System for automatically matching hyper-personalized CRM content through customer data platform

Abstract

The present invention relates to a system for automatically matching personalized CRM content through a customer data platform that matches customized CRM content for each individual on the basis of personal profiling, the system comprising: a customer data platform (CDP) for collecting behavior data of a customer and generating a profile of the customer; a CRM content generation unit for generating CRM content on the basis of the profile of the customer by using generative artificial intelligence according to a predefined message type; a matching unit for matching the CRM content suitable for the customer according to the profile of the customer; and a CRM transmission unit for transmitting, to each customer terminal, customer-specific CRM content matched through the matching unit, wherein the CDP analyzes customer response data for the CRM content transmitted by the CRM transmission unit, and applies the analyzed data to the profile of the customer.

Description

Hyper-personalized CRM content automatic matching system via customer data platform

[0001] The present invention relates to an automatic personalized CRM content matching system through a customer data platform that matches personalized content based on personal profiling.

[0002] Customer Relationship Management (CRM) refers to the management of customer relationships. It is an integrated system that plays a pivotal role in the modern business environment, facilitating communication with customers, promoting customer retention, and ultimately driving revenue generation. Defined as the ability to understand, analyze, and utilize customer behavior data, CRM serves as a strategic solution that enables companies to foster lasting relationships with customers, tailor marketing strategies to customer characteristics, and maximize customer-centric resources.

[0003] In other words, CRM is a method of strengthening long-term relationships with customers and executing customized marketing strategies by analyzing their behavior and needs; the core objective of a CRM marketing strategy is to increase customer satisfaction and enhance customer loyalty.

[0004] However, existing CRM (Customer Relationship Management) systems generally define CRM sending conditions based on rules and generate and send CRM messages accordingly.

[0005] This method requires all rules to be generated mechanically, and automation is possible only within a limited scope.

[0006] For example, when sending CRM messages to customers who have added products to their shopping cart but have not made a purchase, the rule is defined so that messages are no longer sent to customers who have made a purchase, and additional messages are sent to customers who have not made a purchase.

[0007] In addition, existing personalized CRM methods are limited to personalizing only recommended products by utilizing individual behavioral data (e.g., views, shopping cart, wishlist, purchase, etc.) from CRM messages.

[0008] As a result, rules and CRM messages must be continuously updated to improve and optimize performance, which requires significant resources and time.

[0009] The present invention provides an automatic matching system for hyper-personalized CRM content through a customer data platform that automatically matches hyper-personalized CRM content based on personal profiling generated through a CDP and sends it to each individual.

[0010] The technical problems that the present invention aims to solve are not limited to those mentioned above, and other unmentioned technical problems will be clearly understood by those skilled in the art to which the present invention belongs from the description below.

[0011] The hyper-personalized CRM content automatic matching system through the customer data platform of the present invention for achieving the above objective comprises: a Customer Data Platform (CDP) that collects customer behavior data and generates a customer profile; a CRM Content Generation Unit that generates CRM content based on the customer profile by generative artificial intelligence based on a predefined message type; a Matching Unit that matches the CRM content suitable for the customer according to the customer profile; and a CRM Dispatch Unit that sends the customer-specific CRM content matched through the Matching Unit to each customer terminal; wherein the CDP analyzes customer response data to the CRM content sent by the CRM Dispatch Unit and applies it to the customer profile.

[0012] In addition, the customer profile is characterized by being data that includes one or more of the customer's purchasing journey, preferences, satisfaction, and responsiveness to CRM content.

[0013] In addition, the above customer profile is characterized by being updated in real time.

[0014] In addition, the matching unit is characterized by setting priorities through a rule-based algorithm or through customer responses via a machine learning model.

[0015] In addition, the above rule-based algorithm is characterized by grouping and classifying customers based on customer behavior data and conditions, and matching the classified groups with pre-set high-priority CRM content.

[0016] In addition, the machine learning model is characterized by analyzing customer behavior data and continuously learning each customer's response to match CRM content according to the customer type.

[0017] In addition, the above CRM sending unit is characterized by evaluating the ROAS (Return on Advertising Spend) for each sending channel and sending CRM content in order of highest ROAS.

[0018] In addition, the CRM content generation unit is characterized by generating CRM content in real time by considering one or more external factors based on the customer's location, weather, and time.

[0019] As explained above,

[0020] The present invention has the effect of maximizing marketing performance by automatically matching optimal CRM content to customer behavior data through machine learning.

