AI-powered intelligent marketing multi-scenario competitor analysis methods

Abstract

This invention discloses a multi-scenario competitive analysis method for AI-powered intelligent marketing, belonging to the field of competitive analysis technology. The method includes: S1, collecting multi-source data from the entire AI-powered intelligent marketing process and preprocessing it to obtain a structured feature matrix X; S2, dividing the multiple scenarios of AI-powered intelligent marketing into 5 core scenarios and key indicators; S3, combining real-time scenario popularity and indicator importance, proposing a scenario-indicator dynamic weight calculation method to generate a scenario-indicator dynamic weight matrix W; S4, using a Transformer model, inputting the feature matrix X and the dynamic weight matrix W, and outputting three tasks; S5, based on the advantage scores output by the competitive advantage assessment task, constructing an advantage heatmap, and generating differentiated strategies. The multi-scenario competitive analysis method for AI-powered intelligent marketing provided by this invention achieves accurate adaptation to multiple scenarios, better supporting enterprise marketing decisions.

Description

Technical Field

[0001] The present invention relates to the technical field of competitor analysis, and in particular to a method for analyzing competitors in multiple scenarios of AI intelligent marketing. Background Art

[0002] In the traditional field of competitor analysis in AI intelligent marketing, competitor analysis often uses fixed index weights to evaluate the performance of competitors in different scenarios. However, the scenarios involved in AI intelligent marketing are extremely rich, such as traffic acquisition scenarios, content conversion scenarios, promotional activity scenarios, etc. In different scenarios, the impact degrees of various indicators of competitors (such as the coverage rate of investment channels, the completion rate of content, the discount strength, etc.) on the marketing effect are significantly different. At the same time, the popularity of the scenarios themselves will also change in real time with factors such as time (such as approaching major promotion nodes) and industry dynamics (such as policy changes). The traditional analysis method with fixed weights cannot accurately capture the dynamic relationship between scenarios and indicators, resulting in a large deviation between the competitor analysis results and the actual marketing effect, and it is difficult to effectively provide strong support for enterprises to formulate differential competition strategies. Summary of the Invention

[0003] The purpose of the present invention is to provide a method for analyzing competitors in multiple scenarios of AI intelligent marketing to solve the problems existing in the above background art.

[0004] To achieve the above purpose, the present invention provides a method for analyzing competitors in multiple scenarios of AI intelligent marketing, including the following steps:

[0005] S1. For the whole-link process of AI intelligent marketing, collect multi-source data including competitor basic data, user behavior data and environmental dynamic data, and preprocess the collected multi-source data to obtain a structured feature matrix ;

[0006] S2. Divide the multiple scenarios of AI intelligent marketing into 5 core scenarios, and give the key indicators corresponding to each scenario;

[0007] S3. Combine the real-time scenario popularity and indicator importance, propose a method for calculating the dynamic weights of scenarios and indicators, and generate a dynamic weight matrix of scenarios and indicators , and realize the differential weight allocation of competitor indicators in different scenarios;

[0008] S4. Use the Transformer model, input the feature matrix in step S1 and the dynamic weight matrix in step S3 , and output three tasks: competitor strategy classification, competitor effect prediction and competitor advantage evaluation;

[0009] S5. Based on the advantage scores output by the competitor advantage evaluation task, construct an advantage heat map and generate a differential strategy.

[0010] Preferably, the preprocessing in step S1 specifically includes:

[0011] First, multi-source data is processed using integrity, uniqueness, validity, consistency, and outlier dynamic filtering. The formula for dynamic outlier filtering is as follows:

[0012] ;

[0013] in, The original data field value; The average value of the field; Standard deviation;

[0014] Then, for data with different dimensions, max-min standardization is applied:

[0015] ;

[0016] in, The minimum value of the field containing the data; This represents the maximum value of the field containing the data.

[0017] Finally, we construct marketing scenario-specific features to form a structured feature matrix. ,in, For the number of competing products, For feature dimensions.

[0018] Preferably, in step S2, the multiple scenarios of AI intelligent marketing are divided into 5 core scenarios, including traffic acquisition scenario, content conversion scenario, promotional activity scenario, user retention scenario, and public opinion influence scenario. The key indicators corresponding to the 5 core scenarios are: channel coverage, exposure cost, click conversion rate; content completion rate, product mention frequency, live broadcast conversion rate; discount strength, minimum spend threshold, gift value, number of participants; number of membership benefits, repurchase discount strength, user retention rate; public opinion sentiment value, public opinion spread speed, and public opinion influence coefficient on conversion rate.

