Ai-based customer data analysis, lead scoring and automated content generation system

Abstract

The present invention relates to an AI-based customer data analysis, lead scoring, and automatic content generation system comprising: a collection unit that collects and preprocesses customer information to extract customer characteristic information; a prediction unit that predicts potential customers whose lead score calculated based on the customer characteristic information is above a preset threshold and generates sales recommendation information; a content generation unit that generates sales content based on the sales recommendation information; and a performance analysis unit that generates sales performance information based on the results of utilizing the sales content.The system includes an output unit that visualizes and outputs the sales recommendation information and sales performance information in the form of a dashboard, wherein the sales performance information includes performance indicators derived by analyzing the sending success rate, response rate, and open rate of the sales content sent by a sales representative to a customer for each customer or sales activity unit, and the collection unit includes a collection module that collects the customer information, a normalization module that unifies the format of the customer information and verifies duplicates and missing values, and a conversion module that converts the customer information to fit the input format of the prediction unit to extract the customer characteristic information, wherein the collection unit searches for related words including product names, group names, and industry names related to the customer in documents, articles, and posts published on the web connected through an external integration channel, generates related word information by aggregating the frequency of occurrence of the related words, maps the related word information to the customer information, and adds it to the customer characteristic information, and the prediction unit calculates the lead score by quantifying the potential customer's purchase conversion probability through a lead scoring model that performs an evaluation operation applying weights to attributes of the customer information, and the prediction unit corrects the weights based on the related word information, wherein the aggregation period of the related word information, the related words The ratio at which the associated word information is reflected in the weight is adjusted according to the reliability of the collected source and the matching form of the associated word, and the collection unit counts the occurrence of the associated word only once when the associated word is repeatedly detected from the same source, and excludes from the counting of the associated word when the same string is structurally repeated on pages automatically generated during web crawling or on simple list-type pages.

Description

Technology Field

[0001] The present invention relates to an AI-based customer data analysis, lead scoring, and automatic content generation system. Background Technology

[0002] Recently, corporate sales activities are undergoing rapid digital transformation to enhance the efficiency of customer management and transaction expansion. In the past, the primary method involved relying on the experience and intuition of sales representatives to identify customers and draft proposals; however, this approach has limitations, such as the difficulty of systematically managing data and the disruption of sales know-how when personnel change within the organization.

[0003] Furthermore, customer-related information held by companies is accumulated in a dispersed manner across various forms, such as product documents, quotation details, and transaction records; however, due to a lack of interrelationships, this data is difficult to utilize effectively for formulating actual sales strategies or customer targeting.

[0004] In particular, during the process of selecting promising customers or determining customer-specific response strategies, subjective judgment rather than quantitative evidence intervenes, leading to a problem where consistency in performance is not ensured.

[0005] To address these limitations, Customer Relationship Management (CRM) systems and marketing automation tools are being introduced; however, existing technologies are primarily focused on data input and storage and fall short of providing real-time analytical results applicable in the field. Consequently, sales representatives continue to spend significant time on repetitive customer scouting and documentation tasks, making it difficult to formulate differentiated strategies for each individual customer. Furthermore, from a management perspective, the lack of visibility into individual performance and the overall sales flow hinders the smooth execution of data-driven decision-making.

[0006] Therefore, there is a need to develop a system that can comprehensively manage various customer-related information held by a company and, based on this, systematically analyze customer characteristics and the situation at each sales stage. Prior art literature

[0007] Republic of Korea Registered Patent Publication No. 10-2608849 The problem to be solved

[0008] The present invention was devised to solve the aforementioned conventional problems and aims to provide an AI-based customer data analysis, lead scoring, and automatic content generation system that enables efficient sales decision-making and systematic customer management by integrally managing customer-related information, product information, and sales activity history dispersed within an organization, and by predicting customer purchase potential and sales performance through artificial intelligence based on customer characteristics and sales stages using the collected information. means of solving the problem

[0009] The above objective is achieved, according to the present invention, by a collection unit that collects and preprocesses customer information to extract customer characteristic information; a prediction unit that predicts potential customers whose lead score calculated based on the customer characteristic information is greater than or equal to a preset threshold and generates sales recommendation information; a content generation unit that generates sales content based on the sales recommendation information; and a performance analysis unit that generates sales performance information based on the results of utilizing the sales content.The system includes an output unit that visualizes and outputs the sales recommendation information and sales performance information in the form of a dashboard, wherein the sales performance information includes performance indicators derived by analyzing the sending success rate, response rate, and open rate of the sales content sent by a sales representative to a customer for each customer or sales activity unit, and the collection unit includes a collection module that collects the customer information, a normalization module that unifies the format of the customer information and verifies duplicates and missing values, and a conversion module that converts the customer information to fit the input format of the prediction unit to extract the customer characteristic information, wherein the collection unit searches for related words including product names, group names, and industry names related to the customer in documents, articles, and posts published on the web connected through an external integration channel, generates related word information by aggregating the frequency of occurrence of the related words, maps the related word information to the customer information, and adds it to the customer characteristic information, and the prediction unit calculates the lead score by quantifying the potential customer's purchase conversion probability through a lead scoring model that performs an evaluation operation applying weights to attributes of the customer information, and the prediction unit corrects the weights based on the related word information, wherein the aggregation period of the related word information, the related words This is achieved by an AI-based customer data analysis, lead scoring, and automatic content generation system that adjusts the ratio at which the associated word information is reflected in the weight according to the reliability of the collected source and the matching form of the associated word, and the collection unit counts the occurrence of the associated word only once when the associated word is repeatedly detected from the same source, and excludes from the aggregation target of the associated word when the same string is structurally repeated on pages automatically generated during web crawling or on simple list-type pages.

[0010] delete

[0011] In addition, the AI-based customer data analysis, lead scoring, and automatic content generation system further includes a feedback unit that retrains the lead scoring model based on the sales performance information and updates the lead score.