[0021] In addition, the present invention has the effect of maximizing the purchase conversion rate by automatically providing recommendations for similar products and customized CRM content based on customer behavior data collected in the CDP.

[0022] In addition, the present invention has the effect of establishing a sophisticated CRM marketing strategy by analyzing and applying the results of CRM performance through a feedback loop.

[0023] FIG. 1 is a block diagram showing an automatic matching system for hyper-personalized CRM content through a customer data platform according to an embodiment of the present invention.

[0024] FIG. 2 is a diagram showing the classification of personal profiles based on customer behavior data of a hyper-personalized CRM content automatic matching system through a customer data platform according to an embodiment of the present invention.

[0025] FIG. 3 is a diagram showing examples of type and step definitions in a hyper-personalized CRM content automatic matching system through a customer data platform according to an embodiment of the present invention.

[0026] Figure 4 is a diagram showing an example of CRM content matching in a hyper-personalized CRM content automatic matching system through a customer data platform according to an embodiment of the present invention.

[0027] FIG. 5 is a diagram showing content to which hyper-personalization variables of a hyper-personalized CRM content automatic matching system through a customer data platform according to an embodiment of the present invention have been applied.

[0028] Hereinafter, embodiments according to the present invention will be described in detail with reference to the attached drawings.

[0029] Furthermore, the size or shape of components depicted in the drawings may be exaggerated for clarity and convenience of explanation, and terms specifically defined in consideration of the configuration and operation of the present invention may vary according to the intent or convention of the user or operator, and the definitions of such terms must be based on the content throughout this specification.

[0030] FIG. 1 is a block diagram showing an automatic matching system for hyper-personalized CRM content through a customer data platform according to an embodiment of the present invention. As shown in FIG. 1, the present invention is composed of a CDP (Customer data platform, 100), a CRM content generation unit (200), a matching unit (300), and a CRM sending unit (400).

[0031] The CDP (Customer data platform, 100) collects customer behavior data and creates customer profiles.

[0032] Here, customer behavioral data includes membership status, marketing consent status, purchase history, coupon history, shopping cart status, and search term lists.

[0033] The profile is data including one or more of the customer's purchase journey (site activity, purchase history), preferences (interests), satisfaction, and responsiveness to CRM content (210), and the customer's profile is updated in real time.

[0034] In addition, the customer profile based on customer behavior data is divided into five stages as shown in Fig. 2.

[0035] The five stages are divided into Acquisition, Explore, Consideration, Buy, and Churn.

[0036] Here, the Acquisition stage includes whether the customer has signed up or is registered as an open chat room or messenger friend that is convenient for CRM marketing. For example, Kakao Plus Friends falls under this category.

[0037] In addition, the Explore stage includes customer actions such as viewing product detail pages or launching specific products, the Consideration stage includes customer actions such as wishlisting or adding products to a cart, and the Buy stage includes the state where the customer has actually purchased a product. Furthermore, the Churn stage includes a state where the customer has no purchase history for a certain period, or has a purchase history but is unlikely to re-engage.

[0038] Customer profiles based on customer behavior data are classified into five stages, and depending on the type of customer, they can be further classified in detail as shown in Fig. 3.

[0039] Customer types can be matched with any one of the following: social connectors, casual subscribers, full subscribers, browsers, information subscribers, keyword registrants, active explorers, interested parties, purchase considerers, new buyers, silent buyers, satisfied buyers, dissatisfied buyers, potential churners, high-value dormant customers, and low-value dormant customers.

[0040] The RFM values ​​in Fig. 3 correspond to Recency (R), Monetary (M), and Frequency (F), and the RMF score (Total RMF score) is derived by the following Equation 1.

[0041] [Equation 1]

[0042]

[0043]

[0044] At this time, represents the weight, R_score represents the Recency score, M_score represents the Monetary score, and F_score represents the Frequency score.

[0045] Recency decreases as the purchase date increases, Monetary increases as the average purchase price increases, and Frequency increases as the number of purchases increases.

[0046] Based on this, the RMF score for an arbitrary customer is calculated as follows.

[0047] When assigning weights, the purchase date is the most important, the purchase amount is also important but less important than the purchase date, and the number of purchases has the same importance as the purchase amount, After assigning weights of 0.4, 0.3, and 0.3 respectively, the RMF scores derived according to the conditions of an arbitrary customer are as follows.

