Intelligent marketing decision analysis method and system for real-time competitor strategy

Abstract

The application provides a real-time competitor strategy-oriented intelligent marketing decision analysis method and system, which comprises the following steps: collecting marketing content published by competitors on a social media platform, extracting account number, influencer portrait features, content structure labels, platform identifiers and publishing time, and constructing a standardized strategy behavior vector; aligning the strategy behavior vector with our marketing indicators in time sequence, identifying a subset of competitor strategy behaviors that have a significant impact on our indicators; generating a structured response strategy based on the subset, including recommended influencer types, content structure labels, delivery platforms and time periods; and performing multi-dimensional matching with the influencer resource library and the content template library to output executable marketing resource combinations and scheduling instructions. The application realizes a closed loop from competitor behavior perception, causal influence identification to response strategy automatic generation and landing execution, and improves the real-time, accuracy and automation level of marketing response.

Description

Technical Field

[0001] This invention belongs to the field of data processing, and in particular relates to an intelligent marketing decision analysis method and system for real-time competitor strategies. Background Technology

[0002] In today's content dissemination environment dominated by social media, brand marketing campaigns are characterized by high-frequency changes, strong competition, and multi-platform heterogeneity. Competitors often vie for mindshare by adjusting influencer types, content structure, posting times, and platform choices, leaving brands frequently in a passive position during campaigns. Existing technologies primarily rely on static content analysis and human experience to formulate campaign strategies, failing to capture real-time changes in competitors' strategies across different platforms and time periods, or to identify whether these changes truly impact our conversion metrics. Furthermore, existing systems generally only provide display-style analysis of competitor content, lacking the ability to transform it into structured strategic actions, resulting in a lack of quantifiable strategic pathways for subsequent analysis. When competitor campaigns cause fluctuations in our marketing performance, current systems struggle to determine which strategic dimensions are at play, let alone automatically generate response strategies that comply with business rules. Existing strategy generation methods typically rely on manually written campaign summaries, heavily dependent on experience, leading to slow response times, severe strategy fragmentation, and an inability to stably link influencer selection, creative retrieval, and campaign scheduling, exhibiting a broken and discontinuous strategy chain. In real-world business scenarios, the impact of competitor strategies on our operations is often short-lived, sudden, and cross-platform. Without a structured representation of these strategies, time-sensitive causal analysis, and automated execution scheduling capabilities, massive waste of resources and delayed market response are inevitable. Therefore, the industry urgently needs a comprehensive solution that enables structured representation of competitor behavior, time-window causal identification, automatic generation and coordinated execution of response strategies, thereby maintaining consistency between marketing rhythm and strategy execution in a dynamic competitive environment. Summary of the Invention

[0003] The purpose of this invention is to design an intelligent marketing decision analysis method and system for real-time competitor strategies, which can meet the needs of rapid response, accurate matching and continuous expression of marketing strategies in a dynamic competitor environment.

[0004] To achieve the above objectives, a first aspect of the present invention provides an intelligent marketing decision analysis method for real-time competitor strategy, the method comprising:

[0005] Collect marketing content published by competitors on social media platforms, extract account numbers, influencer profile characteristics, content structure tags, platform identifiers, and publication times, and construct standardized strategy behavior vectors;

[0006] The strategy behavior vectors are arranged in chronological order to form a strategy behavior sequence, which is then time-sequentially aligned with our marketing indicator sequence.

[0007] Based on the sliding time window, combined with time decay weight and platform disturbance correction term, the sensitivity contribution of each strategic behavior dimension to our marketing metrics is calculated, and strategic behaviors with sensitivity contribution exceeding the preset impact threshold are screened to form a subset of competitor strategic behaviors that have a significant impact on our marketing metrics.

[0008] Based on the subset of competitor strategies and behaviors and their corresponding sensitivity contributions, a structured response strategy is generated that includes the type of recommended influencer, the structure tags of the recommended content, the recommended delivery platform, and the suggested delivery time period.

[0009] The structured response strategy is matched in multiple dimensions with the influencer profile information in the influencer resource library and the content templates in the content template library to generate executable marketing resource combinations and scheduling instructions.

[0010] Furthermore, the step of constructing the standardized policy behavior vector includes:

[0011] Normalize the characteristics of the influencer profile;

[0012] Maintain a five-dimensional classification tag vector format for content structure tags;

[0013] Discrete embedding encoding is used for platform identifiers;

[0014] Convert the release time into two one-hot encoded variables: intraday time period and weekday type;

[0015] The variables processed above are combined into a strategy behavior vector with a total number of dimensions not exceeding fifty.

[0016] Furthermore, the step of generating the structured response strategy includes:

[0017] Introduce a competitor differentiation suppression term to suppress response dimensions that converge with the recent high-performing strategies of competitors;

[0018] Introduce a platform consistency adjustment item to enhance the consistency between the response strategy and the historically high-performing content of the current platform;

[0019] The structured response strategy is generated by aggregating factors based on causal impact weights, competitor difference suppression terms, and platform consistency adjustment terms.

