Influencer automatic matching system for marketing

Abstract

The present invention relates to an automatic influencer matching system for marketing, and more specifically, to an automatic influencer matching system and method for marketing that maximizes the overall efficiency and reliability of a marketing campaign by organically linking online channels such as influencer marketing and offline channels such as digital outdoor advertising, analyzing the performance of each channel in real time to dynamically optimize the advertising budget, and automatically evaluating and selecting advertising content and advertising slots to be used in the campaign based on data.

Description

Technology Field

[0001] The present invention relates to an automatic influencer matching system for marketing, and more specifically, to an automatic influencer matching system and method for marketing that maximizes the overall efficiency and reliability of a marketing campaign by organically linking online channels such as influencer marketing and offline channels such as digital outdoor advertising, analyzing the performance of each channel in real time to dynamically optimize the advertising budget, and automatically evaluating and selecting advertising content and advertising slots to be used in the campaign based on data. Background Technology

[0002] Recently, corporate marketing activities have been evolving into an integrated form that combines online channels utilizing social media with offline channels using outdoor electronic billboards. Conventional marketing systems have attempted to manage these online and offline channels in an integrated manner.

[0003] For example, Korean Registered Patent No. 10-2594812 discloses a technology for conducting online or offline marketing by identifying regional demand for a seller's products. The aforementioned prior art focuses on fixedly 'determining' a specific region as an online or offline target based on historical static data, such as regional sales data, and 'calculating' the contribution of influencers afterward.

[0004] However, these conventional technologies have the following obvious limitations.

[0005] A static channel selection method based on historical data has the drawback of failing to proactively respond to real-time market changes or fluctuations in channel efficiency that occur during a campaign. For example, even if a specific influencer's content suddenly goes viral and the efficiency of an online channel skyrockets, the fixed budget allocation cannot be changed, resulting in opportunity costs.

[0006] There is a problem in that while online and offline channels are simply executed in parallel, there is a lack of specific means to quantitatively measure the organic interaction between the two channels—specifically, the synergistic effect of offline advertising on online search volume or brand awareness—and to reflect this in campaign strategies.

[0007] There is a problem in that the focus is solely on selecting influencers or advertising locations, while lacking the functionality to pre-evaluate and optimize the quality of the advertising 'content' itself—which has a decisive impact on actual marketing performance—or the value of each of the numerous offline advertising 'slots' based on objective data, leading to reliance on the marketer's subjective judgment.

[0008] Therefore, there is an urgent need for the development of new technology that can fundamentally improve the efficiency and predictability of marketing campaign operations by comprehensively providing dynamic resource allocation based on real-time data, quantitative analysis of synergy effects between channels, and objective valuation functions for content and ad slots themselves. The problem to be solved

[0009] The first objective of the present invention is to provide an integrated online-offline marketing device and method that maximizes the efficiency of marketing investment by dynamically redistributing the advertising budget in response to the performance of each channel changing in real time during a campaign, thereby solving the problems of the prior art as described above.

[0010] The second objective of the present invention is to provide an apparatus and method that quantitatively measures the synergistic effect of offline advertising on online channels and reflects this in the budget allocation logic, thereby organically linking online and offline channels and fully utilizing the potential value of integrated marketing.

[0011] The third objective of the present invention is to provide an apparatus and method that enhance the predictability and reliability of marketing performance by pre-evaluating the channel suitability of the advertising 'content' itself and the dynamic value of each 'offline advertising slot' based on objective data and sophisticated computational logic, and thereby automatically selecting the most effective creative and advertising location.

[0012] The problems of the present invention are not limited to those mentioned above, and other problems not mentioned will be clearly understood by those skilled in the art from the description below. means of solving the problem

[0014] In a device for performing integrated online and offline marketing according to an embodiment of the present invention for solving the above problem,

[0015] An influencer information management module that stores and manages information on multiple influencers;

[0016] Campaign settings module for setting the goals and targets of a marketing campaign;

[0017] An influencer matching module that recommends and selects the most suitable influencer among the influencers stored in the influencer information management module based on information input from the campaign setting module;

[0018] An advertising integration module that provides an interface for linking with multiple online and offline advertising media;

[0019] A performance analysis module that collects and analyzes performance data of advertisements transmitted through the above advertising media in real time; and

[0020] An integrated ad delivery module that dynamically redistributes the budget between online channels and offline channels based on real-time performance data collected from the above performance analysis module;

[0021] Includes,

[0022] The above-described integrated advertising delivery module is characterized by executing a first operation logic to calculate the relative ratio of the value per unit cost of an online channel and the value per unit cost of an offline channel to calculate a real-time channel efficiency index, and if the calculated real-time channel efficiency index exceeds a preset first threshold, redistributing the budget of the offline channel to the online channel, and if it is less than a second threshold, redistributing the budget of the online channel to the offline channel.

[0023] The above first operation logic is,

[0024] The method further includes a step of multiplying the above-calculated relative ratio by the value obtained by adding the inter-channel synergy coefficient, and

[0025] The above-mentioned synergy coefficient between channels is characterized by being calculated based on the growth rate of online search volume of campaign-related keywords that occurred while offline advertising was being executed.

[0026] The above first operation logic is,

[0027] It further includes a step of calculating a final real-time channel efficiency index by multiplying the value reflecting the above synergy effect by a periodic volatility coefficient calculated based on the current time.

[0028] The above periodic variability coefficient is characterized by being calculated based on the difference between the current time and a preset maximum efficiency time phase using a cosine function with a period of 24 hours, and

[0029] The value per unit cost of the above online channel is,

[0030] The real-time conversion value of an online channel, including the number of purchase conversions and cart additions generated through online advertising, is calculated by dividing the real-time cost executed on the said online channel.

[0031] The value per unit cost of the above offline channel is,

[0032] The real-time conversion value of an offline channel, including the number of QR code scans and campaign-exclusive shortened URL clicks by users exposed to offline advertisements, is calculated by dividing the real-time cost executed on the said offline channel.

[0033] It further includes a content management module for registering and managing ad creatives to be used in campaigns, and

[0034] The above influencer matching module is,

[0035] It is characterized by performing primary filtering based on the degree of match between the demographic information of target customers input from the campaign setting module and the demographic information of influencer followers stored in the influencer information management module.

[0036] The above performance analysis module is,

[0037] It is characterized by integrating all collected performance data after the campaign ends to generate a training dataset to improve the prediction accuracy of future campaigns, and

[0038] The above influencer information management module is,

[0039] It may be characterized by periodically collecting and updating the number of followers, posts, engagement rate, and follower demographic information of the influencer by linking with the API of a social media platform. Effects of the invention

[0041] According to the present invention, by calculating a real-time channel efficiency index and dynamically redistributing the advertising budget based thereon, it is possible to respond quickly to unexpected market changes or fluctuations in the performance of specific channels that occur during the campaign. This minimizes the opportunity cost incurred from static budget allocation methods and has the effect of significantly improving the return on investment (ROI) of the entire campaign by always investing the limited budget intensively in the most efficient places.

[0042] By introducing an inter-channel synergy coefficient that measures the impact of offline advertising on online search volume, it is possible to quantitatively capture positive inter-channel interactions—which were overlooked in conventional technologies—and reflect them in campaign optimization. This results in maximizing the effectiveness of integrated marketing by managing online and offline channels as an organic ecosystem rather than a simple parallel relationship.

[0043] Through sophisticated computational logic that calculates ad content suitability scores and offline ad slot value scores, it transforms content and ad placement selection—which previously relied on marketers' subjectivity or intuition—into data-driven, automated decision-making. This effectively guarantees the reliability and reproducibility of marketing activities by predicting campaign success probabilities in advance and deriving the most efficient execution plans.

[0044] Through a virtuous cycle structure that utilizes performance data collected throughout the entire campaign as training data for the next campaign, the system learns and evolves autonomously, continuously improving the accuracy of predictions and recommendations over time. This provides the effect of systematically accumulating and capitalizing on the marketing organization's experience and know-how.

[0045] The effects according to the present invention are not limited to those exemplified above, and a wider variety of effects are included within the present invention. Brief explanation of the drawing

[0047] Figure 1 illustrates an overall relationship diagram according to the present invention. Figure 2 illustrates an overall flowchart according to the present invention. FIG. 3 illustrates a step-by-step flowchart of the first control logic according to the present invention. FIG. 4 illustrates a step-by-step flowchart of the second control logic according to the present invention. FIG. 5 illustrates a step-by-step flowchart of the third control logic according to the present invention. FIG. 6 is a configuration diagram showing the interaction between a system and an external environment according to one embodiment of the present invention. FIG. 7 is a diagram showing the data flow between the main component modules of the present invention. FIG. 8 is a diagram showing an example of a user interface provided by the campaign setting module of the present invention. FIG. 9 is a diagram showing an example of an integrated dashboard provided by the performance analysis module of the present invention. FIG. 10 is a diagram illustrating the conceptual operation of the first operation logic. FIG. 11 is a diagram illustrating the conceptual operation of the second operation logic. FIG. 12 is a diagram illustrating the conceptual operation of the third operation logic. FIG. 13 is a conceptual diagram illustrating the effect of dynamic budget redistribution according to the present invention. FIG. 14 is an illustration for explaining the concept of calculating the inter-channel synergy coefficient of the present invention. FIG. 15 is a conceptual diagram illustrating the system learning and optimization virtuous cycle structure of the present invention. FIG. 16 is a hardware and network configuration diagram of a system for implementing the present invention. FIG. 17 is an example of a building advertising service according to the present invention. FIG. 18 is an example of a shelter advertising service according to the present invention. FIG. 19 is an example of a subway advertising service according to the present invention. FIG. 20 is an example of a platform advertising service according to the present invention. Specific details for implementing the invention

[0048] Hereinafter, various embodiments are described in more detail with reference to the attached drawings. The embodiments described in this specification may be modified in various ways. Specific embodiments may be depicted in the drawings and described in detail in the detailed description. However, specific embodiments disclosed in the attached drawings are intended only to facilitate understanding of various embodiments. Accordingly, the technical concept is not limited by specific embodiments disclosed in the attached drawings, and it should be understood that it includes all equivalents or substitutions that fall within the spirit and scope of the invention.

[0049] Terms including ordinal numbers, such as first, second, etc., may be used to describe various components, but these components are not limited by the aforementioned terms. The aforementioned terms are used solely for the purpose of distinguishing one component from another.