[0048] If customer A's last purchase date was 10 days ago, the recent purchase amount was 45,000 won, and the total number of purchases was 4 times, the RMF score is as follows.

[0049]

[0050]

[0051] In addition, if customer B's last purchase date was 30 days ago, the recent purchase amount was 30,000 won, and the total number of purchases was 2, the RMF score is as follows.

[0052]

[0053]

[0054] Finally, if customer C's last purchase date was 5 days ago, the recent purchase amount was 35,000 won, and the total number of purchases was 3 times, the RMF score is as follows.

[0055]

[0056] The CRM content generation unit (200) generates CRM content (210) based on the customer's profile by a generative artificial intelligence based on a predefined message type.

[0057] Here, message types are classified into product information, event information, promotion information, and tone and manner.

[0058] Additionally, the CRM content generation unit (200) generates CRM content (210) in real time by considering one or more external factors based on the customer's location, weather, and time.

[0059] Additionally, the CRM content generation unit (200) can generate messages of various types, and if the message type is defined, it can automatically generate messages as prompts through the LLM (Large Language Model).

[0060] At this time, the generated CRM content (210) is transmitted to a marketer terminal (not shown) so that the marketer can modify and confirm it.

[0061] The matching unit (300) matches CRM content suitable for the customer according to the customer's profile.

[0062] A detailed personalization unit (not shown) can be added to match CRM content suitable for a customer through the matching unit (300) and to make a judgment on what percentage of coupon the customer will respond most effectively to, whether it is free shipping, shipping discount, or if it is a discount, what percentage is appropriate.

[0063] As illustrated in FIG. 4, the CRM content (210) matching selects the most suitable message from a predefined set of messages based on the customer's segmented behavioral data and conditions, and the matching unit (300) sets the priority through a rule-based algorithm or through the customer's response through a machine learning model.

[0064] Here, the rule-based algorithm groups and classifies customers based on their behavioral data and conditions, and matches the classified groups with pre-set high-priority CRM content types.

[0065] In this case, the priority may be set by the marketer, fluctuate over time, or change depending on specific events. Specific events may refer to promotion periods.

[0066] In addition, the classified groups may be classified into customers who add items to their cart, customers who add items to their wishlist, and customers who repeatedly view the same product.

[0067] For example, for customers who have added items to their cart, CRM content containing discount coupon messages related to the items in the cart is matched first.

[0068] Machine learning models analyze customer behavior data and continuously learn each customer's response to match CRM content according to the customer type.

[0069] Machine learning models are capable of finding patterns in previously unencountered datasets or making decisions based on them; through parsing, they can correctly recognize the intent of unfamiliar weapons or word combinations.

[0070] When matching CRM content, it is desirable to consider one or more parameters among customer responsiveness, message exposure frequency and fatigue index, message response rate (conversion rate, click-through rate), customer state variables, model accuracy, customer behavior data (purchase, click, view, wishlist, etc.), product characteristics, event characteristics, and recommendation accuracy according to various algorithms.

[0071] Machine learning models can be classified by application case, and the required parameters may also differ.

[0072] In machine learning algorithms for mode switching designed to avoid sending the same content to customers as fatigue builds up when a customer's status does not change for a certain period, customer responsiveness, message exposure frequency, and a fatigue index are considered. Customer responsiveness sets rewards or penalties based on customer response data when changing messages, message exposure frequency determines how many times the same message is sent before switching to a new message, and the fatigue index tracks fatigue when customers do not respond to messages to optimize the timing of message changes.

[0073] In addition, rule-based message sending allows customers to receive various messages depending on their state. Since the most efficient message must be received, the machine learning algorithm used to determine which message is the most efficient considers the message response rate, customer state variables, and model accuracy. The message response rate trains the model based on the customer's response to each message and is used when selecting messages. Customer state variables utilize various variables to enable the machine learning model to make predictions based on the customer's current state, and model accuracy evaluates the predictive performance of the machine learning model and selects the optimal CRM content.

[0074] Furthermore, machine learning algorithms designed to determine which events are most efficient to offer to customers or which products are most effective to recommend consider customer behavior data, product and event characteristics, and recommendation accuracy. This is because customer behavior data determines the timing of recommendations and event delivery based on the customer's past behavior; product and event characteristics optimize the recommendation model by considering the characteristics of the products to be recommended or the events to be provided; and recommendation accuracy optimizes the model based on the customer's response to the recommended products or events.