[0020] Furthermore, the profile information of each influencer in the influencer resource library includes: account number, platform coverage, fan level, influencer tag code, historical content style clustering vector, and recent interaction performance mean vector.

[0021] Furthermore, each content template in the content template library includes: content structure tags, platform adaptation encoding, and available time window tags.

[0022] Furthermore, the multidimensional matching steps include:

[0023] Calculate the matching degree between influencer profile information and recommended influencer types;

[0024] Calculate the matching degree between the content template and the recommended content structure tags;

[0025] Combine the activity level of influencers on the platform with the suitability of the recommended placement platform;

[0026] Combine the activity level of influencers with the suitability of the suggested ad placement time slots;

[0027] A resource adaptation score is generated by combining various matching factors.

[0028] Furthermore, after the step of generating executable marketing resource combinations and scheduling instructions, the method further includes:

[0029] Introduce a strategy inertia penalty term to evaluate the consistency between the current resource combination and historical response strategies;

[0030] The candidate resource combinations are ranked based on the combined results of the resource adaptation score and the strategy inertia penalty.

[0031] Select the resource combination that ranks highest as the final execution unit.

[0032] Furthermore, when the talent resources are unavailable, alternative combinations with similar scores are re-evaluated from the alternative resource pool, and the scheduling instructions are updated.

[0033] Furthermore, the preset impact threshold is dynamically determined based on the historical marketing indicator fluctuation range and the statistical distribution of competitor behavior.

[0034] A second aspect of the invention provides an intelligent marketing decision analysis system for real-time competitor strategy, the system comprising:

[0035] The strategy behavior construction module is used to collect marketing content published by competitors on social media platforms, extract account numbers, influencer profile features, content structure tags, platform identifiers and publication time, and construct standardized strategy behavior vectors.

[0036] The causal identification module is used to assemble the strategy behavior vectors into a strategy behavior sequence in chronological order and align it with our marketing indicator sequence in time. Based on a sliding time window, combined with time decay weight and platform disturbance correction term, the module calculates the sensitivity contribution of each strategy behavior dimension to our marketing indicators, filters out strategy behaviors whose sensitivity contribution exceeds a preset impact threshold, and forms a subset of competitor strategy behaviors that have a significant impact on our marketing indicators.

[0037] The response strategy generation module is used to generate a structured response strategy that includes the type of recommended influencer, the structure tags of the recommended content, the recommended delivery platform, and the suggested delivery time period, based on the subset of the competitor's strategy behavior and the corresponding sensitivity contribution.

[0038] The resource matching and execution module is used to perform multi-dimensional matching between the structured response strategy and the influencer profile information in the influencer resource library and the content templates in the content template library to generate executable marketing resource combinations and scheduling instructions.

[0039] The beneficial technical effects of the present invention are at least as follows:

[0040] To address the aforementioned issues, this invention provides an intelligent marketing decision analysis method and system for real-time competitor strategies. It provides a unified, structured representation of competitor influencer profiles, content structures, platform attributes, and time information, transforming unstructured content into strategic behavior vectors capable of computation and modeling. Subsequently, by introducing causal modeling methods incorporating time decay, platform difference perturbations, and multi-dimensional derivative sensitivity, it identifies a subset of strategic behaviors that actually impacts our marketing metrics from the behavioral sequences, providing verifiable evidence for response strategies. Building upon this, the invention further transforms causal behaviors and contribution information into structured marketing strategy variables by constructing difference suppression terms, platform consistency adjustment terms, and response inertia control terms. These variables include target influencer types, content structure tags, placement platforms, and recommendation time periods, giving response strategies a clear and executable structure. Finally, by performing multi-dimensional matching of structured strategy variables with execution elements such as influencer profiles, content templates, and resource activity, this invention generates resource combinations and execution plans that can be directly deployed for production and scheduling, achieving automated linkage from strategy identification to strategy implementation. The technical approach of this invention achieves rapid response, accurate matching, and continuous expression of marketing strategies in a dynamic competitive environment by abstracting the structure of strategy information, characterizing the causal relationship of the influence mechanism, and controlling the scheduling of the execution process. It has significant engineering value and application advantages. Attached Figure Description

[0041] The present invention will be further described with reference to the accompanying drawings, but the embodiments in the drawings do not constitute any limitation on the present invention. For those skilled in the art, other drawings can be obtained based on the following drawings without creative effort.

[0042] Figure 1 This is a flowchart of the intelligent marketing decision analysis method for real-time competitor strategy according to the present invention.