[0012] In addition, the above-mentioned AI-based customer data analysis, lead scoring, and automatic content generation system further includes a strategy generation unit that collects product information and generates sales strategy information, which is information regarding sales methods and sales strategies for each product.

[0013] delete Effects of the invention

[0014] According to the present invention, by integrating and analyzing customer information, product information, and sales activity data distributed within an organization based on artificial intelligence, customer characteristics and the likelihood of purchase conversion can be quantitatively predicted, and based on this, sales strategies can be established and sales documents such as proposals, quotations, and contracts can be automatically generated, thereby improving the efficiency of the entire sales process and ensuring objectivity in decision-making.

[0015] In addition, by collecting the utilization results of generated sales content in real time to quantitatively evaluate sales activity performance and reflecting them in the training of the prediction model, the accuracy of sales forecasts and the precision of strategy recommendations can be continuously improved over time.

[0016] Meanwhile, the effects of the present invention are not limited to those mentioned above, and various effects may be included within the scope obvious to a person skilled in the art from the contents described below. Brief explanation of the drawing

[0017] FIG. 1 illustrates an AI-based customer data analysis, lead scoring, and automatic content generation system according to one embodiment of the present invention. FIG. 2 illustrates the connections between the components of a collection unit of an AI-based customer data analysis, lead scoring, and automatic content generation system according to an embodiment of the present invention. FIG. 3 illustrates an example of a prediction unit visualized as a dashboard of an AI-based customer data analysis, lead scoring, and automatic content generation system according to an embodiment of the present invention. FIGS. 4 to 9 illustrate an example of sales content visualized on a dashboard of an AI-based customer data analysis, lead scoring, and automatic content generation system according to an embodiment of the present invention. FIG. 10 illustrates an example of sales performance information visualized on a dashboard of an AI-based customer data analysis, lead scoring, and automatic content generation system according to an embodiment of the present invention. FIGS. 11 and 12 illustrate an example of an output unit of an AI-based customer data analysis, lead scoring, and automatic content generation system according to an embodiment of the present invention. Specific details for implementing the invention

[0018] Hereinafter, some embodiments of the present invention will be described in detail with reference to the exemplary drawings. It should be noted that in assigning reference numerals to the components of each drawing, the same components are given the same reference numeral whenever possible, even if they are shown in different drawings.

[0019] In addition, when describing embodiments of the present invention, if it is determined that a detailed description of related known configurations or functions would hinder understanding of the embodiments of the present invention, such detailed description is omitted.

[0020] In addition, terms such as first, second, A, B, (a), (b), etc., may be used when describing the components of the embodiments of the present invention. These terms are used merely to distinguish the components from other components, and the essence, order, or sequence of the components is not limited by the terms used.

[0021] Additionally, in the specification, the singular form includes the plural form unless specifically stated otherwise in the text. The terms “comprising” and / or “comprising” as used in the specification do not exclude the presence or addition of one or more other components in addition to the mentioned components.

[0022] Additionally, in describing embodiments of the present invention, each "part," "module," or "step" may be implemented through a processor and memory. The term "processor" should be broadly interpreted to include general-purpose processors, central processing units (CPUs), microprocessors, digital signal processors (DSPs), controllers, microcontrollers, state machines, etc. In some environments, the term "processor" may refer to an application-specific integrated circuit (ASIC), a programmable logic device (PLD), a field programmable gate array (FPGA), etc. The term "processor" may also refer to a combination of processing devices, such as, for example, a combination of a DSP and a microprocessor, a combination of multiple microprocessors, a combination of one or more microprocessors combined with a DSP core, or any other combination of such configurations.

[0023] Furthermore, memory should be broadly interpreted to include any electronic component capable of storing electronic information. Memory may also refer to various types of processor-readable media such as random access memory (RAM), read-only memory (ROM), non-volatile random access memory (NVRAM), programmable read-only memory (PROM), eraseable-programmable read-only memory (EPROM), electrically eraseable PROM (EEPROM), flash memory, magnetic or optical information storage devices, registers, etc. If a processor can read information from memory or write information to memory, memory is said to be in an electronic communication state with the processor, and each "part," "module," or "stage" may be implemented through a program or application based on the processor and memory in an electronic communication state.

[0024] Furthermore, in the present invention, Artificial Intelligence (AI) refers to a technology that imitates human learning ability, reasoning ability, and perceptual ability, and implements them on a computer, and may include concepts such as machine learning and symbolic logic. Machine Learning (ML) is an algorithmic technology that classifies or learns the characteristics of input information on its own. AI technology can analyze input information as a machine learning algorithm, learn from the results of that analysis, and make judgments or predictions based on the results of that learning. Additionally, technologies that mimic the functions of the human brain, such as cognition and judgment, by utilizing machine learning algorithms can also be understood as falling within the category of AI. For example, technological fields such as linguistic understanding, visual understanding, reasoning / prediction, knowledge representation, and motion control may be included.

[0025] Artificial intelligence learning models or neural network models can be designed to implement the structure of the human brain on a computer and may include multiple network nodes that simulate neurons of a human neural network and have weights. These multiple network nodes can have interconnected relationships by simulating the synaptic activity of neurons, where neurons exchange signals through synapses. In an artificial intelligence learning model, multiple network nodes can be located in layers of different depths and exchange information according to convolutional connections. An artificial intelligence learning model may be, for example, an Artificial Neural Network (ANN) or a Convolutional Neural Network (CNN). An artificial intelligence learning model can be machine learned according to methods such as supervised learning, unsupervised learning, and reinforcement learning. Machine learning algorithms that can be used to perform machine learning include Decision Trees, Bayesian Networks, Support Vector Machines, Perceptrons, and Clustering.

[0027] Now, with reference to the attached drawings, an AI-based customer data analysis, lead scoring, and automatic content generation system (100) according to one embodiment of the present invention will be described in detail.

[0028] As illustrated in FIG. 1, an AI-based customer data analysis, lead scoring, and automatic content generation system (100) according to one embodiment of the present invention includes a collection unit (110), a prediction unit (120), a content generation unit (130), a performance analysis unit (140), a feedback unit (150), and / or an output unit (160).