[0075] In the implementation of the present invention, if a model trained on past data shows a response to any CRM content of a specific type of customer, that CRM content is preferentially matched to that customer.

[0076] By continuously learning customer responses, the accuracy of the model can be improved, and this feedback can enhance the efficiency of message matching.

[0077] Here, customer responses can be improved by continuously learning from past reactions to delivered CRM content, such as message clicks, searches for products included in the message, and purchases, thereby enhancing model accuracy and maximizing marketing performance.

[0078] The CRM sending unit (400) sends the customer-specific CRM content (210) matched through the matching unit (300) to each customer terminal.

[0079] The CRM sending unit (400) evaluates the ROAS (Return on Advertising Spend) for each channel of the sending channel and sends CRM content in order of highest ROAS.

[0080] At this time, if the CRM content (210) fails to be sent, it can be retried through another channel, and the optimal channel can be selected based on the customer's response data.

[0081] For example, send CRM content by selecting the most effective channel among various channels such as email, SMS, and open chat rooms, or send CRM content through the channel preferred by the customer.

[0082] As illustrated in Fig. 5, similar product recommendations and customized CRM content are automatically provided based on customer behavior data collected in the CDP, and generated CRM content (210) is delivered to the customer terminal according to customer information and customer classification collected through the CDP to maximize the purchase conversion rate.

[0083] Therefore, the present invention has the effect of maximizing marketing performance by automatically matching optimal CRM content suitable for customer behavior data through machine learning.

[0084] In addition, the present invention has the effect of maximizing the purchase conversion rate by automatically providing recommendations for similar products and customized CRM content based on customer behavior data collected in the CDP.

[0085] In addition, the present invention has the effect of establishing a sophisticated CRM marketing strategy by analyzing and applying the results of CRM performance through a feedback loop.

[0086] Although embodiments according to the present invention have been described above, they are merely illustrative and those skilled in the art will understand that various modifications and equivalent embodiments are possible therefrom. Accordingly, the true technical scope of protection of the present invention should be determined by the following claims.

[0087] The present invention relates to an automatic personalized CRM content matching system through a customer data platform that matches personalized content based on personal profiling.

[0088] As explained above, the present invention has the effect of maximizing marketing performance by automatically matching optimal CRM content to customer behavior data through machine learning.

Claims

1. A Customer Data Platform (CDP) that collects customer behavior data and creates customer profiles; A CRM content generation unit that generates CRM content based on customer profiles by generative artificial intelligence based on predefined message types; A matching unit that matches the above CRM content suitable for the customer according to the above customer profile; and A CRM sending unit that sends customer-specific CRM content matched through the above-mentioned matching unit to each customer terminal; comprising: The above CDP is a hyper-personalized CRM content automatic matching system through a customer data platform characterized by analyzing customer response data to CRM content sent by the above CRM sending unit and applying it to the customer's profile.

2. In Claim 1, A hyper-personalized CRM content automatic matching system through a customer data platform, characterized in that the above customer profile is data including one or more of the customer's purchase journey, preferences, satisfaction, and responsiveness to CRM content.

3. In Claim 1, A hyper-personalized CRM content automatic matching system through a customer data platform characterized by the above customer profile being updated in real time.

4. In Claim 1, The above-mentioned matching unit is a hyper-personalized CRM content automatic matching system through a customer data platform, characterized by setting priorities through a rule-based algorithm or through customer responses through a machine learning model.

5. In Claim 4, The above rule-based algorithm is characterized by grouping and classifying customers based on customer behavior data and conditions, and matching the classified groups with pre-set high-priority CRM content types, thereby creating a hyper-personalized CRM content automatic matching system through a customer data platform.

6. In Claim 4, The above machine learning model is characterized by analyzing customer behavior data and continuously learning each customer's response to match CRM content according to the customer type, thereby creating a hyper-personalized CRM content automatic matching system through a customer data platform.

7. In Claim 1, The above CRM sending unit is characterized by evaluating the ROAS (Return on Advertising Spend) for each sending channel and sending CRM content in order of highest ROAS, thereby creating a hyper-personalized CRM content automatic matching system through a customer data platform.

8. In Claim 1, The above CRM content generation unit is characterized by generating CRM content in real time by considering one or more external factors among the customer's location, weather, and time, in a hyper-personalized CRM content automatic matching system through a customer data platform.

Images