[0050] Functions related to artificial intelligence according to the present disclosure are operated through a processor and memory. The processor may be composed of one or more processors. In this case, the one or more processors may be general-purpose processors such as CPUs, APs, and DSPs (Digital Signal Processors), graphics-dedicated processors such as GPUs and VPUs (Vision Processing Units), or artificial intelligence-dedicated processors such as NPUs. The one or more processors control the processing of input data according to predefined operation rules or artificial intelligence models stored in memory. Alternatively, if the one or more processors are artificial intelligence-dedicated processors, the artificial intelligence-dedicated processors may be designed with a hardware structure specialized for processing a specific artificial intelligence model.

[0051] The predefined rules of operation or artificial intelligence models are characterized by being created through learning. Here, being created through learning means that a basic artificial intelligence model is trained using a number of training data by a learning algorithm, thereby creating predefined rules of operation or artificial intelligence models configured to perform desired characteristics (or objectives). Such learning may be performed on the device itself where the artificial intelligence according to the present disclosure is executed, or it may be performed through a separate server and / or system. Examples of learning algorithms include supervised learning, unsupervised learning, semi-supervised learning, or reinforcement learning, but are not limited to the examples described above.

[0052] An artificial intelligence model can be composed of multiple neural network layers. Each of the multiple neural network layers has multiple nodes and weight values, and performs neural network operations through calculations between the results of previous layers and the multiple weights. The multiple weights possessed by the multiple neural network layers can be optimized based on the learning results of the artificial intelligence model. For example, multiple weights can be updated so that the loss value or cost value obtained by the artificial intelligence model during the learning process is reduced or minimized. Additionally, to minimize the loss value or cost value, multiple weights can be updated in a direction that minimizes the gradient associated with the loss value or cost value. Artificial neural networks may include deep neural networks (DNNs), such as Convolutional Neural Networks (CNNs), Deep Neural Networks (DNNs), Recurrent Neural Networks (RNNs), Restricted Boltzmann Machines (RBMs), Deep Belief Networks (DBNs), Bidirectional Recurrent Deep Neural Networks (BRDNNs), or Deep Q-Networks, but are not limited to the examples mentioned above.

[0053] A network is a network that serves as a transmission path for web pages; it may be a closed network such as a LAN (Local Area Network) or WAN (Wide Area Network), but it is desirable for it to be an open network such as the Internet. The Internet refers to a global open computer network structure that provides the TCP / IP protocol and various services existing at its upper layers, namely HTTP (HyperText Transfer Protocol), Telnet, FTP (File Transfer Protocol), DNS (Domain Name System), SMTP (Simple Mail Transfer Protocol), SNMP (Simple Network Management Protocol), NFS (Network File Service), and NIS (Network Information Service).

[0054] Terminals can be implemented in various forms. For example, the terminals described in this specification may include mobile terminals such as smartphones, tablet PCs, PDAs, portable multimedia players, and MP3 players, as well as fixed terminals such as smart TVs and desktop computers.

[0055] In this specification, terms such as “comprising” or “having” are intended to specify the existence of the features, numbers, steps, actions, components, parts, or combinations thereof described in the specification, and should be understood as not precluding the existence or addition of one or more other features, numbers, steps, actions, components, parts, or combinations thereof. When a component is described as being “connected” or “connected” to another component, it should be understood that it may be directly connected to or connected to that other component, or that there may be other components in between. On the other hand, when a component is described as being “directly connected” or “directly connected” to another component, it should be understood that there are no other components in between.

[0056] Meanwhile, a "module" or "part" for a component as used in this specification performs at least one function or operation. Furthermore, a "module" or "part" may perform a function or operation by hardware, software, or a combination of hardware and software. Additionally, a plurality of "modules" or a plurality of "parts," excluding a "module" or "part" that must be performed on specific hardware or on at least one processor, may be integrated into at least one module. A singular expression includes a plural expression unless the context clearly indicates otherwise.

[0057] In addition, power, power transmission, and control therefor for the following assembly configurations and embodiments, including "by control," follow conventional technology including terminals, applications, hardware control modules, etc., so they are omitted to avoid redundancy.

[0058] In addition, the operation embodiments and configurations described in a general manner without being explained in detail below follow the prior art and are omitted in order to focus on describing the purpose of the present invention and the resulting effects.

[0059] Furthermore, in describing the present invention, if it is determined that a detailed description of related known functions or configurations may unnecessarily obscure the essence of the invention, such detailed description is abbreviated or omitted.

[0060] Hereinafter, the specific functions of each module constituting the online-offline integrated marketing system according to the present invention and embodiments of organic operation between them will be described in detail. The system largely comprises an influencer information management module (100), a campaign setting module (200), an influencer matching module (300), a content management module (400), an advertising linkage module (500), an integrated advertising transmission module (600), and a performance analysis module (700).

[0061] The above-mentioned influencer information management module (100) is a component that collects, stores, updates, and manages quantitative and qualitative data regarding multiple influencers that can be utilized in marketing campaigns. It serves as a database that provides basic data for effective influencer matching.

[0062] In addition, by linking with APIs of major social media platforms such as YouTube and Instagram, the number of followers, posts, engagement rates (likes, comments, shares), and demographic information of followers (age, gender, region) of registered influencers are collected periodically (e.g., once a day).

[0063] Data Cleaning and Processing: Process the collected raw data to calculate meaningful metrics such as average engagement rate, follower growth rate, and major content categories (e.g., 'Beauty', 'Games').

[0064] In addition, a unique ID is assigned to each influencer, and the collected and processed data is stored and updated in a database in a structured form.

[0065] In addition, receive channel information for a specific influencer 'A' from the YouTube API in JSON format: {"subscriberCount": 150000, "videoCount": 250, "viewCount": 30000000}.

[0066] Additionally, when a request for a list of influencers with the condition of '100,000 or more followers, beauty category' is received from the influencer matching module (300), the database is queried to obtain a list of influencer IDs ['ID_A', 'ID_C', 'ID_F'] that meet the condition and key indicator data for each influencer are transmitted to the influencer matching module (300).

[0067] The above campaign setting module (200) is a component that provides a user interface (UI) and processing functions for an advertiser or marketing manager to input and set the basic strategy of a new marketing campaign, such as the goal, budget, period, target customers, and channels to be used.

[0068] In addition, through a web-based dashboard, input is received for the campaign name, total budget (e.g., 30 million won), campaign period (e.g., 2025-01-10 ~ 2025-01-24), target audience (e.g., 'women in their 20s and 30s, residing in Seoul'), target performance (e.g., 'number of website sign-ups'), and main channels (online / offline).

[0069] In addition, it verifies whether the entered budget is in numerical format and whether the period setting is logically correct.

[0070] In addition, based on the input information, a new campaign object with a unique campaign ID is created so that it can be shared throughout the entire system.

[0071] In addition, the campaign settings entered by the user on the web screen are received in the body of the HTTP POST request: {"campaignName": "New Year Cosmetics Discount", "budget": 30000000, "targetAge": "20-39", ...}.

[0072] In addition, information about the generated campaign object, particularly 'target target' and 'campaign category' information, is transmitted to the influencer matching module (300), content management module (400), and ad integration module (500) to serve as a standard for subsequent operations.

[0073] The above-mentioned influencer matching module (300) is a component that finds, recommends, and selects the optimal influencer that best matches the campaign goal and target information received from the campaign setting module (200) from the database of the influencer information management module (100).

[0074] In addition, a pool of influencer candidates is selected where the degree of agreement between the campaign target (age, gender, region) and the influencer's follower demographic data is above a specific threshold (e.g., 60%).

[0075] In addition, the pool of candidates is narrowed down by evaluating the alignment between the campaign category (e.g., 'Cosmetics') and the influencer's primary content category, as well as past campaign history.

[0076] In addition, it synthesizes factors such as estimated advertising costs and average engagement rates from the compressed pool of candidates to generate and present to the manager the final recommendation list most suitable for the campaign objectives.

[0077] In addition, information {"targetAge": "20-39", "targetGender": "female", "category": "cosmetics"} is received from the campaign setting module (200).

[0078] In addition, the final selected influencer ID list ['ID_A', 'ID_C'] and their estimated unit price information are transmitted to the campaign setting module (200) to be reflected in the budget plan, and are transmitted to the content management module (400) to generate customized content guidelines.

[0079] The above-mentioned content management module (400) is a component that registers, manages, evaluates, and optimizes all advertising materials (images, videos, text, etc.) to be used in a campaign. This includes the second operation logic of the present invention.

[0080] In addition, it stores multiple ad creative candidates uploaded by the administrator and manages them by version.

[0081] In addition, for each registered creative, a second calculation logic is executed based on the target channel (e.g., Instagram feed, subway screen advertisement) to calculate the ad content suitability score.

[0082] In addition, Optimal Content Recommendation: Recommends the top creatives with the highest calculated ad content suitability scores as A / B testing candidates or automatically selects them as the final creatives for execution.

[0083] In addition, the selected material is automatically resized or formatted to fit the technical specifications of each channel (e.g., 1:1 ratio, 16:9 ratio).

[0084] In addition, five image files and advertising text are received from the administrator. A list of channels to be broadcast (['instagram_feed', 'subway_screen']) is received from the advertising integration module (500).

[0085] In addition, the content ID that received the highest ad content suitability score for each channel (e.g., 'IMG_03_v2') and the converted file (e.g., 'IMG_03_v2_1x1.jpg', 'IMG_03_v2_16x9.mp4') are transmitted to the integrated ad delivery module (600).

[0086] The above-mentioned advertising linkage module (500) is a component that technically links with various external online and offline advertising platforms (e.g., meta advertising API, digital outdoor advertising reservation system) and performs the role of an interface for querying, reserving, and controlling advertising slot information. This includes the third operation logic of the present invention.

[0087] In addition, information on all available offline ad slots within the campaign target area (location, time, cost, competition status, etc.) is collected in real time via external platform APIs.

[0088] In addition, a third operation logic is executed for each collected slot information to calculate the dynamic value evaluation score of the ad slot.

[0089] In addition, by considering the calculated dynamic value scores and costs of the ad slots, it configures and recommends slot combinations that produce the maximum effect within a given budget.

[0090] In addition, with administrator approval, selected slots are automatically reserved via external platform APIs.