[0029] First, the collection unit (110) collects customer information and preprocesses it to extract customer characteristic information, and is electrically connected to the prediction unit (120).

[0030] The collection unit (110) can receive customer information from a customer management system, a sales activity record storage unit, or an external linkage channel, and preprocess the information into a form suitable for artificial intelligence analysis to extract customer characteristic information.

[0031] Here, "customer information" refers to basic information regarding potential or existing customers targeted for sales, and refers to raw data used as input for the prediction and lead scoring of the artificial intelligence model in this invention.

[0032] Customer information may include company information, transaction history information, response history information, sales stage information, and customer classification information.

[0033] Corporate information refers to information used to identify basic attributes of a customer, such as the company name, industry group, company size, location, job title, and contact information of the person in charge; this information can be utilized in the sales strategy formulation stage by identifying the customer's industrial characteristics or market position.

[0034] Transaction history information refers to information regarding past transaction records with customers and may include data such as the date and time of purchase of products or services, transaction amount, contract period, delivery history, and product family used. Transaction history information can be used to predict the likelihood of repurchase or upselling opportunities by analyzing customer purchasing patterns and contract cycles.

[0035] Response history refers to information indicating a customer's response to sales activities and may include whether a proposal was received, email open and click history, meeting participation, and inquiry or response history. As a key indicator of customer interest and engagement, response history is utilized by AI models to generate personalized content for each customer or calculate lead scores.

[0036] Sales stage information refers to data indicating the sales progress stage where a customer is currently located, and can be classified into the prospect stage, lead stage, proposal stage, quotation stage, and / or contract stage. Sales stage information can be used to track progress by customer, prioritize sales activities, and predict the likelihood of success at each stage.

[0037] Customer classification information refers to classification data based on customers' transaction relationships or retention status, and may include classifications such as new customers, retained customers, renewing customers, and churned customers. Customer classification information can be utilized to manage the customer lifecycle, efficiently allocate sales resources, and train segmented predictive models.

[0038] As illustrated in FIG. 2, the collection unit (110) may include a collection module (111), a normalization module (112), and / or a conversion module (113).

[0039] The collection module (111) collects customer information from a customer management system, a sales activity record storage unit, an external data linkage server, or a user input interface, and is electrically connected to the normalization module (112).

[0040] The collection module (111) can collect customer information in various formats in bulk and store the source and time of collection of each data as metadata to ensure the reliability and traceability of the data.

[0041] In addition, the collection module (111) can synchronize data not only in real-time input but also in a periodic batch manner, and can perform a filtering function to minimize the occurrence of duplicate data. Through this structure, the collection module (111) can secure customer information that exists dispersed as internal CRM data, external marketing channel data, or customer contact logs in an integrated form.

[0042] The normalization module (112) unifies the format of the collected customer information and verifies whether there is duplication or missing values ​​in the data, and is electrically connected to the collection module (111) and the conversion module (113).

[0043] The normalization module (112) converts customer information collected in different formats into a standardized data structure to maintain a consistent schema and can automatically detect and correct incomplete or error-containing data.

[0044] In addition, the normalization module (112) can perform data normalization processes such as string cleaning, standardization of date and numeric units, encoding conversion, and removal of unnecessary spaces, and ensures data quality that allows the lead scoring model described later to learn stably. Through this, the normalization module (112) plays a role in maintaining the integrity of customer information and improving the reliability of analysis and prediction results.

[0045] The conversion module (113) converts customer information, for which data normalization is completed, to the input format of the prediction unit (120) to extract customer characteristic information, and is electrically connected to the normalization module (112).

[0046] The conversion module (113) analyzes each attribute value of the normalized customer information and structures it into a form that can be used for learning by an artificial intelligence model, thereby converting the customer's behavioral patterns, transaction tendencies, and reaction characteristics into numerical feature values.

[0047] Customer characteristic information is a standardized feature value generated through preprocessing and conversion processes in the collection unit (110), and is data used as an input variable for an artificial intelligence prediction model in the prediction unit (120) to be described later.

[0048] Customer characteristic information may include numerical and behavioral indicators extracted from customer information, such as transaction frequency, purchase cycle, average transaction amount, proposal response rate, email open rate, click-through rate, product interest, time spent at each sales stage, customer retention rate, and contract ratio relative to proposal. This customer characteristic information is a value that quantifies and expresses each customer's behavioral patterns and transaction propensity, and is used as a key input variable for lead scoring and purchase conversion possibility analysis performed by the prediction unit (120).

[0049] In addition, the collection unit (110) can collect group information together and map it to customer characteristic information.

[0050] Here, group information refers to data used to group and manage sales leads and customers based on specific product or service units, and may include information such as the group name, linked product identifier, and group description.

[0051] The collection unit (110) can provide sales activity data of customers belonging to the same group in a structured form by normalizing the collected group information in conjunction with customer information. Through this, the prediction unit (120), which will be described later, can perform analysis at the group level and evaluate the conversion potential of customers by group.

[0052] Meanwhile, the collection unit (110) can search for related words such as product names, group names, and industry group names related to customers in documents, articles, and posts published on the web connected through an external linkage channel, and generate related word information by counting the frequency of occurrence of the words.

[0053] The collection unit (110) considers the product name and identifier registered in the product information, the group name of the group information, and the industry group notation of the company information as a dictionary of related words, and can search for a match within an external document based on this.

[0054] The collection module (111) can store the source, collection date and time, and document identification information of each document in which an associated word is detected as metadata, and can remove duplicates by counting only the first occurrence when the same associated word appears repeatedly from the same source.

[0055] Additionally, the collection module (111) can identify pages automatically generated during web crawling or simple list-type pages, and if the same string is structurally repeated, it can determine this as a duplicate and exclude it from the count. Through this, the collection module (111) can secure the frequency of occurrence and temporal distribution of related words by source in a highly reliable form.

