition prediction module collects the first interaction dataset between the first customer and the corresponding first service personnel in the real-time relationship stage, performs stage feature diagnosis and analysis in the cloud center, and predicts a first predicted probability of transitioning from the real-time relationship stage to the next relationship stage. The missing factor analysis module performs interaction missing factor analysis based on the first interaction dataset to generate a first interaction missing factor when the first predicted probability is less than a preset probability threshold. The auxiliary reminder module uses the first interaction missing factor to filter data in the shared knowledge database of the customer management platform and recommends the first filtered data and the first interaction missing factor to the first service personnel for customer relationship management auxiliary reminders. In other words, by reading the current customer's conversion features, matching their relationship stage, collecting interaction data between the customer and the service personnel, analyzing, and predicting whether the customer will enter the next stage. If the predicted probability is lower than the preset threshold, the interaction missing factors are analyzed and relevant suggestions are generated. Relevant data is filtered through the shared knowledge database, and the missing factors and screening results are recommended to service personnel to provide auxiliary reminders for customer relationship management, better respond to changes in customer status, and thus improve customer conversion rate and customer satisfaction.

[0077] In the second embodiment, based on the same inventive concept as the customer relationship management system based on cloud computing evaluation and diagnosis in the first embodiment, this application also provides a customer relationship management method based on cloud computing evaluation and diagnosis, please refer to the attached Figure 2 The customer relationship management method based on cloud computing evaluation and diagnosis includes:

[0078] S100: Connect to the customer management platform, collect historical customer conversion data sets to perform customer conversion stage segmentation, and construct a customer relationship stage sequence; S200: Read the first conversion feature data of the first customer currently in progress, and match the real-time relationship stage to which it belongs in the customer relationship stage sequence; S300: Collect the first interaction data set between the first customer and the corresponding first service personnel in the real-time relationship stage, perform stage feature diagnosis and analysis in the cloud center, and predict the first prediction probability of transitioning from the real-time relationship stage to the next relationship stage; S400: When the first prediction probability is less than a preset probability threshold, perform interaction missing factor analysis based on the first interaction data set to generate a first interaction missing factor; S500: Use the first interaction missing factor to screen data in the shared knowledge database of the customer management platform, and recommend the first screened data and the first interaction missing factor to the first service personnel for customer relationship management auxiliary reminders.

[0079] Furthermore, the connection to the customer management platform collects historical customer conversion data sets to segment customer conversion stages and construct a customer relationship stage sequence, including:

[0080] Based on the historical customer conversion data set, customer continuous behavior analysis is performed to construct a plurality of customer continuous behavior sequences; similar continuous behaviors are respectively merged on the plurality of customer continuous behavior sequences to generate a plurality of merged behavior sequences; and customer relationship stage decomposition is performed based on the plurality of merged behavior sequences to generate the customer relationship stage sequence.

[0081] Furthermore, the similarity merging of the continuous behaviors of the plurality of customer continuous behavior sequences to generate a plurality of merged behavior sequences includes:

[0082] Extracting a first customer continuous behavior sequence from the plurality of customer continuous behavior sequences; merging continuous behavior features in the first customer continuous behavior sequence according to behavior similarity to generate a first merged behavior sequence, and adding the plurality of merged behavior sequences.

[0083] Furthermore, collecting a first interaction dataset between the first customer and the corresponding first service personnel in the real-time relationship stage, performing stage feature diagnosis and analysis in a cloud center, and predicting a first predicted probability of transitioning from the real-time relationship stage to a next relationship stage includes:

[0084] Based on the historical customer conversion dataset, a first historical interaction dataset of the first relationship stage in the customer relationship stage sequence is extracted, wherein each data in the first historical interaction dataset carries a binary classification label indicating whether the second relationship stage has been entered; a logistic regression model is trained using the first historical interaction dataset to establish a first probability prediction channel, and multiple probability prediction channels are added; the multiple probability prediction channels are embedded in a cloud center, and corresponding probability prediction channels are matched in the multiple probability prediction channels based on the real-time relationship stage and the next relationship stage in the cloud center, the first interaction dataset is analyzed, and the first prediction probability is output.

[0085] Furthermore, the first relationship stage and the second relationship stage are two adjacent relationship stages, and the first relationship stage is located before the second relationship stage.