[0091] In addition, information {"targetArea": ​​"Gangnam-gu", "offlineBudget": 10000000} is received from the campaign setting module (200).

[0092] In addition, the optimal slot combination list [{"slotID": "KNG-003-18", "cost": 250000}, ...] configured by the third operation logic is sent to the administrator dashboard to request approval, and upon approval, a reservation command is sent to the corresponding digital outdoor advertising platform API.

[0093] The above integrated ad delivery module (600) is an automated execution engine that executes and controls actual ad execution according to a set campaign schedule based on the finally selected content and ad channels (including influencers).

[0094] In addition, the optimized advertising material received from the content management module (400) is transmitted to each reserved online and offline advertising platform.

[0095] In addition, content guidelines and a dedicated tracking link are provided to the selected influencers, and notifications are sent to ensure that the content is posted at the scheduled time.

[0096] In addition, the real-time channel efficiency index is periodically received from the performance analysis module (700), and a command to dynamically adjust the budget distribution between channels according to the method of utilizing the first calculation logic is transmitted to the advertising linkage module (500).

[0097] In addition, a reservation completion signal for {'content_id': 'IMG_03_v2_16x9.mp4'} is received from the content management module (400), and for {'slot_id': 'KNG-003-18'} from the ad integration module (500).

[0098] In addition, at the start time of the campaign, an ad creation command is sent to the meta ad API and a content delivery start command is sent to the DOOH platform through the ad integration module (500).

[0099] The above performance analysis module (700) is a component that collects, integrates, and analyzes all performance data generated from each online and offline channel during and after the execution of a campaign to generate a comprehensive report and provides feedback to other modules of the system. This includes the role of providing real-time data for calculation according to the first operation logic of the present invention.

[0100] In addition, all performance indicators, such as impressions, clicks, conversions, QR scans, and costs, are collected in real time through each advertising platform API (web analytics, social media, DOOH).

[0101] In addition, a first operation logic is executed based on the collected real-time data to calculate the real-time channel efficiency index, and the result is transmitted to the integrated advertisement transmission module (600).

[0102] In addition, all collected online and offline performance data is visualized on a single dashboard, enabling managers to intuitively grasp the overall status of the campaign.

[0103] Additionally, after the campaign ends, the final performance data is organized to generate a training dataset to be used for retraining the prediction model of the content management module (400) and the recommendation algorithm of the influencer matching module (300).

[0104] In addition, real-time N_conv and N_cart data are received from the Google Analytics API, N_url data from the Bitly API, and P_on data from the Meta Ads API.

[0105] In addition, the calculated real-time channel efficiency index value '1.288' is sent to the integrated ad delivery module (600). The final campaign performance report {"total_conversions": 1250, "total_roi": 1.85, ...} is sent to the administrator dashboard.

[0106] In one embodiment, each component may be performed as follows:

[0107] The advertiser registers the ‘New Year’s Cosmetics Campaign Targeting Women in Their 20s’ in the campaign setting module (200) with a total budget of 30 million won.

[0108] Next, the influencer matching module (300) queries the data of the influencer information management module (100) and recommends three beauty influencers with a high proportion of female followers in their 20s.

[0109] The advertiser uploads five ad creatives to the content management module (400).

[0110] The content management module (400) uses a second calculation logic to calculate the ad content suitability score and automatically selects the materials 'IMG_A' and 'VID_B' that received the highest scores for Instagram and YouTube, respectively.

[0111] Next, the ad linkage module (500) collects DOOH slot information near the campaign target areas, ‘Gangnam-gu’ and ‘Hongdae Station’.

[0112] For each slot, a third operation logic is executed to calculate the dynamic value assessment score for the ad slot, and the slot combination of 'Gangnam Station A LED screen 18-20' and 'Hongik Univ. Station B screen 19-21' with the highest score within the budget is automatically reserved.

[0113] Next, the integrated ad delivery module (600) requests the influencer to post content on the campaign start date and starts delivering ads to the reserved offline slot.

[0114] On the third day of the campaign, the performance analysis module (700) collects real-time data and calculates the real-time channel efficiency index value according to the first calculation logic, which is 1.45, indicating that the online efficiency is very high.

[0115] The integrated ad delivery module (600) dynamically redistributes 10% of the offline budget into an online influencer ad boosting budget based on this real-time channel efficiency index value.

[0116] After the campaign ends, the performance analysis module (700) generates a final report integrating the performance of all channels and provides it to the advertiser.

[0117] The successful content ('IMG_A'), influencers, and the characteristics of the highly effective offline slots from this period are utilized as training data for the prediction model and recommendation algorithm of the next campaign, enabling the system to make increasingly sophisticated decisions.

[0118] The dynamic budget redistribution module (800) according to the present invention calculates a real-time channel efficiency index using the following first calculation logic to dynamically adjust the budget between online and offline marketing channels. The first calculation logic includes the step of calculating the relative ratio of the 'value per unit cost of online' which is calculated by dividing the real-time conversion value of an online channel by the execution cost of that channel, and the 'value per unit cost of offline' which is calculated by dividing the real-time conversion value of an offline channel by the execution cost of that channel; the step of multiplying the calculated relative ratio by the value obtained by adding a synergy coefficient between channels; and the step of calculating the final real-time channel efficiency index by multiplying the multiplied value again by a periodic volatility coefficient calculated based on the current time.

[0119] The real-time channel efficiency index is a final calculated value that indicates the relative marketing efficiency of online channels compared to offline channels at a specific point in time. The purpose of this index is to provide a quantitative and objective basis for determining which channels should be focused more on with campaign budgets.

[0120] The real-time conversion value of online channels refers to the real-time value generated through online advertising. This is calculated by adding the number of final purchase conversions to a value obtained by multiplying the number of cart additions—an action taken by potential customers with purchasing intent—by a specific weight. A key feature here is that the aforementioned cart addition weight is not a fixed constant, but a dynamic variable that learns and updates itself based on the actual performance of the campaign. Specifically, at the beginning of the campaign, the average 'purchase conversion rate after cart addition' of the relevant industry is set as the initial value; however, as the campaign progresses, a new weight value is calculated periodically (e.g., every 24 hours) by dividing the cumulative number of purchase conversions up to that point by the cumulative number of cart additions, and this is applied to the next cycle. This implies that this control logic does not rely on an arbitrary constant, but rather is a learning parameter that optimizes itself by reflecting actual consumer behavior patterns.

[0121] The real-time conversion value of offline channels refers to the value derived from offline advertising driving online behavior. This is calculated by summing the number of times users exposed to outdoor advertisements scanned QR codes and the number of times they clicked by directly entering campaign-specific shortened URLs; its purpose is to evaluate the effectiveness of offline advertising by converting it into measurable digital behaviors.

[0122] The real-time cost executed on each channel is the advertising cost incurred on each channel during a specific period. The purpose of this variable is to serve as a standard for calculating the return on investment for each channel.

[0123] The inter-channel synergy coefficient is a measure representing the positive interaction effect that offline advertising has on online channels. Specifically, it is measured by the search volume growth rate, indicating how much the online search volume for campaign-related keywords has increased compared to usual levels during the offline advertising execution period; its purpose is to evaluate the two channels as being in a mutually complementary relationship rather than as independent entities.

[0124] The periodic variability coefficient is a coefficient used to mathematically model the real-world phenomenon where advertising efficiency changes periodically over time. Fundamentally, this invention utilizes a cosine function, which is the most intuitive and computationally easy to simulate the universal single-peak pattern where efficiency is concentrated during specific times of the day. Furthermore, this invention can selectively provide an advanced mode to address complex periodicities involving multiple peaks, such as commuting times. In this case, the system ensures the scalability and precision of the model by applying signal processing techniques, such as Fourier transform, to historical performance data to extract key periodic components and superimposing them to generate a more sophisticated periodic variability coefficient.

[0125] The current time is the current time information from 0:00 to 23:00. It is used for the purpose of mathematically modeling the actual phenomenon where advertising efficiency changes periodically over time.

[0126] The peak efficiency time phase is a correction constant that refers to the time of day when advertising efficiency is highest. Its purpose is to flexibly adjust the peak efficiency time to suit the characteristics of the campaign.

[0127] The above-mentioned first computational logic was derived by fusing the economic theory of return on investment with the signal processing theory of periodicity modeling to overcome the limitation of the prior art, which statically selects a channel based on past data. The derivation process is as follows.

[0128] The basic efficiency of each channel is defined as 'cost-to-value'. The value per unit cost of online channels is calculated by dividing the real-time conversion value of online channels by the execution cost of online channels, while the value per unit cost of offline channels is calculated by dividing the real-time conversion value of offline channels by the execution cost of offline channels. By calculating the ratio of the two values ​​per unit cost, a normalized metric is generated to compare the relative efficiency of the two channels regardless of budget size.

[0129] The prior art had a problem in that it could not consider the interaction between channels. The present invention models the synergistic effect of offline advertising increasing online search volume as a term obtained by adding the inter-channel synergy coefficient to 1, and multiplies it by the basic efficiency ratio calculated in step 1. This is based on the technical fact that the two channels may be in a cooperative relationship rather than a simple competitive relationship, and it measures the effect of integrated marketing more accurately.

[0130] User responses to advertisements have periodic characteristics that are concentrated during specific times, such as commuting hours and lunch breaks. To mathematically model these characteristics, the present invention introduces a cosine function with a period of 24 hours. This is calculated using the difference between the current time and the maximum efficiency time phase, and provides a technical basis for overcoming the limitations of static analysis by reflecting the actual advertising efficiency that changes over time.

[0131] By multiplying all the elements of the above steps, a complex dynamic channel efficiency index is completed that considers relative efficiency, inter-channel synergy, and temporal periodicity.

[0132] The above purchase conversions and cart additions are collected in real time through APIs of web analytics platforms such as Google Analytics or Adobe Analytics.

[0133] For example, the system sends a query to the Google Analytics Real-time Reporting API every 30 minutes. Among the API response data, the total number of occurrences of the event named 'purchase' is obtained as the purchase conversion count, and the total number of occurrences of the event named 'add_to_cart' is obtained as the cart addition count. For example, if the API response results from 14:00 to 14:30 are 12 purchases and 55 cart additions, the system confirms the purchase conversion count as 12 and the cart addition count as 55.

[0134] The number of QR code scans and shortened URL clicks mentioned above are tallied in real time through tracking APIs of shortened URL and QR code generation platforms such as Bitly and Rebrandly.