[0056] Afterwards, the normalization module (112) can refine the collected string data to remove unnecessary symbols, HTML tags, special characters and spaces, and combine the same words repeated within the same document into one.

[0057] The normalization module (112) can distinguish between a complete match and a partial match and convert each data into a standardized data structure. Here, a complete match refers to a case where a term registered in the related word dictionary is included in the same form within a document, and a partial match may be limited to a case where a product name and an industry group name are included simultaneously within the document. In this case, if only a single partial match exists, it may be stored only as reference data and not reflected when generating related word information.

[0058] The conversion module (113) can analyze normalized data to convert the number of associated words aggregated per customer into numerical attributes and generate associated word information based on this.

[0059] At this time, the conversion module (113) can classify and store data into three intervals based on the occurrence time of the related word: the last 7 days, 8 to 30 days, and 31 to 90 days, and can manage the aggregation results for each interval separately.

[0060] The conversion module (113) can divide the most recent 7-day period into a priority reflection period, the 8 to 30-day period into an intermediate reflection period, and the 31 to 90-day period into a reference reflection period, and data exceeding 90 days can be moved to a separate storage area and excluded from the input target of the lead scoring model described later.

[0061] Additionally, the conversion module (113) can calculate the final number of associated words by merging data classified by source, period, and matching type after removing the repetition count of the same associated words detected in the same document.

[0062] The collection unit (110) can map the generated associated word information to customer information and add it as one item of customer characteristic information.

[0063] Accordingly, customer characteristic information includes not only internal data such as transaction frequency, average transaction amount, proposal response rate, email open rate, and click-through rate, but also related keyword information based on external web data, enabling the quantitative expression of customers' external exposure and market responsiveness.

[0064] The collection unit (110) performs the role of providing customer characteristic information in a form suitable for artificial intelligence analysis by refining unstructured data that exists in a scattered manner through the processes of collecting, normalizing, and converting customer information. Through this, the collection unit (110) increases the accuracy of lead scoring and customer prediction performed by the prediction unit (120), which will be described later, and enables the entire system to operate consistently based on data.

[0065] Next, the prediction unit (120) generates sales recommendation information by predicting potential customers whose lead score is above a preset threshold based on customer characteristic information received from the collection unit (110), and is electrically connected to the collection unit (110), content creation unit (130), feedback unit (150), and / or output unit (160).

[0066] The prediction unit (120) utilizes an artificial intelligence-based prediction model to analyze customer characteristic information refined by the collection unit (110), evaluates the likelihood of customer purchase conversion, and performs the function of identifying and prioritizing sales opportunities. Through this, the organization can structure vast amounts of customer data into an analyzable form and efficiently establish sales strategies and allocate resources based on the results.

[0067] The prediction unit (120) can calculate a lead score by quantifying the potential customer's purchase conversion probability through a lead scoring model that performs an evaluation operation applying weights to each attribute of customer information.

[0068] The lead scoring model is an artificial intelligence prediction algorithm based on supervised learning that takes various attribute values ​​included in customer characteristic information as input, applies weights to each attribute, and combines them to calculate the probability of a customer's purchase conversion.

[0069] Lead scoring models may utilize supervised learning-based logistic regression, gradient boosting, random forest, or deep neural network models, but are not limited to these; any artificial intelligence prediction algorithm capable of quantifying a customer's conversion potential can be used.

[0070] The lead scoring model represents variables included in customer characteristic information, such as transaction frequency, purchase cycle, average transaction amount, proposal response rate, email open rate, click-through rate, time spent at each sales stage, and / or contract-to-proposal ratio, as multidimensional vectors, and can calculate attribute-specific weights that reflect the contribution of each variable.

[0071] The lead scoring model performs an evaluation operation by multiplying each attribute value of customer characteristic information by an attribute-specific weight and accumulating the sum, and inputs this result into a probability output function to produce a lead score expressed as a real value between 0 and 1.

[0072] The lead score is a metric that numerically expresses a customer's likelihood of conversion, and the higher the value, the more likely the customer is to be judged to be converting.

[0073] For example, if customer A’s customer characteristic information is measured as an average monthly transaction frequency of 8 times, an average transaction amount of 4.2 million won, a proposal response rate of 76%, an email open rate of 82%, and a click-through rate of 54%, and there is a record of 2 meetings participated in within the last 2 weeks, the prediction unit (120) can calculate a high lead score by applying weights that reflect the high values ​​of each attribute. In this case, customer A can be classified as a promising lead with a high probability of purchase conversion, and the priority of sales activities can be set high.

[0074] Conversely, if customer B’s customer characteristic information is measured as having a monthly average transaction frequency of 1 time, an average transaction amount of 700,000 won, a proposal response rate of 18%, an email open rate of 21%, and a click-through rate of 5%, and there is no history of proposal response or meeting participation in the last 6 months, the prediction unit (120) can reflect weights by reflecting the low values ​​of each attribute, and a low lead score may be calculated. In this case, customer B may be classified as a general lead with a low probability of purchase conversion, and the priority of sales activities may be set to a low level.

[0075] Meanwhile, the lead scoring model can adjust attribute-specific weights based on associated word information included in customer characteristic information.

[0076] The lead scoring model includes associated word information as input variables along with internal attributes such as transaction frequency, average transaction amount, proposal response rate, email open rate, and click-through rate, and can calculate a lead score by combining the weights of each attribute.

[0077] In this case, the lead scoring model may apply different reflection ratios depending on the aggregation period of related word information. For example, the number of related words in the recent 7-day period may be set as the highest priority reflection period and reflected at 60% of the total reflection weight, the 8 to 30-day period may be set as the intermediate reflection period and reflected at 30% of the total reflection weight, and the 31 to 90-day period may be set as the reference reflection period and reflected at 10% of the total reflection weight. Additionally, data exceeding 90 days may be excluded from reflection.

[0078] Accordingly, customers whose number of associated words exceeds a certain standard in the past 7 days are judged to have high interest in the external market, and the weight of the corresponding attribute may be adjusted upward.