[0043] Figure 2 This is a framework diagram of the intelligent marketing decision analysis system for real-time competitor strategy of the present invention. Detailed Implementation

[0044] Embodiments of the present invention are described in detail below. Examples of these embodiments are shown in the accompanying drawings, wherein the same or similar reference numerals denote the same or similar elements or elements having the same or similar functions throughout. The embodiments described below with reference to the accompanying drawings are exemplary and are only used to explain the present invention, and should not be construed as limiting the present invention.

[0045] In one or more embodiments, such as Figure 1 As shown, an intelligent marketing decision analysis method for real-time competitor strategies is disclosed, the method comprising the following:

[0046] S1: Collect marketing content published by competitors on social media platforms, extract account numbers, influencer profile features, content structure tags, platform identifiers, and publication times, and construct standardized strategy behavior vectors;

[0047] Specifically, this step aims to transform the content actually published by competitors on social media platforms into a unified format of strategic behavior representations, laying the foundation for subsequent analysis and modeling. The collected information includes not only the published text and images but also elements such as platform, account, and publication time. Combined with influencer profiles and content structure characteristics, this forms a set of standardized strategic behavior variables, which are further transformed into a sequence of strategic behavior vectors suitable for modeling. This structure is specifically designed for "marketing placement strategies" rather than the semantics of the content itself, focusing on revealing the strategic intentions behind competitors, such as publication time selection, persona placement, platform distribution, and content style preferences, to support downstream influence identification and analysis.

[0048] The data collection sources for this step include the following three categories: The first category is basic metadata of competitor accounts and content, obtained from platform open interfaces or web scraping programs, including account number, content publication time, platform to which the content belongs, text, image or video cover, etc.; the second category is influencer profile feature information, obtained through image and text content parsing models. Image feature extraction uses an image classification module to identify the number and type of people appearing in the video, outputting terms such as "single," "couple," etc.; text feature extraction uses a recurrent neural network model to process the homepage text and the last five pieces of content to identify their style type, such as "story-based," "recommendation-based," "review-based," etc.; the third category is content structure feature tags, which use a content structure classifier to identify the expression structure type of the content, with tags including but not limited to five categories such as "emotional story," "product display," and "contextual immersion," outputting a five-dimensional classification label vector.

[0049] The above information is constructed into a set of policy behavior variables, over time... The competitive bidding behavior at any given moment can be represented by the following quintuple:

[0050] ;

[0051] in, This indicates the account number, which is directly obtained from the content attribution field provided by the platform. The feature vector representing an expert is composed of image recognition results and text style vectors. The image recognition part consists of one-hot vectors of three types of labels, and the text part consists of thirty-two sentence vectors. The final combination results in a thirty-five-dimensional feature representation of an expert. The content structure tag vector is a five-dimensional vector, where each dimension represents a structural type to which the content belongs, and is calculated by the content classifier. This represents a platform-encoded variable, with values ​​that are integers, representing different content publishing platforms. This indicates the content publication time and is retained as a timestamp for subsequent time window synchronization modeling.

[0052] Considering the inconsistent dimensions of the variables above—some are discrete variables and some are dense vectors—direct use would lead to modeling difficulties. Therefore, a unified mapping function is introduced:

[0053] ;

[0054] This function performs standardized encoding on each item in the quintuple, specifically as follows: The feature vectors of top performers are normalized to ensure balanced weights in the vector space; Maintain the original five-dimensional label vector format; Two-dimensional discrete embedding encoding vectors are used to represent platform features; for Convert into two time-labeled variables, representing "intraday time period" and "weekday type" respectively, and represent them using one-hot encoding.

[0055] Ultimately, the policy behavior vector This includes a standardized combination of all five dimensions, with the total number of dimensions kept below fifty, making it suitable for direct entry into the modeling phase. Multiple behaviors are arranged in chronological order to form a behavior sequence. It is used for time-series alignment and causal modeling with subsequent marketing performance data.

[0056] For example: A competitor's account posted a short video on the platform "Hongshu" at 10:15 AM on October 12, 2025. The content was an emotional wedding unboxing video featuring a couple, with the text written in a "memories + blessings" style. The data collected by the system includes: account number 114514, platform number 2 (representing "Hongshu"), posting time 10:15 AM, corresponding to the "morning + weekday" tag; the influencer's image feature is tagged "couple," and the text style output is a "story-like" vector; the content structure tag is "emotional story," with the first dimension of the five-dimensional tag being 1 and the rest being 0. This campaign behavior was standardized using a function. After processing, a behavior vector is formed. This information is stored in the behavior sequence. The output of this step is a set of standardized policy behavior vector sequences. This sequence is indexed by behavioral time, has a unified structure and clear dimensions, and can be used as an input variable set for subsequent causal analysis steps. It supports time window alignment and causal relationship modeling with our marketing performance indicators.