[0086] Further, outputting the first predicted probability includes:

[0087] Extracting the last relationship stage from the customer relationship stage sequence and constructing a last stage adaptation evaluation index; if the real-time relationship stage is the last relationship stage, performing a stage interaction fitness evaluation using the last stage adaptation evaluation index to generate an interaction fitness evaluation result; performing normalized mapping based on the interaction fitness evaluation result to generate the first prediction probability.

[0088] Furthermore, performing interaction missing factor analysis based on the first interaction data set to generate a first interaction missing factor includes:

[0089] A conversion success data set is collected for any customer relationship stage in the customer relationship stage sequence; a central trend analysis of interaction features is performed on the conversion success data set to generate a calibrated interaction concentration feature for any customer relationship stage; the calibrated interaction concentration feature corresponding to the real-time relationship stage is extracted, and a missing feature comparison is performed on the first interaction data set to generate the first interaction missing factor.

[0090] Furthermore, the performing of data screening in the shared knowledge database of the customer management platform using the first interaction missing factor includes:

[0091] Collect multiple customer interaction reference data that meet predetermined data sharing constraints to build the shared knowledge database, wherein the multiple customer interaction reference data carry target interaction labels; input the first interaction missing factor into the shared knowledge database, extract customer interaction reference data whose target interaction labels match the first interaction missing factor, and generate the first screening data.

[0092] The various embodiments in this specification are described in a progressive manner, and each embodiment focuses on the differences from other embodiments. Figure 1 The customer relationship management system based on cloud computing evaluation and diagnosis and the specific examples in Example 1 are also applicable to the customer relationship management method based on cloud computing evaluation and diagnosis in this embodiment. Through the above detailed description of the customer relationship management system based on cloud computing evaluation and diagnosis, those skilled in the art can clearly understand the customer relationship management method based on cloud computing evaluation and diagnosis in this embodiment, so for the sake of brevity of the specification, it will not be described in detail here.

[0093] In the third embodiment, based on the same inventive concept as the customer relationship management system based on cloud computing evaluation and diagnosis in the aforementioned first embodiment, the present application also provides a computer-readable storage medium, on which a computer program is stored. When the computer program is executed, it implements the customer relationship management system based on cloud computing evaluation and diagnosis described in any one of the aforementioned first embodiments.

[0094] The above description of the disclosed embodiments is intended to enable one skilled in the art to implement or use the present application. Various modifications to these embodiments will be readily apparent to one skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the present application. Therefore, the present application is not limited to the embodiments shown herein, but is intended to conform to the widest scope consistent with the principles and novel features disclosed herein.

[0095] Obviously, those skilled in the art may make various changes and modifications to the present application without departing from the spirit and scope of the present application. Thus, if these modifications and variations of the present application fall within the scope of the present application and its equivalents, the present application is intended to include these modifications and variations.

Claims

1. A customer relationship management system based on cloud computing evaluation and diagnosis, characterized by: include: The stage segmentation module is used to connect to the customer management platform, collect historical customer conversion data sets, segment customer conversion stages, and build a customer relationship stage sequence; a stage matching module, configured to read the first conversion feature data of the first customer currently in progress and match the corresponding real-time relationship stage in the customer relationship stage sequence; a stage transition prediction module, configured to collect a first interaction dataset between the first customer and the corresponding first service personnel in the real-time relationship stage, perform stage feature diagnosis and analysis in a cloud center, and predict a first prediction probability of transitioning from the real-time relationship stage to a next relationship stage; a missing factor analysis module, configured to perform an interaction missing factor analysis based on the first interaction data set to generate a first interaction missing factor when the first predicted probability is less than a preset probability threshold; an auxiliary reminder module, configured to screen data in a shared knowledge database of the customer management platform using the first interaction loss factor, and recommend the first screened data and the first interaction loss factor to the first service personnel for customer relationship management auxiliary reminder; The stage change prediction module includes: a first interaction data extraction unit configured to extract, based on the historical customer conversion dataset, a first historical interaction dataset for a first relationship stage in the customer relationship stage sequence, wherein each data item in the first historical interaction dataset carries a binary classification label indicating whether the customer has entered the second relationship stage; a first prediction channel establishing unit, configured to train a logistic regression model using the first historical interaction data set, establish a first probability prediction channel, and add a plurality of probability prediction channels; A probability prediction channel matching unit is used to embed the multiple probability prediction channels into a cloud center, match corresponding probability prediction channels in the multiple probability prediction channels based on the real-time relationship stage and the next relationship stage in the cloud center, analyze the first interaction data set, and output the first prediction probability.