[0079] For example, if Customer A's associated word information is tallied as 6 exact matches and 3 partial matches in the last 7-day period, the lead scoring model determines that this corresponds to a high-impact period and may increase the weights of the transaction frequency, average transaction amount, and proposal response rate attributes by 20% each. On the other hand, if Customer B's associated word information is tallied as only 3 exact matches in the last 31 to 90-day period, it is classified as a reference period, and the weights of the same attributes may be increased by only 5%, or no adjustment may be applied if it falls below a certain threshold.

[0080] Additionally, the lead scoring model can adjust the reflection ratio of related word information according to the reliability of the source by referring to metadata stored in the collection module (111).

[0081] Related keywords collected from credible sources, such as official corporate websites, public institution publications, and press releases, may be reflected with a higher weight than general articles or community posts. Additionally, if data from different sources regarding the same customer is mixed, data collected from official sources is reflected first, and data from other sources may be applied within the remaining weight.

[0082] The lead scoring model can correct for quality deviations in external data and minimize weight distortion caused by noise through this reliability correction process.

[0083] In addition, the lead scoring model can distinguish the reflection strength based on the matching form of related word information.

[0084] Exact matches are reflected as is at a 1:1 ratio, and partial matches can be reflected at a 0.5x ratio only when both the product name and the industry group name are detected simultaneously within the same document.

[0085] As described above, the lead scoring model can calculate a lead score that simultaneously reflects the customer's external market interest and internal sales responsiveness by comprehensively considering the aggregation period, source reliability, matching type, and response history information, and finely adjusting the weighting of related word information.

[0086] Through this, customers with rich and recent related word information are placed at the top of sales recommendation information and classified as priority targets, while customers with insufficient or outdated related word information are classified as general leads and managed as monitoring targets.

[0087] In this way, the prediction unit (120) can precisely calculate each customer's lead score to be close to the actual sales potential by dynamically applying weights for each attribute while comprehensively considering the transaction patterns, reaction behaviors, and external environmental changes that differ for each customer.

[0088] The lead scoring model can classify the status of customers by applying pre-set thresholds based on the calculated lead score.

[0089] The threshold is a reference value for interpreting lead scores, representing the minimum predicted probability that a customer must meet to be determined as a promising lead. The threshold can be predefined based on the sales policies, target conversion rates, or the scale of available sales resources established by the organization.

[0090] For example, customers whose lead score is above a threshold are classified as promising leads and may be prioritized for sales activities. Conversely, customers whose lead score is below the threshold are classified as general leads or hold targets and may be designated for monitoring of future sales activities or for the collection of additional information.

[0091] The threshold can be set by considering the relationship between the distribution of scores generated by the lead scoring model, the characteristics of the customer group, and the conversion rate targeted by the organization. Setting a higher threshold reduces the number of customers classified as promising leads but increases prediction accuracy, while setting a lower threshold widens the customer range but may reduce the selectivity of the prediction. For example, if the threshold is set to 0.7, customers with a lead score of 0.7 or higher are considered promising leads, while those with a score below 0.7 may be classified as general or pending.

[0092] In addition, the threshold can be set as a fixed value and can be adjusted according to the distribution of lead scores or changes in the sales environment.

[0093] A lead scoring model can generate sales recommendation information by predicting potential customers whose calculated lead score is above a preset threshold.

[0094] In this case, customers with high lead scores are classified as priority response targets, while customers with low lead scores may be classified as targets for maintaining interest or future contact.

[0095] Sales recommendation information refers to information that quantitatively expresses a customer's conversion potential and product fit, based on lead scores calculated through a lead scoring model and the results of applying thresholds.

[0096] Sales referral information may include details such as the customer's lead score, referral probability, customer name, address, and industry, and customers with higher lead scores may be displayed higher.

[0097] In addition, sales recommendation information may include the results of a comprehensive analysis of the customer's transaction history, response tendencies, and sales stage information in the form of a summary sentence, and this summary sentence can be automatically generated based on the key attribute values ​​of the customer characteristic information analyzed by the lead scoring model.

[0098] For example, if a customer's recent proposal response rate is high, the average transaction amount is above a certain level, and the time spent at each sales stage is measured within the normal range, the customer is evaluated as having high product suitability and can be displayed with a high recommendation probability in sales recommendation information.

[0099] The prediction unit (120) can transmit the generated sales recommendation information to the output unit (160) to be described later for visualization, or transmit it to the content creation unit (130) to be used as reference data for creating customer-tailored proposals, quotations, or contracts.

[0100] Additionally, the prediction unit (120) can allow a sales representative to check recommendation results on a user interface, select a recommended customer to register as a sales lead, or exclude them from the recommendation list as needed.

[0101] As illustrated in FIG. 3, the prediction unit (120) can generate sales recommendation information for each customer based on the lead score calculated through the lead scoring model and the result of applying the threshold.

[0102] “Recommendation score 70%” is an example in which the prediction unit (120) converts the lead score calculated through the lead scoring model into a percentage and displays it, representing a result in which the likelihood of a customer’s purchase conversion is quantified.

[0103] In addition, the sentence presented at the bottom of the customer section is a summary automatically generated through sales recommendation information, which may reflect the customer's industry, market activity, transaction volume, response tendencies, and characteristics of the sales stage.

[0104] The prediction unit (120) can perform the role of quantitatively evaluating the purchase conversion potential of each customer and generating it in the form of lead scores and sales recommendation information by performing AI-based lead scoring and prediction calculations based on customer characteristic information provided by the collection unit (110). Through this, the prediction unit (120) objectively predicts the conversion potential of potential customers based on collected customer data, and enables the organization to establish efficient sales strategies and allocate resources in a data-centric manner.

[0105] Next, the content generation unit (130) generates sales content based on sales recommendation information received from the prediction unit (120) and is electrically connected to the prediction unit (120), the performance analysis unit (140) and / or the output unit (160).

[0106] The content generation unit (130) can automatically generate sales content suitable for each customer's sales situation based on sales recommendation information.