[0057] S2: Arrange the strategy behavior vectors into a strategy behavior sequence in chronological order and align it with our marketing indicator sequence in time. Based on the sliding time window, combined with the time decay weight and platform disturbance correction term, calculate the sensitivity contribution of each strategy behavior dimension to our marketing indicators, and screen strategy behaviors with sensitivity contribution exceeding the preset impact threshold to form a subset of competitor strategy behaviors that have a significant impact on our marketing indicators.

[0058] Specifically, this step aims to identify the competitive strategies and behaviors that truly have a significant impact on our marketing effectiveness, serving as input for subsequent strategy generation. This step doesn't simply calculate correlations or predict trends; instead, it uses time-aware modeling to determine whether fluctuations in our marketing performance at the current point in time are caused by specific competitive strategies and behaviors from previous periods. This step forms a closed loop with the previous one, entirely dependent on the sequence of strategic behaviors output from the previous step. Each of them Each vector contains information such as influencer characteristics, content structure tags, platform ID, and time tags, providing a unified representation of the campaign actions. This step integrates these behavioral vectors with our marketing metric sequence. The joint analysis outputs the affected subset of behaviors and their corresponding policy impact structure, which is used to drive the construction of subsequent response strategies.

[0059] Inputting This represents a vector of competitor behaviors arranged by time. This represents our key marketing performance indicators (including conversion rate, click-through rate, and content interaction frequency) within the same period, which are derived from the internal indicator logs of the advertising platform. Since multiple actions often occur simultaneously in real-world scenarios, and changes in marketing indicators have a time lag, this step employs a sliding time window modeling approach, using each of our key performance indicators as a reference. For the response point, the backward backtracking length is Time segment behavior sequences, constructing modeling samples and The mapping relationship.

[0060] The model structure designed in this step not only possesses time sensitivity but also introduces two targeted innovative mechanisms: First, considering the impact of emotional content, which is characterized by its rapid peak and subsequent decay within a short period, a time-series decay weight function is introduced during the modeling process. Second, considering the differences in the expressive inertia of strategies and behaviors across different platforms, such as the same influencer strategy having a sustained impact on short video platforms while often concentrating on the day of content publication on text and image platforms, platform factors are modeled separately as a disturbance moderating term. After comprehensive consideration, the modeling formula for this step is as follows:

[0061] ;

[0062] In this formula, The model represents time intervals. Our projected marketing metrics, Indicates the position of behavioral lag Learning weights This is a time decay term, representing the time-limited effect of the behavior. This is an adjustable coefficient; This is a platform disturbance correction item, numbered by the platform. The resulting perturbation vector is of fixed length. This serves as an adjustment coefficient. The platform disturbance term is used to adjust the intensity of the impact of the same type of behavior on the indicator on different platforms, making the model more consistent with the heterogeneity of actual marketing strategies.

[0063] Due to behavior vectors It inherently contains multiple dimensions. To further identify specific strategy factors that significantly impact the indicators, this step introduces a strategy feature contribution matrix, defined as follows:

[0064] ;

[0065] This matrix represents the position at lag. At that moment, the first strategic action The first dimension variable affects the marketing metrics. The degree of influence of each dimension refers to the sensitivity of each behavioral variable to the indicator result. After the model is trained, the matrix is ​​normalized, and an influence threshold is set. The behaviors of strategy variables whose influence exceeds a threshold are selected as "effective causal behaviors" and form a set of behaviors of concern. And retain their corresponding weights.

[0066] For example, if the system predicts a decline in conversion rate at a certain point on October 12, 2025, the model analysis reveals that the behavior of posting "high-emotion content" concentrated on the text and image platform three hours prior exhibits a high sensitivity to the current decline in the causal matrix. Furthermore, the partial derivatives of both the influencer characteristic dimension and the content structure dimension of this behavior are greater than a set threshold. Therefore, this behavior will be recorded. It is included in the set and used as input for the downstream strategy generation module.

[0067] The output of this step includes two parts: first, a subset of competitor behaviors after causal filtering. Each item represents a competitor's advertising action that the system identifies as having a real impact on our metrics; the second is the contribution matrix. This provides a ranking of the strength of policy factors and a reference for selecting target variables in the policy generation stage. All output data formats are consistent with the previous step, with no new variable names added, maintaining consistency between upstream and downstream structures.

[0068] S3: Based on the subset of competitor strategy behaviors and their corresponding sensitivity contribution, generate a structured response strategy that includes the type of recommended influencer, the structure tags of the recommended content, the recommended delivery platform, and the suggested delivery time period;

[0069] Specifically, the goal of this step is to identify the subset of significant competitor strategies and behaviors selected in the previous step. and its contribution matrix Building upon this foundation, a structured set of marketing response strategy variables is constructed to drive downstream modules such as influencer selection, content style matching, and ad scheduling, achieving a systemic transformation from "identifying strategic threats" to "generating strategic responses." This step is not merely a variable transformation, but rather introduces three types of adjustment mechanisms—targeted expression suppression, platform strategy coordination, and response inertia control—into the response mechanism to "competitive disturbances" in marketing scenarios.