2. The customer relationship management system based on cloud computing evaluation and diagnosis according to claim 1, characterized in that: The stage segmentation module includes: A behavior analysis unit, configured to perform customer continuous behavior analysis based on the historical customer conversion data set and construct a plurality of customer continuous behavior sequences; A similarity merging unit, configured to perform similarity merging on the plurality of customer continuous behavior sequences to generate a plurality of merged behavior sequences; The stage decomposition unit is used to perform customer relationship stage decomposition based on the plurality of merged behavior sequences to generate the customer relationship stage sequence.

3. The customer relationship management system based on cloud computing evaluation and diagnosis according to claim 2, characterized in that: The similar merging unit includes: A first sequence extraction subunit is configured to extract a first customer continuous behavior sequence from the plurality of customer continuous behavior sequences; The first merging subunit is configured to merge the continuous behavior features in the first customer's continuous behavior sequence according to behavior similarity to generate a first merged behavior sequence, and add the plurality of merged behavior sequences.

4. The customer relationship management system based on cloud computing evaluation and diagnosis according to claim 1, characterized in that: The first relationship stage and the second relationship stage are two adjacent relationship stages, and the first relationship stage is located before the second relationship stage.

5. The customer relationship management system based on cloud computing evaluation and diagnosis according to claim 1, characterized in that: The probability prediction channel matching unit further includes: An evaluation index construction subunit is used to extract the last relationship stage in the customer relationship stage sequence and construct an evaluation index adapted to the last stage; a fitness evaluation subunit, configured to, if the real-time relationship stage is the final relationship stage, perform stage interaction fitness evaluation using the final stage fitness evaluation index to generate an interaction fitness evaluation result; The prediction probability determination subunit is used to perform normalized mapping based on the interactive fitness evaluation result to generate the first prediction probability.

6. The customer relationship management system based on cloud computing evaluation and diagnosis according to claim 1, characterized in that: The missing factor analysis module includes: A data collection unit, configured to collect a conversion success data set for any customer relationship stage in the customer relationship stage sequence; a central tendency analysis unit, configured to perform a central tendency analysis of interaction features on the conversion success data set to generate a calibrated interaction central tendency feature for any customer relationship stage; The missing feature comparison unit is used to extract the features in the calibration interaction set corresponding to the real-time relationship stage, perform missing feature comparison on the first interaction data set, and generate the first interaction missing factor.

7. The customer relationship management system based on cloud computing evaluation and diagnosis according to claim 1, characterized in that: The auxiliary reminder module includes: a database construction unit, configured to collect a plurality of customer interaction reference data satisfying predetermined data sharing constraints to construct the shared knowledge database, wherein the plurality of customer interaction reference data carry target interaction tags; The data matching and screening unit is configured to input the first interaction missing factor into the shared knowledge database, extract customer interaction reference data whose target interaction tag matches the first interaction missing factor, and generate the first screening data.

8. A customer relationship management method based on cloud computing evaluation and diagnosis, characterized in that: The customer relationship management system based on cloud computing evaluation and diagnosis according to any one of claims 1 to 7 is implemented, and the customer relationship management method based on cloud computing evaluation and diagnosis comprises: Connect to the customer management platform, collect historical customer conversion data sets to segment customer conversion stages, and build a customer relationship stage sequence; Reading first conversion feature data of a first customer currently in progress, and matching the corresponding real-time relationship stage in the customer relationship stage sequence; collecting a first interaction dataset between the first customer and the corresponding first service personnel in the real-time relationship stage, performing stage feature diagnosis and analysis in a cloud center, and predicting a first prediction probability of transitioning from the real-time relationship stage to a next relationship stage; When the first predicted probability is less than a preset probability threshold, performing interaction missing factor analysis based on the first interaction data set to generate a first interaction missing factor; The first interaction missing factor is used to screen data in the shared knowledge database of the customer management platform, and the first screened data and the first interaction missing factor are recommended to the first service personnel for customer relationship management auxiliary reminder.

9. A computer-readable storage medium, characterized in that The computer-readable storage medium stores a computer program, which, when executed, implements the customer relationship management method based on cloud computing evaluation and diagnosis as described in claim 8.

Images