[0107] Here, sales content refers to data configured to correspond to the customer's status and sales stage based on sales recommendation information generated by the prediction unit (120), and specifically expresses the purpose, procedure, and items to be performed of sales activities.

[0108] Sales content automatically structures and displays the actual tasks that sales representatives must perform, thereby supporting activities tailored to each customer's sales stage to proceed according to consistent standards.

[0109] The content creation unit (130) can create sales content corresponding to each customer's status by reflecting the customer's conversion potential and product suitability included in the sales recommendation information.

[0110] For example, if the customer's status corresponds to the proposal stage, items for proceeding with the proposal can be automatically configured; if it corresponds to the quotation stage, items for verifying transaction conditions; and if it corresponds to the contract stage, items related to preparing for contract conclusion can be automatically configured.

[0111] In this way, the content creation unit (130) can improve the efficiency of sales activities by identifying the customer's sales stage through sales recommendation information and automatically configuring necessary items according to the purpose of the activity at each stage.

[0112] The content creation unit (130) can create suitable sales content according to the activity type specified by the user.

[0113] For example, as shown in Fig. 4, for customer meeting types, you can select the meeting format and meeting method, set the number of participants through the meeting attendee designation function, and check the registered schedule through the calendar screen.

[0114] In addition, for customer meeting types, automatic content generation for efficient meeting preparation can be performed, including features for setting meeting goals, defining success metrics, structuring anticipated questions and answers, providing checklists of necessary materials, and offering progress scenarios.

[0115] As illustrated in FIG. 5, for the mail sending type, the content creation unit (130) can automatically generate a mail draft based on a template associated with sales recommendation information provided by the prediction unit (120).

[0116] The selected template automatically includes items such as product introduction, proposal summary, and inquiry guidelines, and users can modify parts of the content as needed before sending.

[0117] As shown in Fig. 6, in the proposal sending type, a draft proposal can be automatically generated based on registered product information, and the generated proposal can be directly modified through a document editing tool.

[0118] Items for product configuration, application effects, expected performance upon implementation, and ROI (Return on Investment) analysis can be automatically inserted into the proposal, and the completed proposal can be sent to the customer via the email sending function.

[0119] After sending the proposal, the content creation unit (130) can automatically record the status of the activity and generate an activity summary to link the customer's response and subsequent sales activities.

[0120] As illustrated in Fig. 7, in the quotation sending type, a draft quotation can be generated based on registered product and transaction information. The generated quotation includes automatically calculated unit prices, quantities, and total sums for each item, and a price adjustment proposal can be automatically inserted depending on the customer's budget range or purchase scale.

[0121] After sending the quotation, the content creation unit (130) can provide reference data for establishing a future response strategy by summarizing and displaying the results of sales activities related to the quotation.

[0122] As illustrated in Fig. 8, a draft contract can be automatically generated in the contract sending type. Product provision conditions, service scope, contract period, cost conditions, and legal clauses can be automatically inserted into the contract, and the generated contract can be modified directly by the user or supplemented with questions through an AI question-and-answer function.

[0123] In addition, the content creation unit (130) can display a list of recommended questions regarding contract-related inquiry items so that they can be referenced during the contract condition review process.

[0124] As illustrated in Fig. 9, in the outbound call type, a call script that a sales representative can refer to before speaking with a customer can be automatically generated.

[0125] The script includes the customer's lead score, offer history, and key product items of interest, which can support sales representatives in conducting conversations efficiently.

[0126] For other types, it can be used as a general item to record tasks that do not correspond to specific business activity statuses or to manage additional communication activities.

[0127] The content creation unit (130) can automate all stages of the sales process under a consistent data structure by automatically generating documents such as proposals, quotations, and contracts according to the type of sales activity for each customer based on the sales recommendation information calculated by the prediction unit (120), and by managing the progress status of each activity.

[0128] Through this, the content creation unit (130) enables sales representatives to avoid having to manually perform repetitive document creation or activity recording, and allows the organization to understand sales situations in real time and establish efficient response strategies.

[0129] Next, the performance analysis unit (140) generates sales performance information based on the results of utilizing sales content and is electrically connected to the content generation unit (130) and / or output unit (160).

[0130] The performance analysis unit (140) can generate sales performance information by analyzing the results of sales activities performed through sales content received from the content creation unit (130).

[0131] Here, sales performance information refers to information derived by analyzing the results of the actual sales process in which sales content generated by the content creation unit (130) is utilized, and represents an indicator for evaluating how much performance the organization's sales activities have achieved.

[0132] Sales performance information is data that quantifies the results of activities performed based on sales recommendation information, and can be used as a basis for objectively evaluating the efficiency of work performed by sales representatives, customer responsiveness, and conversion rates.

[0133] The performance analysis unit (140) can calculate performance for each customer or sales activity unit by referring together to the usage history of sales content transmitted from the content creation unit (130) and sales recommendation information provided by the prediction unit (120).

[0134] For example, by analyzing the rate at which actual customers view or respond to a message after a sales representative has sent a proposal or email, performance indicators such as delivery success rate, response rate, and open rate can be derived.

[0135] These analysis results serve as key data for real-time assessment of how effectively an organization's sales strategy is being executed, and can also be utilized to modify future sales policies or adjust priorities.

[0136] As illustrated in Fig. 10, sending refers to the number of times sales content is transmitted to actual customers, success refers to the number of times the sending was successfully completed, and failure refers to the number of times an error occurred during the transmission process.

[0137] Additionally, "Open" indicates the number of times a customer has actually viewed the sent email, and "Last Sent Date" refers to the time when sales content was last delivered to that customer.

[0138] In this way, the performance analysis unit (140) can quantitatively analyze the effectiveness of sales activities based on whether sales content is sent, responded to, or viewed, and can structure and display the results as sales performance information.

[0139] The performance analysis unit (140) transmits the generated sales performance information to the feedback unit (150) described later and feeds it back to the prediction unit (120), thereby allowing it to be used as training data for the lead scoring model or improving the accuracy of sales recommendation information.