[0019] Preferably, step S3 proposes a method for calculating the dynamic weights of scenarios and indicators, and generates a dynamic weight matrix of scenarios and indicators. Specifically, it includes:

[0020] S31. Calculate scenario activity based on industry data to obtain scenario popularity weight. ;

[0021] S32. Calculate the index using the entropy weight method. In the scene The weights assigned to each indicator determine its importance. ;

[0022] S33. Based on steps S31 and S32, obtain the elements of the scenario-index dynamic weight matrix. Finally, a scenario-metric dynamic weight matrix is ​​generated. ,in, For the number of scenes, This represents the total number of indicators.

[0023] Preferably, the scene heat weight in step S31 The calculation formula is:

[0024] ;

[0025] in, These are the weighting coefficients; The default number of days for the promotional period; The number of days between the current date and the most recent major promotional event. For competitors in the scenario The percentage of investment.

[0026] Preferably, step S32 specifically includes:

[0027] S321, Calculation Indicators In the scene The following proportion: ,in, For competitors In the scene Indicators value, The number of competing products;

[0028] S322, Calculation Indicators Information entropy: ;

[0029] S323, Calculate the importance weight of indicators : ,in, For the scene The total number of indicators below.

[0030] Preferably, in step S4, the competitor strategy classification is used to determine the strategy type of the competitors in each scenario; the competitor effect prediction is used to predict the key indicator values ​​of the competitors in each scenario within a certain period of time in the future; and the competitor advantage assessment is used to output the competitive advantage score of the competitors in each scenario.

[0031] Preferably, the Transformer model using a multi-task attention mechanism in step S4 incorporates a scene-index attention mechanism in its attention layer to calculate the correlation between features and the scene, expressed as:

[0032] Softmax ;

[0033] in, For query matrix; The key matrix; It is a value matrix; Key matrix The dimension; Softmax is the activation function; and ; ; This enables focused attention on high-weight scenarios and metrics.

[0034] A weighted loss function is used to balance the training priorities of the three tasks:

[0035] ;

[0036] in, Cross-entropy loss for competitor strategy classification tasks; The mean absolute error (MAE) loss for the competitor performance prediction task; The mean squared error (MSE) loss for the competitive advantage assessment task; , , All are loss weighting coefficients; , and The expressions are as follows:

[0037] ;

[0038] ;

[0039] ;

[0040] in, For the scene Next strategy The true label, To predict probabilities, The number of strategy categories; For the scene Next The true index value of the day, For predicted values, The predicted time in days; For competitors In the scene True advantage score, For the predicted score.

[0041] Preferably, in step S5, the advantage score is based on the output. Construct a heatmap of competitive advantages, represented as follows:

[0042] ;

[0043] in, For competitors Advantage score in all scenarios; For competitors The maximum advantage score across all scenarios; For competitors The minimum advantage score across all scenarios.

[0044] Preferably, the differentiation strategy in step S5 calculates the optimal strategy for the enterprise in different scenarios by constructing a cross-scenario decision-making benefit function:

[0045] ;

[0046] in, For the scene The following strategy Expected returns; Adopting strategies for businesses In the scene The predicted value of the indicator; For competitors in the scenario Indicators average value; For the scene The following strategy Cost; by traversing all strategies ,choose The biggest strategy for enterprises in a given scenario The optimal competitive strategy under the given conditions.

[0047] Therefore, the above-mentioned AI-powered intelligent marketing multi-scenario competitor analysis method adopted in this invention has the following beneficial effects:

[0048] (1) By constructing a scenario-indicator dynamic weight matrix, the indicator weights can be adjusted in real time according to the popularity of different scenarios and the importance of each indicator in the scenario;

[0049] (2) Enterprises can evaluate competitors’ strategies more scientifically based on the weight information provided by the matrix, and then make more targeted and effective marketing decisions.