[0070] All inputs for this step come from step two, and the variables include: first, the identified subset of causal behaviors. Each of them Vector structure and step one Maintaining consistency; secondly, the causal contribution matrix. Each element express The Middle The first dimension for marketing metrics Derivative sensitivity in each dimension. The variable structure is not expanded to maintain upstream consistency.

[0071] The response strategy variable is defined as four types of output items: This represents the vector of recommended influencer types (such as couples appearing in photos, working women, etc.). This indicates the structure tags of the recommended content (such as emotion-driven, product review). This indicates the recommended advertising platform (such as a text and image platform or a short video platform). This indicates the suggested delivery time (e.g., weekend lunchtime, weekday evening). These four types of variables will form the structured response strategy. In subsequent processes, these instructions flow into downstream calling modules in the form of structured instructions.

[0072] To achieve the above strategy generation, from multiple The common features are extracted from the behavior and aggregated to form policy variables. The response aggregation formula is designed as follows in this step:

[0073] ;

[0074] in Indicates the first One response strategy variable ( ), This is the set of behavioral dimensions associated with this policy variable; Indicates the first The variable values ​​of each dimension; Weight its causal influence; This is a competitor differentiation suppression term, used to prevent the system's response strategy from converging with the efficient strategies of competitors. It is calculated as the squared difference between the current behavioral variable and the recent average strategy of competitors in this dimension. The specific calculation method is as follows:

[0075] ;

[0076] in This represents the average of the same behavioral dimension across top-performing content from competitors over the past week. A larger value indicates that the behavior closely resembles a competitor's usual strategy, and the system will suppress this dimension when generating our response strategy. (Coefficient) Control the intensity of the compression.

[0077] Meanwhile, to avoid style fragmentation caused by an overemphasis on a single behavior in the response strategy, this step introduces a platform content consistency adjustment item. This parameter emphasizes the consistency of our strategy regarding past high-performing content on the current platform. It is calculated from the cosine of the angle between the current dimensional variable and the statistical median of historically high-performing content on the platform, enhancing the continuity of our response strategy and its platform adaptability. The coefficient... Control its adjustment intensity.

[0078] The final generated structured response variables It will be formatted into system commands with fixed fields and then called by the subsequent expert selection module and content generation module. For example, This can be interpreted as "Recommended mid-tier couples". The analysis defines it as a "warm and sentimental narrative content structure". Mapped to "Xiaohongshu platform priority" This is mapped to the time period of "18:00-21:00 Saturday to Sunday". The system will use this strategy structure as the basis for call scheduling control during execution.

[0079] S4: Perform multi-dimensional matching between the structured response strategy and the influencer profile information in the influencer resource library and the content templates in the content template library to generate an executable marketing resource combination and scheduling instruction;

[0080] Specifically, the core objective of this step is to apply the structured response strategy output in the previous step. This process transforms resources into executable resource scheduling actions, enabling coordinated scheduling of four key elements: influencers, content, time, and platform, thus constructing a complete closed-loop strategy response system. Unlike traditional methods that rely on manual decision-making, this step introduces a resource combination mechanism oriented towards strategy instruction structures, ensuring that response actions maintain consistency in direction while also satisfying the practical availability of deployed resources.

[0081] The input to this step is the set of policy variables output from step three, where The target influencer profile vector indicates which influencer types with specific tag characteristics should be selected, such as "couples taking photos together" or "posting a lot of content related to domestic products". The recommended content structure tags indicate the content expression methods that the system should prioritize, such as "emotional story type" and "product comparison type". Suggested platform identifiers, such as "text and image platform (Xiaohongshu)" or "short video platform (Douyin)". The recommended time window is usually coded as a combination of "intraday time period + weekly type", such as "weekend afternoon" or "weekday evening".

[0082] To complete the action, the system needs to call two resource libraries: one is the expert resource library. Each expert It possesses structured profile information, including basic account information (account number, platform coverage, follower level) and behavioral feature vectors. The behavioral characteristics consist of three parts: first, the influencer's tag encoding (such as whether there are wedding-related notes, whether it is a couple's account, etc.); second, the historical content style clustering vector; and third, the mean vector of recent interaction performance. These data are generated by the system through batch collection of influencer content via platform interfaces, followed by processing using a text classifier and image recognition network. For example, the system will extract all text and image notes from the influencer's content over the past 30 days, use a text and image sentiment style classification model to determine whether they belong to "warmth" or "information," and then combine this with the words appearing in the text and the cover image information to determine the content structure type and tone. Secondly, there is a content template library. Each content template Includes content structure tags Platform adaptation coding and available time window labels The data comes from high-conversion content fragments reused by the system and undergoes structured processing by a multimodal content structure annotation system.