[0140] In this way, the performance analysis department (140) systematically analyzes the results of the use of sales content to support the organization in performing data-based performance evaluation and strategy formulation.

[0141] Meanwhile, an AI-based customer data analysis, lead scoring, and automatic content generation system (100) according to one embodiment of the present invention may further include a feedback unit (150) that retrains a lead scoring model based on sales performance information and updates the lead score.

[0142] The feedback unit (150) can correct the prediction accuracy by comparing the prediction result performed by the lead scoring model of the prediction unit (120) with the result of the actual sales activity based on the sales performance information received from the performance analysis unit (140), and is electrically connected to the prediction unit (120) and / or the performance analysis unit (140).

[0143] The feedback unit (150) can improve the prediction accuracy of the model by evaluating the reliability of the lead score calculated by the prediction unit (120) based on sales performance information and readjusting the weights of the attributes of the lead scoring model to minimize the difference between the prediction and the actual result.

[0144] The feedback unit (150) applies the lead scoring model updated through retraining to the prediction unit (120), and the prediction unit (120) can recalculate the lead score for customer characteristic information based on the model.

[0145] The lead score updated by the feedback unit (150) maintains the same calculation criteria while reflecting the latest sales data, so the organization can immediately respond to customer reactions and market conditions that change over time. Accordingly, the feedback unit (150) can maintain consistency and reliability of lead scoring by cyclically correcting the deviation between the predicted results and actual sales performance.

[0146] Additionally, the feedback unit (150) links the updated lead score and prediction results to the prediction unit (120) and the output unit (160) so that the entire system operates based on the latest data. The feedback unit (150) can control the relearning cycle and update frequency by periodically collecting and reflecting sales performance information continuously provided by the performance analysis unit (140), thereby preventing overfitting or performance degradation of the lead scoring model.

[0147] In this way, the feedback unit (150) directly utilizes sales performance information to periodically update the lead scoring model and maintains the lead score calculated by the prediction unit (120) in an up-to-date state, thereby enabling the organization to continuously perform data-based precise predictions and decisions even in a sales environment that changes in real time.

[0148] Next, the output unit (160) visualizes and outputs sales recommendation information and sales performance information in the form of a dashboard, and is electrically connected to the prediction unit (120), content creation unit (130), performance analysis unit (140) and / or feedback unit (150).

[0149] The output unit (160) integrates sales data generated from the entire system and displays it on a single screen, thereby allowing the user to intuitively check the organization's sales status and performance.

[0150] The output unit (160) can visualize sales recommendation information to display prediction results such as lead scores, conversion potential, and product suitability for each customer, and can output sales performance information calculated by the performance analysis unit (140) by classifying it into items such as key indicators, sales status, sales efficiency analysis and / or time required for each stage.

[0151] The main indicator screen of the output unit (160) is configured to display key indicators such as total sales, transaction amount, number of leads, contract success rate, and average time taken per lead, so that the organization's sales efficiency can be grasped at a glance. Each indicator can be updated in real time, and the sales volume, contract rate, and customer acquisition status can be comprehensively compared and analyzed.

[0152] In addition, as illustrated in FIG. 11, the output unit (160) visualizes and displays the actual amount achieved relative to the target on an annual, quarterly, or monthly basis through the overall sales screen, and outputs the difference between the achievement rate and the target amount in the form of a graph, thereby allowing the trend of sales performance to be intuitively grasped. Through this, the user can monitor sales trends by period and perform strategic adjustments based on the target achievement rate.

[0153] Additionally, as illustrated in FIG. 12, the output unit (160) may include a sales efficiency analysis function and may divide the progress stages of sales activities into seven stages: lead acquisition, initial contact, needs identification, proposal, negotiation, contract, and termination, and visually output the conversion rate for each stage. Through this, the user can identify where bottlenecks occur at each stage and derive directions for improving the sales process.

[0154] Additionally, the output unit (160) provides a screen showing the time required for each stage, which can output the average period required to transition between each sales stage on a monthly basis. By comparing the time required for each stage with the previous month or quarter, the user can analyze the speed of sales progress and set management standards for improving sales efficiency.

[0155] In this way, the output unit (160) integrates and visualizes key indicators, sales status, efficiency analysis, and time required information, thereby supporting the organization to monitor sales status in real time and perform data-based performance analysis and decision-making.

[0156] Meanwhile, although not shown in the drawing, an AI-based customer data analysis, lead scoring, and automatic content generation system (100) according to one embodiment of the present invention may further include a strategy generation unit that collects product information and generates sales strategy information, which is information about sales methods and sales strategies for each product.

[0157] The strategy generation unit can analyze product information collected through the product management screen, such as product names, product files, service guides, proposals, quotations, and contracts, and generate sales strategy information to determine sales directions for each product.

[0158] The Strategy Generation Department analyzes data related to key product attributes, target markets, pricing policies, competitive factors, and introduction effects to derive sales directions and strategies. Through this process, the department models the relationship between each product's strengths and customer needs, and establishes differentiated sales approaches and priorities for each product.

[0159] The strategy generation unit can configure sales strategy information in a format that sales representatives can refer to, based on the analysis results.

[0160] Sales strategy information may consist of data regarding product-specific value propositions, target customer characteristics, market entry strategies, message direction, price negotiation criteria, and proposal directions, and the components or methods of expression may be varied depending on the product information. In this process, the strategy generation unit may reflect the lead score and customer characteristic information calculated by the forecasting unit (120) together to concretize the strategy centered on the customer group with the highest conversion potential for each product.

[0161] For example, the strategy creation unit can analyze registered product information to generate product summaries, sales playbooks, and / or proposal templates.

[0162] The Strategy Generation Unit can analyze product-related documents registered on the product management screen, such as product names, service guides, proposals, quotations, and / or contracts, to extract key values, customer segments, competitive factors, and pricing strategies for each product. These analysis results are structured into a product summary and can be utilized as sales strategy data that concisely expresses the product's key characteristics, differentiation points, and implementation benefits.