[0050] The technical solution of the present invention will be further described in detail below with reference to the accompanying drawings and embodiments. Attached Figure Description

[0051] Figure 1 This is a flowchart of the AI-powered intelligent marketing multi-scenario competitor analysis method of the present invention;

[0052] Figure 2 This is a heatmap showing the advantages of three competing products in this embodiment of the invention. Detailed Implementation

[0053] The following detailed description of embodiments of the invention provided in the accompanying drawings is not intended to limit the scope of the claimed invention, but merely to illustrate selected embodiments of the invention. All other embodiments obtained by those skilled in the art based on the embodiments of the invention without inventive effort are within the scope of protection of the invention.

[0054] Please see Figure 1 AI-powered intelligent marketing multi-scenario competitor analysis methods include the following steps:

[0055] S1. For the entire process of AI-powered intelligent marketing, from "user touchpoints to content dissemination, conversion, and user retention," multi-source data is collected, including competitor basic data, user behavior data, and dynamic environmental data. The collected multi-source data is then preprocessed to obtain a structured feature matrix. Specifically:

[0056] First, multi-source data is processed using integrity, uniqueness, validity, consistency, and outlier dynamic filtering. The formula for dynamic outlier filtering is as follows:

[0057] ;

[0058] in, The original data field value; The average value of the field; Standard deviation;

[0059] Then, for data with different dimensions, max-min standardization is applied:

[0060] ;

[0061] in, The minimum value of the field containing the data; This represents the maximum value of the field containing the data.

[0062] Finally, we construct marketing scenario-specific features to form a structured feature matrix. ,in, For the number of competing products, For feature dimensions.

[0063] S2. AI-powered intelligent marketing scenarios are divided into 5 core scenarios: traffic acquisition, content conversion, promotional activities, user retention, and public opinion influence. The key indicators corresponding to these 5 core scenarios are: channel coverage, exposure cost, click-through rate; content completion rate, product mention frequency, live stream conversion rate; discount strength, minimum spend threshold, gift value, number of participants; number of membership benefits, repeat purchase discount strength, user retention rate; public opinion sentiment value, public opinion spread speed, and the impact coefficient of public opinion on conversion rate.

[0064] S3. Combining real-time scene popularity and indicator importance, a dynamic weight calculation method for scene-indicator is proposed to generate a dynamic weight matrix for scene-indicator. To achieve differentiated weight allocation of competitor metrics in different scenarios, specifically:

[0065] S31. Calculate scenario activity based on industry data to obtain scenario popularity weight. :

[0066] ;

[0067] in, These are the weighting coefficients; The default number of days for the promotional period; The number of days between the current date and the most recent major promotional event. For competitors in the scenario The percentage of investment.

[0068] S32. Calculate the index using the entropy weight method. In the scene The weights assigned to each indicator determine its importance. Specifically, it includes:

[0069] S321, Calculation Indicators In the scene The following proportion: ,in, For competitors In the scene Indicators value, The number of competing products;

[0070] S322, Calculation Indicators Information entropy: ;

[0071] S323, Calculate the importance weight of indicators : ,in, For the scene The total number of indicators below.

[0072] S33. Based on steps S31 and S32, obtain the elements of the scenario-index dynamic weight matrix. Finally, a scenario-metric dynamic weight matrix is ​​generated. ,in, For the number of scenes, This represents the total number of indicators.

[0073] S4. Using a Transformer model with a multi-task attention mechanism, input the feature matrix from step S1. and the dynamic weight matrix of step S3 It outputs three tasks: competitor strategy classification, competitor effect prediction, and competitor advantage assessment; its attention layer introduces a scene-indicator attention mechanism to calculate the correlation between features and scenes, represented as:

[0074] Softmax ;

[0075] in, For query matrix; The key matrix; It is a value matrix; Key matrix The dimension; Softmax is the activation function; and ; ; This enables focused attention on high-weight scenarios and metrics.

[0076] A weighted loss function is used to balance the training priorities of the three tasks:

[0077] ;

[0078] in, Cross-entropy loss for competitor strategy classification tasks; The mean absolute error (MAE) loss for the competitor performance prediction task; The mean squared error (MSE) loss for the competitive advantage assessment task; , , These are all loss weighting coefficients, which can be dynamically adjusted according to the needs of the enterprise. , and The expressions are as follows:

[0079] ;

[0080] ;

[0081] ;

[0082] in, For the scene Next strategy The true label, To predict probabilities, The number of strategy categories; For the scene Next The true index value of the day, For predicted values, The predicted time in days; For competitors In the scene True advantage score, The competitive strategy classification is used to determine the strategy type of competitors in each scenario; the competitive effect prediction is used to predict the key indicator values ​​of competitors in each scenario within a certain period of time in the future; and the competitive advantage assessment is used to output the competitive advantage score of competitors in each scenario.