[0135] For example, the Bitly API link click lookup endpoint is periodically called for a specific shortened URL used in a campaign. If the cumulative number of clicks was 1,205 at the time of the 14:00 call and 1,248 at the time of the 14:30 call, the number of shortened URL clicks during that 30-minute period is calculated as 43. The number of QR code scans is also obtained in the same way by aggregating the number of clicks on the shortened URL linked to the QR code.

[0136] The real-time costs incurred on each of the above channels are obtained in real time through the reporting APIs of each advertising medium (e.g., Meta Ad Manager, outdoor advertising media company).

[0137] For example, the system accesses Meta's marketing API to query the value of the cost field corresponding to a specific campaign ID. If the cost value returned as '54,300 won' upon querying at 14:30, the real-time cost for the online channel is confirmed as 54,300 won. For the offline channel, the hourly billing amount of 150,000 won for the corresponding slot is converted to a 30-minute unit from the settlement system API of the linked outdoor advertising media company and obtained as 75,000 won.

[0138] The search volume growth rate used to calculate the synergy coefficient between the aforementioned channels is calculated by collecting search volume data for a specific time period through the APIs of Naver Data Lab or Google Trends.

[0139] For example, the system queries search volume data for the campaign keyword 'HanPatent' through the Naver Data Lab API. If the search volume index for the offline advertising execution time (14:00-14:30) is measured at 85 and the average search volume index for the past 7 days during the same time period is 70, the search volume growth rate is calculated as approximately 0.214. Therefore, the synergy coefficient between channels becomes 0.214.

[0140] The above current time and maximum efficiency time phase are obtained through the system's internal clock and values ​​entered by the administrator during campaign setup.

[0141] For example, if the system executes the corresponding operation logic at 14:30, it obtains the current time as 14.5 from the system's internal clock. If the campaign manager has previously selected the 'Lunchtime Focus' option, the system automatically sets the maximum efficiency time phase value to the predefined 12.5.

[0142] The dynamic budget redistribution module (800) of the present invention periodically (e.g., every 30 minutes) recalculates the real-time channel efficiency index value. At this time, the threshold value serving as the basis for budget redistribution is not a fixed constant set in advance, but is dynamically determined based on the statistical variability of the real-time channel efficiency index calculated over a recent period (e.g., the past 24 hours). Specifically, the system calculates the standard deviation of the index values ​​and then sets the values ​​obtained by adding and subtracting a specific multiple of the standard deviation based on '1' as the upper threshold and lower threshold values, respectively. Here, the multiplier applied to the standard deviation can be adjusted by an administrator selecting either a 'stable operation mode' or an 'aggressive optimization mode' depending on the campaign goal. Through this, the system reacts sensitively to small inefficiencies when channel performance is stable, and prevents unnecessary overreactions when performance fluctuations are severe, thereby ensuring the stability of the system.

[0143] Furthermore, if the calculated index deviates from the dynamically set threshold range, the ratio for redistributing the budget is not a fixed value but is determined in proportion to the degree to which the index deviates from the threshold. In other words, a proportional-adaptive control method is adopted in which the budget is adjusted by an amount proportional to a minimum ratio (e.g., 2%) when the index deviates slightly from the threshold, and the redistribution ratio is further increased in proportion to the difference when the index deviates significantly. This possesses a technical feature that makes the system's response much more sophisticated and rational by taking small measures for minor inefficiencies and larger measures for serious inefficiencies.

[0144] If the calculated index is greater than 1.1, it is determined that the efficiency of online channels is 10% or higher than offline, and a portion of the total budget (e.g., 5%) is redistributed from offline to online channels.

[0145] If the calculated index is less than 0.9, it is determined that the efficiency of offline channels is 10% or higher than that of online channels, and a portion of the total budget (e.g., 5%) is redistributed from online to offline channels.

[0146] If the calculated index is between 0.9 and 1.1, it is determined that the efficiency of the two channels is balanced, and the current budget allocation is maintained.

[0147] In one embodiment, it is assumed that the following data was collected at a specific time, 2:00 PM.

[0148] Purchase conversions=50, Adds to cart=100, Add to cart weight=0.2

[0149] Online channel cost = 100,000 won

[0150] QR code scans = 30, shortened URL clicks = 20

[0151] Offline channel cost = 80,000 won

[0152] Channel synergy coefficient = 0.15 (15% increase in search volume)

[0153] Maximum efficiency time phase=14 (set to 2 PM)

[0154] In one embodiment,

[0155] Online Channel Real-time Conversion Value = 50 + (0.2 * 100) = 70

[0156] Real-time conversion value of offline channels = 30 + 20 = 50

[0157] In one embodiment,

[0158] Value per online unit cost: 70 / 100,000 = 0.0007

[0159] Value per offline unit cost: 50 / 80,000 = 0.000625

[0160] Default ratio: 0.0007 / 0.000625 = 1.12

[0161] In one embodiment,

[0162] Synergy Effect Coefficient: (1 + 0.15) = 1.15

[0163] Periodicity coefficient: Since the current time (14) and the maximum efficiency time phase (14) are the same, the input value of the cosine function becomes 0 and the result becomes 1.

[0164] In one embodiment,

[0165] Real-time Channel Efficiency Index = 1.12 * 1.15 * 1 = 1.288

[0166] The final calculated index value of 1.288 means that online channels are approximately 28.8% more efficient than offline channels at this point in time, and accordingly, the budget is redistributed to online channels.

[0167] The above first operation logic ensures robustness by including exception handling logic that treats the efficiency of the corresponding channel as 0 when the cost is 0, in order to prevent division errors from occurring when the online channel cost or offline channel cost becomes 0.

[0168] Since the cosine function that calculates the periodicity above always limits the output value to between -1 and 1, it prevents changes in the current time or maximum efficiency time phase value from having an excessive and explosive effect on the final result, thereby ensuring the stability of the system.

[0169] By setting an upper limit (e.g., 1.0, i.e., a 100% increase) on the synergy coefficient value between the above channels, stability is enhanced so that an abnormal surge in search volume does not distort the overall calculation.

[0170] Meanwhile, assume a situation where a campaign is conducted for 10 days with a total budget of 10 million won. The prior art determines a static budget allocation of '60% offline, 40% online' through initial analysis. The present invention dynamically redistributes the budget by calculating a real-time channel efficiency index every day. Performance is evaluated as the sum of the final conversion values.

[0171]

[0172] As a result of the simulation above, the present invention achieved an efficiency approximately 16.4% higher in total performance compared to prior art by optimizing the budget through active response to rapid market changes, such as online viral marketing on day 5 and offline events on day 7.

[0173] The input variables of the real-time channel efficiency index, which is the output value of the first computational logic, are data objectively measured through external APIs or system logs, and this logic is a process of combining these objective input values ​​within a logical framework. In particular, the technical feature of the present invention lies in the following structural superiority.

[0174] First, it represents a paradigm shift from a 'fixed model' to an 'autonomous learning framework.' While conventional technology assumes that the optimal value is known in advance and presents fixed logic, this invention acknowledges that the optimal value is unknown and incorporates a learning mechanism that autonomously finds the correct answer (e.g., cart add weight) through the campaign's own data. This is a technological concept fundamentally different from simply presenting arbitrary numbers.

[0175] Second, there is an evolution from 'discrete control' to 'continuous-proportional control.' While conventional technology simply shifts a fixed amount of budget when a threshold is exceeded, this invention introduces a proportional-adaptive control logic that measures the 'degree' of inefficiency and adjusts the 'intensity' of the response in proportion. This applies the principles of control engineering to marketing optimization, enabling much more sophisticated and stable system operation.

[0176] Third, it represents an advancement from 'static risk management' to 'dynamic risk management.' This invention measures the volatility of channel performance in real time and automatically adjusts the threshold range for determining budget changes. This enables the system to recognize market stability and instability, performing a dynamic risk management function that automatically switches between conservative and aggressive stances. This configuration goes beyond simple optimization to improve the robustness and stability of the system.

[0177] In the present invention, "optimization of budget distribution" means a process of periodically calculating a real-time channel efficiency index value through the first operation logic, and when the dynamically defined threshold criteria are met, automatically transferring a budget amount proportional to the degree of inefficiency from an inefficient channel to an efficient channel.

[0178] "The point where campaign performance is maximized" means a state in which, by repeating the above optimization process during the campaign period, the total sum of cumulative conversion value obtained with the same total budget is achieved at a statistically significant level (e.g., 10% or more) compared to a static budget allocation method or a fixed threshold-based allocation method.

[0179] The content management module (400) according to the present invention calculates an advertising content suitability score using the following second calculation logic to automatically select the most effective content among a plurality of advertising materials for a specific channel and target customer. The second calculation logic includes the step of calculating a 'basic content value' by summing the visual attractiveness and expected user engagement of the advertising material; the step of multiplying the basic content value by a channel format matching coefficient of the advertising material; and the step of calculating a final advertising content suitability score by multiplying the multiplied value again by an emotional matching coefficient of the advertising text. At this time, the visual attractiveness is calculated by taking the square root of the result of summing the squared values ​​of the image clarity score, color harmony score, and composition stability score, and the expected user engagement is calculated by summing the result of log-transforming the expected click-through rate, expected number of shares, and expected number of views, respectively.

[0180] The ad content suitability score is a final output value, a quantitative score representing the overall attractiveness and technical suitability of a specific ad creative for its target channels and audiences. The purpose of this score is to predict and select the content with the highest probability of success from a pool of ad creative candidates prior to A / B testing.

[0181] The intrinsic value of content refers to the inherent quality and potential performance of the advertising material. This is largely composed of visual completeness and anticipated user responsiveness. Visual completeness is a visual quality index of the advertising material evaluated through an artificial intelligence image analysis model. Importantly, this invention does not simply use an abstract value called a "quality score," but rather evaluates specific visual attributes individually and synthesizes them. For example, the system inputs an advertising image into a pre-trained image analysis AI model (e.g., Google Vision API, Amazon Recognition) to receive raw data representing attributes related to sharpness (e.g., analysis of high-frequency components of the image), attributes related to color harmony (e.g., distribution of major colors and complementary contrast), and attributes related to compositional stability (e.g., similarity to the rule of thirds or the golden ratio). This invention converts this raw data into individual attribute scores according to predefined rules (e.g., higher complementary contrast earns more points), and then synthesizes them to derive a final visual completeness score.