[0163] In addition, the strategy generation unit can construct sales playbooks by considering the conversation flow at customer touchpoints and the stages of sales activities. The sales playbook includes content such as customer approach scenarios, step-by-step messaging strategies, anticipated questions and answers, competitor response logic, and price negotiation guides, allowing sales representatives to utilize it as a strategic resource for reference during the actual sales process. Such sales playbooks can be automatically customized and generated based on product characteristics and customer group response patterns.

[0164] In addition, the Strategy Creation Department can generate proposal templates that can be utilized during the proposal activity phase. These proposal templates include customer-tailored components, value proposition stories, ROI calculation models, relevant industry case studies, and post-implementation action plans, and can serve as foundational data when drafting actual proposals or sales proposal documents.

[0165] The Strategy Creation Department supports the organization's sales representatives in clearly identifying the strategic strengths of each product and rapidly generating appropriate response content based on customer needs and market conditions.

[0166] In addition, the strategy generation unit can automatically update sales strategy information based on registered data to keep it up to date when changes occur, such as updates to product information, changes in pricing policies, or entry into new markets.

[0167] Sales strategy information can be visualized and displayed through the output unit (160), or linked with the results of the content creation unit (130) and used as basic data for creating proposals and sales documents.

[0168] In this way, the Strategy Generation Unit integrates and analyzes registered product information and forecast results, and converts them into strategic content optimized for each product, thereby enabling the organization to formulate data-driven sales strategies and establish an efficient sales execution system.

[0169] As described above, according to the AI-based customer data analysis, lead scoring, and automatic content generation system (100) according to one embodiment of the present invention, by integrating and analyzing customer information, product information, and sales activity data distributed within an organization based on artificial intelligence, customer characteristics and purchase conversion potential can be quantitatively predicted, and based on this, sales strategies can be established and sales documents such as proposals, quotations, and contracts can be automatically generated, thereby improving the efficiency of the entire sales process and ensuring objectivity in decision-making.

[0170] In addition, by collecting the utilization results of generated sales content in real time to quantitatively evaluate sales activity performance and reflecting them in the training of the prediction model, the accuracy of sales forecasts and the precision of strategy recommendations can be continuously improved over time.

[0172] Although all components constituting the embodiments of the present invention have been described above as being combined or operating in combination, the present invention is not necessarily limited to such embodiments. That is, within the scope of the purpose of the present invention, all components may be selectively combined in one or more ways to operate.

[0173] Furthermore, terms such as "include," "compose," or "have" described above, unless specifically stated otherwise, mean that the relevant component may be inherent; therefore, they should be interpreted as allowing for the inclusion of additional components rather than excluding them. All terms, including technical or scientific terms, have the same meaning as generally understood by those skilled in the art to which the present invention pertains, unless otherwise defined. Commonly used terms, such as those defined in advance, should be interpreted in accordance with their meaning in the context of the relevant technology and should not be interpreted in an ideal or overly formal sense unless explicitly defined in the present invention.

[0174] Furthermore, the above description is merely an illustrative explanation of the technical concept of the present invention, and those skilled in the art to which the present invention pertains will be able to make various modifications and variations within the scope of the essential characteristics of the present invention.

[0175] Accordingly, the embodiments disclosed in this invention are intended to illustrate, not limit, the technical concept of the invention, and the scope of the technical concept of the invention is not limited by these embodiments. The scope of protection of this invention shall be interpreted by the claims below, and all technical concepts within an equivalent scope shall be interpreted as being included within the scope of rights of this invention. Explanation of the symbols

[0176] 100: AI-based customer data analysis, lead scoring, and automatic content generation system according to an embodiment of the present invention 110: Collection Department 111: Collection Module 112: Normalization Module 113: Conversion Module 120: Prediction section 130: Content Creation Section 140: Performance Analysis Department 150: Feedback Department 160: Output section

Claims

Claim 1 A collection unit that collects and preprocesses customer information to extract customer characteristic information; a prediction unit that predicts potential customers whose lead score calculated based on the customer characteristic information is above a preset threshold and generates sales recommendation information; a content generation unit that generates sales content based on the sales recommendation information; and a performance analysis unit that generates sales performance information based on the results of utilizing the sales content.The system includes an output unit that visualizes and outputs the sales recommendation information and the sales performance information in the form of a dashboard, wherein the sales performance information includes performance indicators derived by analyzing the sending success rate, response rate, and open rate of the sales content sent by a sales representative to a customer for each customer or sales activity unit, and the collection unit includes a collection module that collects the customer information, a normalization module that unifies the format of the customer information and verifies duplicates and missing values, and a conversion module that converts the customer information to fit the input format of the prediction unit to extract the customer characteristic information, wherein the collection unit searches for related words including product names, group names, and industry names related to the customer in documents, articles, and posts published on the web connected through an external linkage channel, generates related word information by aggregating the frequency of occurrence of the related words, maps the related word information to the customer information, and adds it to the customer characteristic information, and the prediction unit calculates the lead score by quantifying the potential customer's purchase conversion probability through a lead scoring model that performs an evaluation operation applying weights to attributes of the customer information, and the prediction unit corrects the weights based on the related word information, wherein the aggregation period of the related word information, the related words An AI-based customer data analysis, lead scoring, and automatic content generation system that adjusts the ratio at which the associated word information is reflected in the weight according to the reliability of the collected source and the matching form of the associated word, wherein the collection unit counts the occurrence of the associated word only once when the associated word is repeatedly detected from the same source, and excludes from the aggregation target of the associated word when the same string is structurally repeated on a page automatically generated during web crawling or a simple list-type page. Claim 2 delete Claim 3 The AI-based customer data analysis, lead scoring, and automatic content generation system of claim 1 further comprises a feedback unit that retrains the lead scoring model based on the sales performance information and updates the lead score. Claim 4 delete Claim 5 The AI-based customer data analysis, lead scoring, and automatic content generation system of claim 1 further comprises a strategy generation unit that collects product information and generates sales strategy information, which is information regarding sales methods and sales strategies for each product.

Images