[0083] S5, Output-Based Advantage Score Construct a competitor advantage heatmap. The darker the color of the heatmap, the stronger the competitor's advantage in that scenario. This helps companies quickly identify competitors' weaknesses, represented as follows:

[0084] ;

[0085] in, For competitors Advantage score in all scenarios; For competitors The maximum advantage score across all scenarios; For competitors The minimum advantage score across all scenarios.

[0086] Differentiation strategies calculate the optimal strategy for an enterprise in different scenarios by constructing a cross-scenario decision-making benefit function:

[0087] ;

[0088] in, For the scene The following strategy Expected returns; Adopting strategies for businesses In the scene The predicted value of the indicator; For competitors in the scenario Indicators average value; For the scene The following strategy Cost; by traversing all strategies ,choose The biggest strategy for enterprises in a given scenario The optimal competitive strategy under the given conditions.

[0089] To verify the effectiveness of the above solution, this embodiment takes the AI ​​live-streaming marketing scenario in the beauty industry as an example, selecting three core competitors (A / B / C) and analyzing them using this invention:

[0090] Data collection: Crawling competitor's Douyin live streaming data (live stream duration, product conversion rate, interaction rate) and e-commerce promotion data (discount level, gift information) for the past 6 months; calculating the scenario popularity under the "content conversion scenario". (With the big sale approaching) Indicator weighting , , ;

[0091] The predicted conversion rate of competitor A's live stream over the next 7 days is 12.5% ​​(higher than competitors B / C's 8.2% and 9.1% respectively), with an advantage score of 8.7 (out of 10) in the "content conversion scenario".

[0092] Based on the revenue function, we recommend that businesses adopt a strategy of "inserting a 10-minute limited-time flash sale during the live stream (cost C=5000 yuan)", with an expected revenue of This is higher than the 10.2% of the "conventional live streaming" strategy. Verification results show that this solution can effectively identify competitive advantage scenarios, as shown in the advantage heatmap below. Figure 2 As shown, the output strategy increased the enterprise's live streaming conversion rate by 35%, meeting the multi-scenario analysis needs of AI intelligent marketing.

[0093] Therefore, the present invention adopts the above-mentioned AI intelligent marketing multi-scenario competitive product analysis method. Compared with the shortcomings of traditional fixed weights that cannot adapt to multiple scenarios and have poor dynamic adaptability, this method achieves accurate adaptation to multiple scenarios, improves the scientific nature of analysis and decision-making, and enhances dynamic adaptability by constructing a scenario-indicator dynamic weight matrix, which can better support enterprise marketing decisions.

[0094] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit them. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can still be made to the technical solutions of the present invention, and these modifications or equivalent substitutions cannot cause the modified technical solutions to deviate from the spirit and scope of the technical solutions of the present invention.