[0083] To complete the mapping from "response strategy to deployment action", this step introduces a resource adaptation scoring function. Used to evaluate experts and content templates The degree of compatibility between the combined components and the strategy instructions. The function is defined as follows:

[0084] ;

[0085] In the formula, Represents the vector of an expert's profile; Represents a content structure vector; This represents the distribution vector of influencer performance intensity across different platforms, derived from the average exposure and interaction volume on each platform over the past seven days. This represents the activity level of influencers at different time periods, and calculates the average interaction performance of their content within 24 hours after it is published at different times. Represents the corresponding policy instruction variable; coefficient This represents the importance weight of each adaptation dimension, dynamically updated based on historical campaign performance. (Inner product operation) This represents the directional consistency between strategy variables and resource features; a higher score indicates a higher degree of matching between the strategy recommendation and the resource attributes.

[0086] For example, if To match the "couple photo shoot" style, the system will select from the expert database. Influencers who meet the following criteria: "two people appear on camera, the cover image includes two people of the opposite sex, and the historical tags contain keywords related to 'marriage'"; Content template for "Emotional Story" category The content must have tags indicating high emotional expression, and the system must determine that the cover image contains non-product elements and the copy contains emotionally charged narrative words. I recommend the "Xiaohongshu" platform; it has high user activity. There should be evidence that the influencer's interaction activity on Xiaohongshu over the past 7 days is higher than the baseline level; if If it is "weekend afternoon", then the system will determine The influencer's content posted during the weekend afternoon saw high engagement.

[0087] To further maintain brand strategy continuity and consistency in expression, the system also introduces a strategy inertia penalty item after the combined scoring. This term is defined as a combination of punishments that deviate significantly from past strategies, as follows:

[0088] ;

[0089] in This indicates the first of the system's past five responses. The historical mean of the strategy-like variable is used to measure the consistency between the current strategy and historical strategies. The system incorporates this value into the combinatorial optimization objective during execution to mitigate the risk of cognitive dissonance caused by style shifts.

[0090] Ultimately, the system ranks teams according to their overall scores. Sort and select the top Group As the execution unit for this strategy response, the combined results will be packaged into scheduling instructions, including fields such as influencer ID, content template number, recommended release time, and platform tags. These instructions will be uniformly passed to the content execution controller for scheduling generation and material distribution. The system can further optimize or replace the combinations based on budget constraints. If an influencer resource is temporarily unavailable, the system will re-evaluate alternative combinations with similar scores from the alternative resource pool and reissue the execution instructions.

[0091] In one or more embodiments, such as Figure 2 As shown, an intelligent marketing decision analysis system for real-time competitor strategy is disclosed, the system comprising:

[0092] The strategy behavior construction module is used to collect marketing content published by competitors on social media platforms, extract account numbers, influencer profile features, content structure tags, platform identifiers and publication time, and construct standardized strategy behavior vectors.

[0093] The causal identification module is used to assemble the strategy behavior vectors into a strategy behavior sequence in chronological order and align it with our marketing indicator sequence in time. Based on a sliding time window, combined with time decay weight and platform disturbance correction term, the module calculates the sensitivity contribution of each strategy behavior dimension to our marketing indicators, filters out strategy behaviors whose sensitivity contribution exceeds a preset impact threshold, and forms a subset of competitor strategy behaviors that have a significant impact on our marketing indicators.

[0094] The response strategy generation module is used to generate a structured response strategy that includes the type of recommended influencer, the structure tags of the recommended content, the recommended delivery platform, and the suggested delivery time period, based on the subset of the competitor's strategy behavior and the corresponding sensitivity contribution.

[0095] The resource matching and execution module is used to perform multi-dimensional matching between the structured response strategy and the influencer profile information in the influencer resource library and the content templates in the content template library to generate executable marketing resource combinations and scheduling instructions.

[0096] It is worth noting that the specific workflow of the intelligent marketing decision analysis system for real-time competitor strategy provided in this embodiment of the invention is the same as that of the intelligent marketing decision analysis method for real-time competitor strategy described in the above embodiments, and will not be repeated here.

[0097] This invention also provides an intelligent marketing decision analysis device for real-time competitor strategy, including a processor, a memory, and a computer program stored in the memory and configured to be executed by the processor. When the processor executes the computer program, it implements the steps described in the above embodiments of the intelligent marketing decision analysis method for real-time competitor strategy. Figure 1 The steps S1 to S4 described above; or, when the processor executes the computer program, it implements the functions of each module in the above system embodiments.

[0098] For example, the computer program may be divided into one or more modules, which are stored in the memory and executed by the processor to complete the present invention. The one or more modules may be a series of computer program instruction segments capable of performing specific functions, which describe the execution process of the computer program in the intelligent marketing decision analysis device for real-time competitor strategy.