[0182] Predicted user responsiveness refers to the potential level of user engagement predicted based on data from similar past campaigns. A key feature here is that the present invention does not rely on an ambiguous "black box" known as a "prediction model," but rather defines specific features serving as the basis for the prediction and the model's learning method. Specifically, the prediction model uses the visual characteristics of newly registered ad creatives (e.g., text ratio within images, presence of people, primary color) and campaign attributes (e.g., target age, industry category) as input variables. This model is a Gradient Boosting Regression Model or a validated machine learning model with similar performance, which has learned what actual click-through rates, shares, and views were recorded by ad creatives with the same attributes across thousands of past campaigns. This clarifies that the predictions of the present invention are not unfounded speculations, but rather statistical inferences based on empirical historical data.

[0183] Channel format match is a coefficient indicating how well the aspect ratio of the ad content matches the recommended aspect ratio of the channel where the ad will be displayed. To impose a strong penalty for technical non-compliance, this is calculated using an exponential function where the value rapidly converges to zero as the difference in aspect ratio increases. In this case, the coefficient that controls the intensity of the penalty is not a fixed constant, but is designed to allow the manager to adjust it by selecting either 'Quality First Mode (high penalty)' or 'Creative Diversity Mode (low penalty)' depending on the campaign objective.

[0184] Text sentiment match is a coefficient indicating how well the sentiment index analyzed from the ad copy matches the campaign target sentiment. For example, if the ad copy "Amazing discount! Experience it right now!" is input into a natural language processing sentiment analysis API to obtain a 'Positive' score of 0.8, and the campaign target sentiment is 'Very Positive' (predefined 0.9), the match coefficient is calculated based on the difference between the two. The purpose of the above coefficient is to evaluate whether the tone and manner of the message intended by the brand is accurately reflected in the content.

[0185] The above-mentioned second computational logic was derived by combining computer vision theory, natural language processing theory, and statistical predictive modeling to overcome the limitation of prior art in being unable to evaluate the qualitative aspects of the content itself. The derivation process is as follows.

[0186] The visual quality of content is regarded not as a single metric, but as a point in a vector space composed of three independent axes: clarity, color, and composition. Visual appeal is defined by calculating the Euclidean distance from the origin to this point—that is, by taking the square root of the sum of the squared values ​​of each element. Based on the Pythagorean theorem, this method comprehensively evaluates the visual completeness of the content by applying engineering principles that measure the combined size of various components.

[0187] Furthermore, clicks, shares, and views differ significantly in their units and frequency of occurrence. By applying a logarithmic function to the value obtained by adding 1 to each metric, the range of values ​​is stabilized, and the influence of extreme values ​​is reduced. This is similar to the method of measuring the value of information in information theory and provides a technical basis for summing the values ​​of different types of participation activities on an equal scale.

[0188] Furthermore, even slight differences in channel format significantly degrade the user experience. To model this non-linear relationship, the channel format match is calculated by multiplying the absolute difference between the content aspect ratio and the recommended channel aspect ratio by a specific penalty factor and applying an exponential function with the result as the negative power of the natural constant. Since this has the characteristic that the value becomes 1 when the difference is 0 and rapidly converges to 0 as the difference increases, it is an effective method for imposing a strong penalty for technical non-conformity.

[0189] Next, the visual appeal and expected user engagement, which are intrinsic attractiveness of the content, are summed to set a base score, and the final score is calculated by multiplying this by the technical and emotional suitability coefficients, namely the channel format match and the text sentiment match. This mathematically implements the actual marketing principle that no matter how attractive the content is, its value decreases significantly if it does not meet channel specifications or fails to convey the intended sentiment.

[0190] The above sharpness score, color harmony score, and composition stability score are obtained by inputting the advertisement image into a pre-trained image analysis AI model, such as Google Vision API or Amazon Recognition, and acquiring the quality score (0~1) returned.

[0191] For example, the system sends the uploaded ad image file to the image attribute analysis function of the Google Cloud Vision API. It analyzes the major color distribution values ​​in the API response data to calculate a color harmony score of 0.88, analyzes the sharpness and noise of the image to calculate a sharpness score of 0.95, and calculates a composition stability score of 0.75 based on the composition suggestion score.

[0192] The above estimated click-through rates, estimated shares, and estimated views are calculated using regression analysis models or machine learning prediction models trained on data from thousands of past campaigns.

[0193] For example, the system converts the features and campaign goals of a newly uploaded ad image into vectors. These vectors are used as inputs for a pre-trained gradient boosting regression model. As outputs of the model, predictions such as an expected click-through rate of 3.5%, an expected number of shares of 210, and an expected number of views of 15,000 are obtained.

[0194] The aspect ratio of the above content is extracted from the metadata of the uploaded image file, and the channel-recommended aspect ratio is retrieved from the channel-specific specification database predefined in the system.

[0195] For example, the system reads the metadata of an image file and confirms that the image width is 1080 pixels and the image height is 1080 pixels. Accordingly, the content aspect ratio is calculated as 1.0. Since the campaign target channel is 'Instagram Feed', the system looks up the recommended aspect ratio for that channel as 1.0 in the internal database.

[0196] The sentiment index of the above phrase and the sentiment index of the campaign goal are obtained by inputting the ad phrase and campaign goal into a natural language processing sentiment analysis API and receiving the returned sentiment index (-1 to +1).

[0197] For example, the ad copy "Amazing discount! Experience it right now!" is sent to the sentiment analysis function of the Google Natural Language API. The API returns a score of 0.8 as a response, and the system uses this value as the phrase sentiment index to finalize it at 0.8. If the target sentiment set by the campaign manager is 'very positive', the system sets it to 0.9, which is the pre-mapped campaign target sentiment index value.

[0198] When planning an advertising campaign, the manager uploads five ad creative candidates to the system.

[0199] The content management module (400) of the present invention calculates an ad content suitability score by executing a second calculation logic based on a target channel (e.g., Instagram feed) for each candidate material.

[0200] The top two materials with the highest calculated scores (e.g., Material B: 1.85 points, Material E: 1.79 points) are automatically selected as final A / B testing targets and recommended to the manager.

[0201] In this case, if a specific score falls below a threshold, it is determined to be 'unsuitable' and automatically excluded from the candidate pool; however, the aforementioned threshold is not a fixed constant. It is set by dynamically calculating a suitability score corresponding to the bottom 5% of content that achieved target performance, based on an analysis of the distribution of performance data from campaigns executed in the same industry category in the past. This is an intelligent filtering method that flexibly adjusts minimum quality standards according to the characteristics of the campaign and market conditions.

[0202] In one embodiment, assume a situation in which ad creative 'B' is posted to an Instagram feed (1:1 ratio).

[0203] Sharpness score=0.9, Color harmony score=0.8, Composition stability score=0.7

[0204] Estimated Click-through Rate = 5%, Estimated Shares = 150, Estimated Views = 10,000

[0205] Content Aspect Ratio=1.0, Recommended Channel Aspect Ratio=1.0, Penalty Factor=5

[0206] Text Sentiment Index = 0.8, Campaign Goal Sentiment Index = 0.9

[0207] In one embodiment,

[0208] Visual attractiveness = square root of (0.9² + 0.8² + 0.7²) 1.39

[0209] Expected user engagement = log of (1+0.05) + log of (1+150) + log of (1+10000) 14.27

[0210] Channel format match = 1

[0211] Text sentiment match = 1 - 0.1 = 0.9

[0212] In one embodiment,

[0213] Ad Content Relevance Score = (1.39 + 14.27) * 1 * 0.9 14.09

[0214] The final calculated score of 14.09 indicates a high suitability of the content.

[0215] When calculating the visual attractiveness above, the square root of the sum of squares is used to prevent the overall score from being excessively distorted even if one specific visual element is abnormally high, and to ensure robustness by ensuring that all elements contribute in a balanced manner.

[0216] When calculating the above-mentioned expected user engagement, a logarithmic function is used to ensure that the calculation result does not diverge and remains within a stable range even if the predicted engagement value is 0 or has a very large value.

[0217] Since the above channel format matching and phrase sentiment matching output values ​​are always normalized to between 0 and 1, they play a role in stably adjusting the final score when used as correction coefficients.

[0218] An example is conducted to select the most effective one from a group of 10 candidate ad creatives. Since the prior art does not have this function, it is assumed that a marketer arbitrarily selects one (subjective selection). The present invention selects the creative with the highest ad content suitability score. Performance is evaluated by the click-through rate after actual execution.

[0219]

[0220] As a result of the simulation above, the present invention showed an effect of improving the average click rate by approximately 76.1% compared to the prior art (subjective selection) by comprehensively evaluating factors such as 'format match' and 'expected participation' that are easily missed by subjective judgment and selecting the optimal material.

[0221] The input variables of the ad content suitability score, which is the output value of the second operation logic, are based on objective and reproducible data sources, such as specific analysis results of artificial intelligence models, statistical predictions of historical data, file metadata, and natural language analysis results. In particular, the present invention has a differentiated structure as follows.

[0222] First, it involves the concretization from 'abstract quality evaluation' to 'concrete attribute-based analysis.' Unlike conventional technology that merely claims to evaluate 'quality,' this invention specifies a concrete computational process that breaks down raw data returned from an artificial intelligence API into interpretable multidimensional attributes such as 'clarity,' 'color harmony,' and 'composition,' and then synthesizes them. This configuration ensures the transparency and reliability of the evaluation process.

[0223] Second, it is a transition from 'black box prediction' to 'transparent feature-based prediction'. The present invention goes beyond simply stating that 'machine learning makes a prediction' by specifying the specific input features used by the prediction model (such as the proportion of text within an image or whether a person is included) and disclosing that verified machine learning techniques (e.g., gradient boosting) are used. This is an approach that enables a third party to understand and reproduce the technology by clarifying the basis of the prediction.

[0224] Third, there is an advancement from 'fixed-criteria filtering' to 'dynamic-adaptive filtering.' This invention does not use a fixed value for the threshold serving as the 'minimum quality standard.' Instead, it statistically analyzes the performance distribution of past similar campaign data to dynamically set the optimal threshold that aligns with the current campaign's objectives. This implements a higher-level intelligent filtering mechanism that flexibly adjusts quality standards in accordance with market changes and campaign goals.