Claims

1. An AI-powered intelligent marketing multi-scenario competitor analysis method, characterized in that, Includes the following steps: S1. For the entire AI-powered intelligent marketing process, collect multi-source data including competitor basic data, user behavior data, and environmental dynamic data, and preprocess the collected multi-source data to obtain a structured feature matrix. ; S2. Divide the multiple scenarios of AI intelligent marketing into 5 core scenarios and give the key indicators corresponding to each scenario; S3. Combining real-time scene popularity and indicator importance, a dynamic weight calculation method for scene-indicator is proposed to generate a dynamic weight matrix for scene-indicator. This enables differentiated weight allocation of competitor metrics in different scenarios; S4. Using the Transformer model, input the feature matrix from step S1. and the dynamic weight matrix of step S3 The task is to output three components: competitor strategy classification, competitor effect prediction, and competitor advantage assessment. S5. Based on the advantage score output from the competitor advantage assessment task, construct an advantage heatmap and generate a differentiation strategy; The preprocessing in step S1 specifically includes: First, multi-source data is processed using integrity, uniqueness, validity, consistency, and outlier dynamic filtering. The formula for dynamic outlier filtering is as follows: ； in, The original data field value; The average value of the field; Standard deviation; Then, for data with different dimensions, max-min standardization is applied: ； in, The minimum value of the field containing the data; This represents the maximum value of the field containing the data. Finally, we construct marketing scenario-specific features to form a structured feature matrix. ,in, For the number of competing products, For feature dimensions; Step S3 proposes a method for calculating the dynamic weights of scenarios and indicators, and generates a dynamic weight matrix of scenarios and indicators. Specifically, it includes: S31. Calculate scenario activity based on industry data to obtain scenario popularity weight. ; S32. Calculate the index using the entropy weight method. In the scene The weights assigned to each indicator determine its importance. ; S33. Based on steps S31 and S32, obtain the elements of the scenario-index dynamic weight matrix. Finally, a dynamic weight matrix of scenarios and metrics is generated. ,in, For the number of scenes, The total number of indicators; Step S4 employs a Transformer model with a multi-task attention mechanism. Its attention layer introduces a scene-metric attention mechanism to calculate the correlation between features and the scene, expressed as: Softmax ; in, For query matrix; The key matrix; It is a value matrix; Key matrix The dimension; Softmax is the activation function; and ; ; This enables focused attention on high-weight scenarios and metrics. A weighted loss function is used to balance the training priorities of the three tasks: ； in, Cross-entropy loss for competitor strategy classification tasks; The mean absolute error (MAE) loss for the competitor performance prediction task; The mean squared error (MSE) loss for the competitive advantage assessment task; , , All are loss weighting coefficients; , and The expressions are as follows: ； ； ； in, For the scene Next strategy The true label, To predict probabilities, The number of strategy categories; For the scene Next The true index value of the day, For predicted values, The predicted time in days; For competitors In the scene The true advantage score, To predict the score, The number of competing products; In step S5, the advantage score is based on the output. Construct a heatmap of competitive advantages, represented as follows: ； in, For competitors Advantage score in all scenarios; For competitors The maximum advantage score across all scenarios; For competitors Minimum advantage score across all scenarios; The differentiation strategy in step S5 calculates the optimal strategy for the enterprise in different scenarios by constructing a cross-scenario decision benefit function: ； in, For the scene The following strategy is adopted Expected returns; Adopting strategies for businesses In the scene The predicted value of the indicator; For competitors in the scenario Indicators average value; For the scene The following strategy is adopted The cost; For the scene The total number of indicators; by traversing all strategies ,choose The biggest strategy for enterprises in a given scenario The optimal competitive strategy under the given conditions.

2. The AI-powered intelligent marketing multi-scenario competitor analysis method according to claim 1, characterized in that: Step S2 divides the multiple scenarios of AI-powered intelligent marketing into five core scenarios: traffic acquisition, content conversion, promotional activities, user retention, and public opinion influence. The key indicators corresponding to these five core scenarios are: channel coverage, exposure cost, click-through rate; content completion rate, product mention frequency, live stream conversion rate; discount strength, minimum spend threshold, gift value, number of participants; number of membership benefits, repeat purchase discount strength, user retention rate; public opinion sentiment value, public opinion spread speed, and the impact coefficient of public opinion on conversion rate.

3. The AI-powered intelligent marketing multi-scenario competitor analysis method according to claim 1, characterized in that, Scene heat weighting in step S31 The calculation formula is: ； in, These are the weighting coefficients; The default number of days for the promotional period; The number of days between the current date and the most recent major promotional event. For competitors in the scenario The percentage of investment.

4. The AI-powered intelligent marketing multi-scenario competitor analysis method according to claim 3, characterized in that, Step S32 specifically includes: S321, Calculation Indicators In the scene The following proportion: ,in, For competitors In the scene Indicators value, The number of competing products; S322, Calculation Indicators Information entropy: ; S323, Calculate the importance weight of indicators : ,in, For the scene The total number of indicators below.

5. The AI-powered intelligent marketing multi-scenario competitor analysis method according to claim 1, characterized in that: The competitor strategy classification in step S4 is used to determine the strategy type of competitors in each scenario; the competitor effect prediction is used to predict the key indicator values ​​of competitors in each scenario within a certain period of time in the future; and the competitor advantage assessment is used to output the competitive advantage score of competitors in each scenario.

Images