[0099] The intelligent marketing decision analysis device for real-time competitor strategy can be a computing device such as a desktop computer, laptop, handheld computer, or cloud server. This device may include, but is not limited to, a processor and memory. Those skilled in the art will understand that the intelligent marketing decision analysis device for real-time competitor strategy may also include input / output devices, network access devices, buses, etc.

[0100] The processor can be a Central Processing Unit (CPU), or other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASACs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The general-purpose processor can be a microprocessor or any conventional processor. This processor is the control center of the intelligent marketing decision analysis device for real-time competitor strategy, connecting all parts of the device via various interfaces and lines.

[0101] The memory can be used to store the computer programs and / or modules. The processor, by running or executing the computer programs and / or modules stored in the memory and calling the data stored in the memory, realizes various functions of the intelligent marketing decision analysis device for real-time competitor strategy. The memory may mainly include a program storage area and a data storage area. The program storage area may store the operating system, at least one application program required for a function, etc.; the data storage area may store data created based on the operation of the air conditioner controller, etc. In addition, the memory may include high-speed random access memory, and may also include non-volatile memory, such as hard disk, memory, plug-in hard disk, smart media card (SMC), secure digital card (SD card), flash card, at least one disk storage device, flash memory device, or other volatile solid-state storage devices.

[0102] The modules integrated into the intelligent marketing decision analysis device for real-time competitor strategies, if implemented as software functional units and sold or used as independent products, can be stored in a computer-readable storage medium. Based on this understanding, all or part of the processes in the above embodiments of the present invention can also be implemented by a computer program instructing related hardware. The computer program can be stored in a computer-readable storage medium, and when executed by a processor, it can implement the steps of the various method embodiments described above. The computer program includes computer program code, which can be in the form of source code, object code, executable files, or certain intermediate forms. The computer-readable medium can include: any entity or device capable of carrying the computer program code, recording media, USB flash drives, portable hard drives, magnetic disks, optical disks, computer memory, read-only memory (ROM), random access memory (RAM), electrical carrier signals, telecommunication signals, and software distribution media, etc.

[0103] Those skilled in the art will understand that all or part of the processes in the above embodiments can be implemented by a computer program instructing related hardware. The program can be stored in a computer-readable storage medium, and when executed, it can include the processes of the embodiments of the above methods. The storage medium can be a magnetic disk, optical disk, read-only memory (ROM), or random access memory (RAM), etc.

[0104] The above description represents the preferred embodiments of the present invention. It should be noted that those skilled in the art can make various improvements and modifications without departing from the principles of the present invention, and these improvements and modifications are also considered to be within the scope of protection of the present invention.

Claims

1. An intelligent marketing decision analysis method for real-time competitor strategies, characterized in that, The method includes: Collect marketing content published by competitors on social media platforms, extract account numbers, influencer profile characteristics, content structure tags, platform identifiers, and publication times, and construct standardized strategy behavior vectors; The strategy behavior vectors are arranged in chronological order to form a strategy behavior sequence, which is then time-sequentially aligned with our marketing indicator sequence. Based on a sliding time window, and combining time decay weights and platform disturbance correction terms, the sensitivity contribution of each strategic behavior dimension to our marketing metrics is calculated. Strategic behaviors with sensitivity contributions exceeding a preset impact threshold are selected to form a subset of competitor strategic behaviors that significantly impact our marketing metrics. The sensitivity contribution is calculated using a strategic feature contribution matrix, specifically: ； This matrix represents the position at lag. At that moment, the first strategic action The first dimension variable affects the marketing metrics. The degree of influence of each dimension, that is, the sensitivity of each behavioral variable to the indicator result, where This represents the input of the k-th dimension variable in the policy behavior vector at time lag position i. The model represents time intervals. The predicted values ​​for our marketing metrics are as follows: ; In this formula, The model represents time intervals. Our projected marketing metrics, Indicates the position of behavioral lag Learning weights This is a time decay term, representing the time-limited effect of the behavior. This is an adjustable coefficient; for The policy behavior vector at each time step; This is a platform disturbance correction item, numbered by the platform. The resulting perturbation vector is of fixed length. Its adjustment coefficient; Based on the subset of competitor strategies and behaviors and their corresponding sensitivity contributions, a structured response strategy is generated that includes the type of recommended influencer, the structure tags of the recommended content, the recommended delivery platform, and the suggested delivery time period. The structured response strategy is matched in multiple dimensions with the influencer profile information in the influencer resource library and the content templates in the content template library to generate executable marketing resource combinations and scheduling instructions.