[0225] In the present invention, "most effective content" refers to content in which the advertising content suitability score calculated through the second calculation logic among a plurality of advertising material candidates belongs to the top 10% or more of the performance distribution of past similar campaign data.

[0226] "Minimum quality standard" means obtaining an ad content suitability score of at least the bottom 5% in the performance distribution of past similar campaign data, as described above, and this is a dynamic threshold set based on statistical facts.

[0227] The advertising linkage module (500) according to the present invention calculates a dynamic value evaluation score for each advertising slot using the following third calculation logic in order to automatically select and reserve the optimal slot that best matches the campaign goal among numerous offline advertising slots. The third calculation logic includes the step of calculating 'total expected utility' by multiplying the potential target exposure amount of the advertising slot by an environmental visibility coefficient; the step of calculating 'total opportunity cost' by multiplying the cost of the slot by a penalty coefficient based on the competitive advertising congestion; and the step of calculating a final dynamic value evaluation score for the advertising slot by dividing the calculated total expected utility by the total opportunity cost. At this time, the potential target exposure amount is calculated by multiplying the number of people passing through the slot by the target customer base match rate, and the penalty coefficient is calculated by taking the square root of the value obtained by adding 1 to the number of competitive advertisements.

[0228] The ad slot value score is a final calculated value, a quantitative score representing the overall marketing value of an ad slot at a specific location during a specific time slot. The purpose of this score is to automatically identify and prioritize offline ad slots capable of generating maximum advertising effectiveness within a limited budget, based on objective data.

[0229] Potential target impressions is an estimate of the valid foot traffic that matches the campaign's target customer base in terms of demographic characteristics, out of the total foot traffic passing through the ad slot during that time. It is calculated by multiplying the simple foot traffic count by the target match rate and aims to measure how much the advertisement reaches actual potential customers rather than meaningless targets.

[0230] The environmental visibility coefficient is a correction factor that reflects external environmental factors affecting the physical visibility of an advertisement. It is calculated by combining real-time weather conditions and time of day. Instead of using fixed weights such as '0.4 for rainy days' or '1.2 for nighttime,' the present invention may include a learning mechanism that periodically corrects these weight values ​​by statistically analyzing the impact of each environmental condition on the actual QR code scan rate or shortened URL click-through rate as campaign data accumulates. For example, if the QR scan rate during nighttime hours in a specific region is found to be 30% higher on average compared to daytime hours, the nighttime weight can be automatically adjusted to 1.3. This is an approach that quantifies the impact of environmental variables based on actual data, rather than arbitrary assumptions.

[0231] Slot cost is the advertising execution cost required to use the corresponding ad slot for one hour. The purpose of this variable is to serve as a criterion for considering economic efficiency, or return on investment, when evaluating the value of an ad slot.

[0232] Competitive ad congestion is the number of other advertisements being broadcast simultaneously in locations visually adjacent to the relevant ad slot. In this case, the scope of 'competition' can be calculated to include not only advertisements split and broadcast within the same screen, but also other adjacent advertising media that can be perceived together within the user's average field of view (e.g., 45 degrees), and this scope can be adjusted according to system settings. The purpose of the above variable is to mathematically model and reflect in the evaluation the effect that the more competing advertisements there are in the surroundings, the more attention is dispersed toward individual advertisements, thereby reducing advertising effectiveness.

[0233] The aforementioned third computational logic was derived by synthesizing the return on investment analysis in marketing, attention theory in environmental psychology, and the law of diminishing marginal utility in economics to overcome the limitation of prior art, which failed to provide specific evaluation criteria for offline advertising execution. The derivation process is as follows.

[0234] The most fundamental element of advertising effectiveness is 'how many potential customers it is seen by.' Unlike prior art, the present invention evaluates the qualitative aspects of advertising effectiveness by defining effective foot traffic matching the campaign target, rather than simple foot traffic, as 'potential target exposure.'

[0235] Furthermore, the value of an ad slot is not fixed but changes according to the real-time environment. The present invention dynamically adjusts the actual value of an advertisement based on environmental changes by introducing a visibility coefficient based on weather and time of day and multiplying it by the potential target exposure amount. This is an evaluation method that goes beyond static location analysis and reflects real-time conditions.

[0236] Next, the basis of the valuation is 'cost relative to expected effect'. A basic structure is established to calculate the effective exposure value obtainable per unit cost by dividing the actual expected effect derived in the previous step by the advertising cost.

[0237] Furthermore, the impact of competitive advertising is not linear. The shock of one competitive ad appearing when there are none is greater than the shock of one additional ad appearing where there are already 10. To model this principle of diminishing marginal utility, a square root function was applied to the value obtained by adding 1 to the competitive ad clutter. Since the square root function has the characteristic that its slope becomes gentler as the input value increases, it effectively expresses this non-linear reduction in effect; by multiplying this by the denominator, the design ensures that the total value decreases as competition intensifies.

[0238] The above floating population figures obtain floating population information for specific regions and times through floating population analysis service APIs based on telecommunications carrier base station data or APIs from public data portals such as the Seoul Open Data Plaza.

[0239] For example, the system queries a telecommunications company's commercial area analysis API with a specific area code (e.g., '11680580', near Gangnam Station) and a time (e.g., '18:00-19:00') as parameters. The API returns an estimated floating population of 4,850 for the area and time period, and the system uses this value as the floating population count and confirms it as 4,850.

[0240] The above target match rate is obtained as a percentage of the degree of match between the demographic characteristics of visitors in the relevant area and the campaign target, through a commercial area information analysis system or a regional consumption pattern data API provided by a credit card company.

[0241] For example, if the campaign target is set to 'women in their 20s and 30s', the system queries the commercial area analysis API with the area code '11680580'. The API returns the demographic distribution of visitors by time of day in that area. The system sums the 25% ratio of 'women in their 20s' and the 20% ratio of 'women in their 30s' corresponding to the target, and calculates the target match rate as 0.45.

[0242] The above weather weights are obtained by receiving the current weather through the Korea Meteorological Administration's real-time weather information API and applying predefined weights (e.g., clear=1.0, rain=0.4).

[0243] For example, the system queries real-time weather information by inputting the grid coordinates of the current location into the Korea Meteorological Administration's short-term forecast API. If the precipitation type code is returned as '1' (rain) in the API response, the system refers to an internal rule table and sets the weather weight value to 0.4. If the precipitation type code is '0' (no) and the sky condition code is '1' (clear), the weather weight value is set to 1.0.

[0244] The above slot costs and competitive ad congestion are provided in real time through the API of an ad reservation platform linked with outdoor advertising media companies.

[0245] For example, the system queries the API of the linked digital outdoor advertising platform for a specific slot ID. The API returns the real-time cost information for the slot, 250,000 won, and the number of third-party ads reserved for that time slot on the same screen, which is 4. Based on this, the system determines the slot cost to be 250,000 and the competitive ad congestion level to be 4.

[0246] In one embodiment, the advertiser sets a campaign budget (e.g., 1 million won per day) and a target area (e.g., Gangnam-gu).

[0247] The ad linkage module (500) of the present invention calculates an ad slot value score by executing the third calculation logic on an hourly basis for all ad slots (hundreds) available for reservation within a target area.

[0248] Ad slots are sorted in order of highest calculated value score, and slot combinations that provide the highest value within a given budget of 1 million won (e.g., Slot A 18-19, Slot B 12-13, Slot C 19-20) are automatically configured and recommended to advertisers, or automatically booked and executed according to settings.

[0249] In one embodiment, it is assumed that the value of the slot 'Gangnam Station A LED display, 19-20 o'clock' is evaluated.

[0250] Foot traffic = 5,000 people

[0251] Target Match Rate = 0.6 (60% of pedestrian traffic matches the target)

[0252] Weather weight=1.0 (Weather: Clear), Time zone weight=1.2 (Time zone: Night)

[0253] Slot cost = 150,000 won

[0254] Competitive ad congestion = 3

[0255] In one embodiment,

[0256] Potential target impressions = 5,000 * 0.6 = 3,000 people

[0257] Environmental Visibility Factor = 1.0 * 1.2 = 1.2

[0258] In one embodiment,

[0259] Total expected utility = 3,000 * 1.2 = 3,600

[0260] In one embodiment,

[0261] Total opportunity cost = 150,000 * square root of (1 + 3) = 150,000 * 2 = 300,000

[0262] In one embodiment,

[0263] Ad Slot Value Score = 3,600 / 300,000 = 0.012

[0264] The final calculated value score of 0.012 is compared with the scores of other slots and used as a criterion for determining priority.

[0265] Robustness is ensured by including exception handling that adds a very small value (e.g., 1) to the cost to prevent division by zero errors in case the slot cost in the denominator of the above third operation logic becomes 0.

[0266] By adding 1 to the above competitive ad congestion and taking the square root, the denominator becomes 1 even when there are no competitive ads, so the calculation is performed stably, and the calculation is maintained stably even for abnormal input values ​​for competitive ad congestion.

[0267] By setting reasonable upper and lower limits for each variable within the system, stability is maintained to prevent the overall calculation result from diverging into unrealistic values ​​even if abnormal data is received from the API.

[0268] In one embodiment, it is assumed that there are 10 offline ad slot candidates and the budget can only purchase 3 slots. Since the prior art does not have this feature, it is assumed that 3 slots are selected in order of lowest ad cost. The present invention selects 3 slots in order of highest ad slot value score. Performance is evaluated as the sum of the 'effective target exposures' that ultimately occurred.

[0269]

[0270] As a result of the simulation above, the prior art (cost-based selection) achieved 1,900 effective target impressions at a total cost of 180,000 won. On the other hand, the present invention achieved 7,200 effective target impressions at a total cost of 380,000 won. When converted to effective impressions per unit cost, the prior art achieved approximately 105 impressions per 10,000 won, while the present invention achieved approximately 189 impressions per 10,000 won, proving that the present invention achieved approximately 80% higher advertising efficiency.

[0271] The input variables for the ad slot value score, which is the output of this third computational logic—namely, pedestrian traffic, target match rate, weather conditions, slot cost, and competitive ad congestion—are all objective and verifiable actual data obtained through APIs from external data providers or advertising platform APIs. The aforementioned computational logic is a deterministic process that combines this objective data within an economic framework, and its characteristics are as follows.

[0272] First, the criteria for evaluating advertisements shift from 'quantity' to 'quality.' Areas with high foot traffic but irrelevant to the target audience are valued lower, while areas with high target density, even with low foot traffic, are valued higher. This fundamentally prevents the waste of advertising budgets and provides the effect of maximizing the 'effective reach' of advertisements.

[0273] Second, it reflects the non-linearity of competitive effects. Reflecting the insight that the attention-reducing effect of competitive advertisements is not linear, this logic applies a penalty through a square root function that reacts sensitively when competition is low and insensitively when competition is high. This implements a realistic valuation model that allows for a reasonable valuation without completely abandoning an oversaturated market.

[0274] Third, it features an integrated analysis structure based on Return on Investment (ROI). This logic is essentially structured to follow the ROI formula. The numerator represents the 'total expected utility' obtainable through advertising, while the denominator represents the 'total opportunity cost and risk' associated with advertising execution. This structure enables the analysis of complex variables by clearly separating them into two core concepts—'utility' and 'cost'—and serves as a systematic approach that reduces decision-making complexity and helps make the most economically rational choices.

[0275] In the present invention, the term 'optimal slot' refers to a slot among all available ad slots within a campaign target area in which the ad slot value score calculated by the third operation logic falls within the top 5% within the same budget group.

[0276] "Maximum advertising effect" means a state in which the sum of the "total effective target exposures" achievable when slots are combined in order of highest value score within a given total budget is statistically significant (e.g., 30% or more) compared to random or simple cost-based selection.

[0277] In the following, the integrated operation process of each component and control logic of the present invention is explained in detail step by step through a scenario in which a fictional cosmetics company, 'XYZ Co., Ltd.', launches a new product, 'Luna Glow Cushion', and executes a campaign with a total budget of 50 million won over two weeks with the goal of 'increasing product awareness and maximizing initial sales volume' targeting 'women in their 20s and 30s residing in the metropolitan area' as the core target.

[0278] A marketing manager of 'XYZ Co., Ltd.' accesses the system of the present invention and inputs the goal, target (women in their 20s and 30s, metropolitan area), budget (50 million won), and duration (2 weeks) of the campaign through a web interface provided by the campaign setting module. Based on the input information, the campaign setting module creates a campaign object with a unique ID and propagates the relevant information to all other modules.

[0279] The influencer matching module receives target information for 'women in their 20s and 30s' from the campaign setting module. The module queries the database of the influencer information management module to request a list of beauty category influencers whose followers consist of 70% or more women in their 20s and 30s. The influencer information management module returns a list of five candidates, including 'XXX' (300,000 followers) and 'YYY' (150,000 followers), along with data on each influencer's average engagement rate and estimated unit price, to the influencer matching module. Considering the budget and expected engagement, the influencer matching module ultimately selects and recommends 'XXX' and 'YYY' as the most suitable influencers for the campaign.

[0280] The marketing manager uploads two types of ad creatives to the Content Management module: 'Creative A (luxurious studio video)' and 'Creative B (natural everyday usage video that looks like it was filmed by an influencer).' Based on the target channel, the 'Instagram Feed,' the Content Management module executes a secondary logic on each creative to calculate an ad content suitability score. 'Creative A' has a high 'visual appeal' score but is predicted to have relatively low 'expected user engagement,' whereas 'Creative B' has average 'visual appeal' but is predicted to have very high 'expected user engagement' based on the characteristics of the Instagram channel and historical data. Ultimately, 'Creative B' obtains a higher ad content suitability score and is automatically recommended as the primary creative for the Instagram campaign.

[0281] The ad integration module collects a list of available digital out-of-home (DOOH) advertising slots around the campaign target areas, 'Gangnam Station' and 'Hongik Univ. Station,' via an external platform API. When comparing the two slots, 'Gangnam Station C Billboard' and 'Hongik Univ. Station D Shelter,' 'Gangnam Station C Billboard' has a high volume of pedestrian traffic but a relatively low target match rate and high competitive advertising congestion. On the other hand, 'Hongik Univ. Station D Shelter' has a low absolute volume of pedestrian traffic but a very high target match rate for women in their 20s and 30s and low competitive advertising congestion. The module executes a third-party computational logic for each slot to calculate an ad slot value score. As a result of the calculation, 'Hongik Univ. Station D Shelter' yields a higher value score, as it can achieve higher-quality valid target exposure at a lower cost. Based on this result, the system selects 'Hongik Univ. Station D Shelter' as the priority reservation target.

[0282] On the campaign start date, the integrated ad delivery module delivers the optimized 'B creative' received from the content management module to 'XXX' and 'YYY' along with a dedicated tracking link and requests publication. At the same time, it commands the ad integration module to deliver the outdoor advertising version of 'A creative' to the reserved 'Hongik Univ. Station D Shelter'.

[0283] On the fifth day of campaign execution, the performance analysis module collects performance data from each channel in real time. The analysis shows that the post by influencer 'XXX' went viral, causing a surge in purchase conversions on online channels, while the number of QR code scans via offline advertising remains steady. Based on the collected data, the module executes the first computational logic to calculate the real-time channel efficiency index, yielding a high value of 1.35. This indicates that, at this point in time, the online channel is 35% more efficient than the offline channel. Upon receiving this index value, the integrated ad delivery module immediately sends a command to the ad integration module to dynamically redistribute 5% of the offline budget as an ad boosting budget for the online 'XXX' post, thereby maximizing high efficiency.

[0284] On the 8th day of the campaign, the performance analysis module discovers that the website conversion rate of Influencer 'YYY's' 'Material B' post is relatively low compared to its high number of 'Likes'. The system suspects a mismatch between the content and the followers' tendencies and requests a re-evaluation from the content management module. Based on historical data indicating that 'YYY's' followers respond more to 'authentic reviews', the content management module re-executes the second calculation logic on 'Material C (a compilation video of reviews by ordinary people)' among the uploaded candidate materials. At this point, as a result of recalculating by reflecting actual data from the past 7 days in the expected user engagement variable, the ad content suitability score for 'Material C' is calculated to be higher than that of 'Material B'. The system recommends to the marketing manager that 'YYY's' content be replaced with 'Material C', and upon approval, immediately replaces the content through the integrated ad delivery module.

[0285] After the two-week campaign ends, the performance analysis module integrates all online and offline performance results to generate a final report. The report shows that a final return on investment 18% higher than the results of a static budget allocation simulation was achieved through dynamic budget allocation and content replacement. In addition, data on success factors obtained through the campaign, such as "natural daily life videos (Material B) are effective for female targets in their 20s" and "the target match rate is very high during weekday evening hours at Hongik University Station," are used to retrain the system's machine learning model. Through this learning process, the system of the present invention further improves the prediction accuracy of the second and third computational logics when conducting similar campaigns in the future.

[0286] In the following, the integrated operation process of the present invention is explained in detail step by step through a scenario in which the FinTech company 'ABC Co., Ltd.' conducts a pre-registration event for its new stock investment service 'Alpha Trader' and executes a campaign with a total budget of 80 million won over one month, targeting 'male office workers in their 30s and 40s' as the core target and aiming to 'maximize the acquisition of event participants (potential customers).'

[0287] A marketer at 'ABC Inc.' registers the 'Alpha Trader Pre-registration Campaign' through the system's campaign configuration module. The core target is set to 'men in their 30s and 40s, residents of the metropolitan financial hubs (Yeouido, Gwanghwamun, Gangnam),' and the target performance is set to 'the number of email addresses submitted via the pre-registration page.' The campaign configuration module generates a campaign ID and propagates information regarding the target audience and performance to relevant modules.

[0288] The influencer matching module queries the influencer information management module based on target information for 'men in their 30s and 40s.' Since conveying credibility is crucial for this campaign, 'expertise' and 'high follower engagement' are set as key criteria in addition to simple follower count. The influencer information management module recommends three candidates: 'Economic YouTuber AAA' (high expertise, 200,000 followers), 'IT Reviewer BBB' (logical, 400,000 followers), and 'Personal Finance Blogger CCC' (high credibility, 100,000 neighbors). The marketer ultimately selects 'Economic YouTuber AAA' and 'Personal Finance Blogger CCC' from among these, as they are the most suitable for conveying credibility.

[0289] Considering the channel characteristics of the two influencers, the marketer registers 'Material A (a 10-minute analysis video emphasizing the technical excellence of Alpha Trader)' and 'Material B (a card-news style blog post explaining the service's stability and revenue model)' in the content management module. The content management module executes a second computational logic for each material. When evaluated against the channel (YouTube) of 'Economic YouTuber AAA', 'Material A' receives high scores in 'Visual Appeal' and 'Channel Format Match'. When evaluated against the channel (Blog) of 'Investment Blogger CCC', 'Material B' obtains high scores in 'Expected User Engagement (Shares and Comments)' and 'Text Sentiment Match (Reliability)'. The system matches and recommends 'Material A' and 'Material B' as content optimized for each influencer and channel, respectively.

[0290] The ad integration module collects digital ad slot information for subway stations and bus stop shelters in the campaign target areas of 'Yeouido', 'Gwanghwamun', and 'Gangnam'. In particular, it intensively analyzes slots during rush hour (08-09 and 18-19). As a result of executing the third computational logic, the 'Yeouido Station Transfer Passage Screen' slot records the highest 'Ad Slot Value Score' because, although the cost is high, the 'Potential Target Exposure' during morning rush hour is overwhelmingly high and the congestion of competing ads is low. On the other hand, the 'Gangnam Station' slot receives a relatively lower score because, despite high foot traffic, the target match rate is lower than that of Yeouido. Considering the calculated value scores and budget, the system selects the 'Yeouido Station Transfer Passage 08-09' and 'Gwanghwamun Bus Stop 18-19' slots as the optimal combination and proceeds with the reservation.

[0291] With the start of the campaign, the integrated ad delivery module requests 'Economic YouTuber AAA' to post an analysis video on 'Material A' and 'Investment Blogger CCC' to post a card news article on 'Material B,' along with their respective tracking links. Simultaneously, it begins broadcasting advertisements with the slogan 'Alpha Trader, Innovating Your Investments' on reserved outdoor billboards in Yeouido and Gwanghwamun.