2. The intelligent marketing decision analysis method for real-time competitor strategy according to claim 1, characterized in that, The steps for constructing standardized policy behavior vectors include: Normalize the characteristics of the influencer profile; Maintain a five-dimensional classification tag vector format for content structure tags; Discrete embedding encoding is used for platform identifiers; Convert the release time into two one-hot encoded variables: intraday time period and weekday type; The variables processed above are combined into a strategy behavior vector with a total number of dimensions not exceeding fifty.

3. The intelligent marketing decision analysis method for real-time competitor strategy according to claim 1, characterized in that, The steps for generating the structured response strategy include: Introduce a competitor differentiation suppression term to suppress response dimensions that converge with the recent high-performing strategies of competitors; Introduce a platform consistency adjustment item to enhance the consistency between the response strategy and the historically high-performing content of the current platform; The structured response strategy is generated by aggregating factors based on causal impact weights, competitor difference suppression terms, and platform consistency adjustment terms. The causal influence weights are derived from the strategy feature contribution matrix in the preceding steps. Specifically, they are the derivative sensitivity of each strategy behavior dimension in the competitor strategy behavior subset to the marketing metrics, used to characterize the strength of the influence of that dimension on the metric changes.

4. The intelligent marketing decision analysis method for real-time competitor strategy according to claim 1, characterized in that, The profile information of each influencer in the influencer resource library includes: account number, platform coverage, fan level, influencer tag code, historical content style clustering vector, and recent interaction performance mean vector.

5. The intelligent marketing decision analysis method for real-time competitor strategy according to claim 1, characterized in that, Each content template in the content template library includes: content structure tags, platform adaptation encoding, and available time window tags.

6. The intelligent marketing decision analysis method for real-time competitor strategy according to claim 1, characterized in that, The steps of the multidimensional matching include: Calculate the matching degree between influencer profile information and recommended influencer types; Calculate the matching degree between the content template and the recommended content structure tags; Combine the activity level of influencers on the platform with the matching degree of the recommended placement platform; Combine the match between the influencer's activity level and the suggested ad placement time slot; A resource adaptation score is generated by combining various matching factors.

7. The intelligent marketing decision analysis method for real-time competitor strategy according to claim 6, characterized in that, Following the step of generating executable marketing resource combinations and scheduling instructions, the method further includes: Introduce a strategy inertia penalty term to evaluate the consistency between the current resource combination and historical response strategies; The candidate resource combinations are ranked based on the combined results of the resource adaptation score and the strategy inertia penalty. Select the resource combination that ranks highest as the final execution unit.

8. The intelligent marketing decision analysis method for real-time competitor strategy according to claim 1, characterized in that, When the talent resources are unavailable, re-evaluate the alternative combinations with similar scores from the alternative resource pool and update the scheduling instructions.

9. The intelligent marketing decision analysis method for real-time competitor strategy according to claim 1, characterized in that, The preset impact threshold is dynamically determined based on the historical fluctuation range of marketing indicators and the statistical distribution of competitor behavior.

10. An intelligent marketing decision analysis system for real-time competitor strategy, characterized in that: The system includes: The strategy behavior construction module is used to collect marketing content published by competitors on social media platforms, extract account numbers, influencer profile features, content structure tags, platform identifiers and publication time, and construct standardized strategy behavior vectors. The causal identification module is used to assemble the strategy behavior vectors into a strategy behavior sequence in chronological order and align it with our marketing metric sequence in time. Based on a sliding time window, combined with time decay weights and platform disturbance correction terms, it calculates the sensitivity contribution of each strategy behavior dimension to our marketing metrics, and filters out strategy behaviors with sensitivity contributions exceeding a preset impact threshold, forming a subset of competitor strategy behaviors that have a significant impact on our marketing metrics. The sensitivity contribution is calculated through a strategy feature contribution matrix, specifically: ； This matrix represents the position at lag. At that moment, the first strategic action The first dimension variable affects the marketing metrics. The degree of influence of each dimension, that is, the sensitivity of each behavioral variable to the indicator result, where This represents the input of the k-th dimension variable in the policy behavior vector at time lag position i. The model represents time intervals. The predicted values ​​for our marketing metrics are as follows: ; In this formula, The model represents time intervals. Our projected marketing metrics, Indicates the position of behavioral lag Learning weights This is a time decay term, representing the time-limited effect of the behavior. This is an adjustable coefficient; for The policy behavior vector at each time step; This is a platform disturbance correction item, numbered by the platform. The resulting perturbation vector is of fixed length. Its adjustment coefficient; The response strategy generation module is used to generate a structured response strategy that includes the type of recommended influencer, the structure tags of the recommended content, the recommended delivery platform, and the suggested delivery time period, based on the subset of the competitor's strategy behavior and the corresponding sensitivity contribution. The resource matching and execution module is used to perform multi-dimensional matching between the structured response strategy and the influencer profile information in the influencer resource library and the content templates in the content template library to generate executable marketing resource combinations and scheduling instructions.

Images