[0292] In the second week of the campaign, the performance analysis module discovers that online search volume for the campaign keyword 'Alpha Trader' surges by 40% compared to usual during commuting hours when offline ads are being executed. This data significantly increases the 'Channel Synergy Coefficient' value when calculating the first computational logic. Simultaneously, on the online channel, videos by 'Economic YouTuber AAA' are recording steady views, driving pre-booked conversions. The calculated 'Real-time Channel Efficiency Index' is 0.95, demonstrating that the two channels are generating a highly balanced synergy effect. Accordingly, the integrated ad delivery module fine-tunes the campaign to further strengthen the synergy effect by maintaining the current budget allocation while requesting the content management module to add the phrase "Search for 'Economic YouTuber AAA' on YouTube now!" to the offline ad creative.

[0293] In the third week of the campaign, the performance analysis module detects that negative public opinion has arisen due to a personal issue involving influencer 'Investment Blogger CCC,' causing an increase in negative comments on the relevant blog post and a sharp decline in the pre-reservation conversion rate. The system determines this to be a 'crisis situation' and immediately sends a warning notification to the marketer. After confirmation by the marketer, the system urgently proposes 'IT Reviewer BBB,' the original candidate, as a replacement influencer through the influencer matching module. Since it is determined that 'BBB' can secure credibility through content that logically reviews the technical aspects of the service, a contract is immediately signed, and new content tailored to the 'BBB' channel is rapidly produced and distributed through the content management module.

[0294] After the one-month campaign concludes, the performance analysis module generates a final performance report showing that the target number of pre-registrants was exceeded by 120%. The report demonstrates with data that the combination of offline advertising in financial hubs during commuting hours and professional online content generated high synergy, and that damage was minimized and performance was defended through the rapid replacement of influencers in the event of a crisis. Specific success formula data, such as "logical and analytical content is effective for male targets in their 30s and 40s" and "advertising during Yeouido's morning commute hours is highly cost-effective," is accumulated in the system, contributing to further enhancing the accuracy of second and third computational logic when conducting future financial product-related campaigns.

[0295] In the following, the integrated operation process of the present invention is explained in detail step by step through a scenario in which the global game developer 'Ganada' executes a global campaign with a total budget of 200 million won for two months, starting three months before the launch of the new MMORPG (Massively Multiplayer Online Role-Playing Game) 'Mabasa', targeting 'core gamers' and 'potential mass gamers' with the goal of 'amplifying pre-launch anticipation and forming an initial community'.

[0296] The marketing manager of 'Ganada' registers the 'Mabasa Global Hype Campaign' through the system's campaign settings module. The core targets are 'core gamers in their teens and twenties' and 'fantasy genre enthusiasts in their twenties and thirties,' and the target performance metrics are set as 'number of official Discord server members,' 'total views of the official trailer video,' and 'social media buzz volume (mentions).' This campaign is characterized by targeting qualitative indicators such as community activation and anticipation, rather than direct revenue.

[0297] The influencer matching module executes a multi-layered matching strategy tailored to the bifurcated targets of 'core gamers' and 'mass gamers.' For the first wave (core target), it queries the influencer information management module to select 10 'micro-influencers' (with 10,000 to 50,000 followers) who possess high expertise in specific genres (MMORPGs) and loyal followers, for campaigns based on the concepts of 'information leaks' or 'first reveal.' For the second wave (mass target), timed with the release of the official trailer, it selects three major game streamers with over 1 million followers for 'trailer reaction' and 'live play' campaigns. This is a sophisticated strategy that sets the scale and roles of influencers differently depending on the target group and campaign stage.

[0298] The marketing manager registers various content in the content management module according to the campaign stages. Stage 1 (Teasing): Materials that arouse curiosity, such as partial game artwork or mysterious screenshots, are registered. The second computational logic selects materials most likely to trigger discussion and speculation within the community by giving high weight to 'expected number of shares' and 'expected number of comments' among 'expected user engagement.' Stage 2 (Information Release): Official trailer videos are registered. The second computational logic evaluates 'visual appeal' and 'expected views' as key indicators and recommends the edited version with the greatest visual impact as the final version. This demonstrates the flexibility of the present invention, which dynamically changes evaluation criteria according to the qualitative goals of the campaign.

[0299] The advertising integration module collects information on outdoor advertising slots around the global game show venue and screen advertising slots inside the venue. When executing the third computational logic, the 'target match rate' for 'potential target exposure' is set to a very high level of 95% or higher. This is intended to focus on 'genuine gamers' attending the event, rather than the general public. As a result of the calculation, the large screen next to the main stage of the venue yields the highest 'ad slot value score,' even though the cost is very high. Within the budget, the system selects and reserves the optimal combination of digital signage at the entrance and exit of the venue, taking into account the movement patterns of gamers, along with the corresponding slot.

[0300] The integrated ad delivery module simultaneously distributes teaser content in the form of 'exclusive information' to first-wave micro-influencers, creating a viral effect that makes it appear as if information has been leaked. All content includes an invitation link to the official Discord server. During the game show, official trailer videos are broadcast on reserved offline billboards, and a large QR code is displayed at the end of the video to encourage immediate connection to the Discord server.

[0301] In addition to simple performance data, the performance analysis module collects buzz volume and key keywords related to 'Mabasa' in real time through the 'social listening' function of major game communities such as Twitter and Reddit. The analysis reveals that gamers are showing an explosive reaction to the design of a specific character, 'Erisa,' and that related secondary creative works are being generated. At this point, the first computational logic calculates the 'conversion value' of online channels by replacing the 'number of pre-registrations' with the 'character-related buzz volume,' confirming that the efficiency index of online channels has skyrocketed. Based on these analysis results, the integrated ad delivery module immediately requests the content management module to produce additional content featuring unreleased artwork and backstory for the character 'Erisa.' The produced content is provided to major streamers, instantly responding to the community's interests and further amplifying the hype.

[0302] Midway through the campaign, a competitor's similar genre game, 'Tapaha,' announces a surprise beta test. The performance analysis module detects, through social listening, the risk that interest in 'Mabasa' will be diverted as the buzz volume for 'Tapaha' surges. The ad integration module re-executes the third-party computation logic and incorporates the 'buzz index of the competitor game' as an additional penalty factor into the 'competitor ad congestion' variable. As a result, the value scores of offline ad slots that overlap with the competitor game's main target audience temporarily drop. Based on this analysis, the integrated ad delivery module temporarily reduces the budget for the relevant offline slots and quickly switches to a strategy of retaining existing users by reallocating the budget to host exclusive events and Q&A sessions within the official Discord server where our game's 'loyal users' gather.

[0303] After the two-month campaign concludes, the performance analysis module generates a final performance report. Data demonstrates that the target number of Discord server subscribers surpassed 300,000, the official trailer recorded a total of 20 million views, and social media buzz volume overwhelmed competing games. In particular, the report analyzes success factors such as "the early information leak strategy targeting core gamers was effective" and "identifying community interests in real-time and reflecting them in content was key to amplifying buzz volume." This active community of 300,000 members remains a significant "asset" for the company, providing continuous virality and feedback even after the game's launch, going beyond mere advertising performance; the data accumulated during this process is utilized to formulate marketing strategies for the next title.

[0304] Although preferred embodiments of the present invention have been illustrated and described above, the present invention is not limited to the specific embodiments described above. It is understood that various modifications can be made by those skilled in the art without departing from the essence of the invention as claimed in the claims, and such modifications should not be understood individually from the technical spirit or perspective of the present invention. Explanation of the symbols

[0305] Influencer information management module (100) Campaign setting module (200) Influencer matching module (300) Content management module (400) Ad linkage module (500) Integrated ad delivery module (600) Performance analysis module (700) Dynamic budget redistribution module (800)

Claims

Claim 1 An automatic influencer matching system for marketing comprises: an influencer information management module that periodically collects, updates, and stores follower counts, posts, engagement rates, and follower demographic information for multiple influencers by linking with an API of a social media platform; an advertising linkage module that provides an interface for linking with multiple online and offline advertising media; a campaign setting module that sets the goals of a marketing campaign and the demographic information of target customers; an influencer matching module that recommends and selects influencers to participate in the campaign based on the degree of match between the demographic information of target customers input from the campaign setting module and the influencer follower demographic information stored in the influencer information management module; a content management module that registers and manages advertising materials to be used in the campaign; and an integrated advertising delivery module that executes and controls actual advertising execution based on influencers selected by the influencer matching module, advertising materials delivered from the content management module, and advertising slots reserved through the advertising linkage module. A performance analysis module that integrates all collected performance data after the campaign ends to generate a training dataset to improve the prediction accuracy of future campaigns;The method comprises: an influencer matching module that first filters influencers whose degree of match exceeds a preset first threshold as a candidate group, and selects a final influencer from the candidate group by comprehensively evaluating campaign category match, past campaign history, estimated advertising cost, and average participation rate; a content management module that calculates an advertising content suitability score by executing a second computational logic that comprehensively evaluates multiple elements including the visual attractiveness of the advertising material, estimated user engagement, and format match with the target channel for each of the registered multiple advertising material candidate groups; selects a top advertising material as an A / B testing candidate or a final execution material based on the advertising content suitability score; converts the selected advertising material to meet the technical specifications of the target channel and transmits it to the integrated advertising delivery module; and the second computational logic includes the step of determining the visual attractiveness by synthesizing the clarity, color harmony, and compositional stability scores of the advertising material calculated through an artificial intelligence image analysis model. and a step of determining the predicted user engagement by combining the predicted click-through rate, predicted number of shares, and predicted number of views calculated through a machine learning prediction model trained on past campaign data;The system includes, wherein the ad integration module executes a third calculation logic for each of a plurality of offline ad slot candidate groups collected from an external ad platform to calculate the total expected utility based on the potential target exposure of the slot and the total opportunity cost based on the slot cost and competitive ad congestion to calculate an ad slot dynamic value evaluation score; configures a slot combination within the allocated offline budget range based on the ad slot dynamic value evaluation score and the slot cost and transmits it to an administrator dashboard; and automatically reserves the selected slot via an external ad platform API upon administrator approval. In the third calculation logic, the total opportunity cost is calculated by multiplying the cost of the slot by a penalty coefficient calculated as the square root of the value obtained by adding 1 to the number of competitive ads simultaneously transmitted in a location adjacent to the slot, and this is characterized by reflecting a non-linear relationship in which the marginal impact of an individual competitive ad decreases as the number of competitive ads increases. The integrated ad delivery module transmits the ad creative sent from the content management module to the reserved online / offline ad platform and delivers the ad according to the set campaign schedule. An influencer automatic matching system characterized by transmitting commands.; Claim 2 delete

Images