Promotion method, system and device based on APP reservation data and medium

Abstract

The invention is suitable for the field of APP promotion, and discloses a promotion method, system, device and medium based on APP reservation data, the method comprises the following steps: firstly, collecting and analyzing user behavior data in a promotion channel, constructing a channel effect evaluation model to evaluate an input-output ratio, and dynamically adjusting a budget according to an evaluation result; then, extracting operation characteristics of the reservation user by using the channel effect evaluation model, and constructing a user downloading intention prediction model to accurately calculate a downloading probability value of the reservation user; in addition, parameters such as crowd attributes, creative themes and delivery time periods in competitive product promotion are further analyzed, and advertisement delivery combinations are constructed and optimized in combination with competitive product APP functions; time data is introduced when the advertisement putting combination is optimized, and a correlation coefficient between the time data and the APP downloading amount is calculated, so that an advertisement putting strategy is adjusted more finely; meanwhile, evaluation texts of the users on the advertisements are collected, and advertisement creativity parameters are optimized based on text analysis. And the APP popularization efficiency and effect are effectively improved.

Description

Technical Field

[0001] The present invention relates to the field of APP promotion, and in particular to a promotion method, system, device and medium based on APP reservation data. Background Art

[0002] With the rapid development of mobile Internet, APP promotion has become a key means for enterprises to acquire users and enhance brand influence. However, traditional APP promotion methods often rely on a large amount of advertising and extensive channel coverage, lacking accurate user positioning and effect evaluation mechanisms, resulting in high promotion costs and low efficiency.

[0003] In order to overcome this problem, the industry has begun to explore data-based APP promotion methods. Among them, user behavior data, as an important indicator reflecting user preferences and needs, is of great significance for guiding the formulation and optimization of promotion strategies. However, how to effectively collect, analyze and use these user behavior data, build a scientific channel effect evaluation model and user download intention prediction model, and how to optimize advertising combinations and creative parameters based on competitive product promotion and user feedback are still urgent issues to be solved in the current APP promotion field. Summary of the invention

[0004] The present invention aims to solve at least one of the technical problems existing in the prior art. To this end, the present invention proposes a promotion method, system, device and medium based on APP reservation data.

[0005] In the first aspect, the present application proposes a promotion method based on APP reservation data, including: obtaining user behavior data in the promotion channel, building a channel evaluation index system based on the user behavior data, establishing a channel effect evaluation model, obtaining input-output data based on the channel effect evaluation model, and triggering a budget adjustment when the input-output ratio is lower than a preset threshold;

[0006] Based on the channel effect evaluation model, extract the operation characteristics of the reserved user, build a user download willingness prediction model based on the operation characteristics, and calculate the download probability value of the reserved user according to the user download willingness prediction model;

[0007] Analyze the crowd attributes, creative themes and delivery time parameters in the competitive product promotion, build an advertising delivery combination based on the functions of the competitive product APP, and optimize the parameters of the advertising delivery combination;

[0008] Obtaining the user's evaluation text of the advertisement, and optimizing the creative parameters of the advertisement based on the evaluation text;

[0009] When optimizing the parameters of the advertising delivery combination, time data is introduced, a correlation coefficient between the time data and the APP download volume is calculated, and the parameters of the advertising delivery combination are adjusted based on the correlation coefficient.

[0010] More specifically, in the above technical solution, user behavior data includes clicks, registrations, downloads and payments; a channel evaluation index system is constructed based on the user behavior data, a channel effect evaluation model is established, input-output data is obtained based on the channel effect evaluation model, and a budget adjustment is triggered when the input-output ratio is lower than a preset threshold, specifically including:

[0011] Cleaning the user behavior data to obtain a basic data set;

[0012] Performing data preprocessing on the basic data set, including performing numerical normalization, performing one-hot encoding on categorical variables, and converting the categorical variables into numerical variables;

[0013] Based on the preprocessed data, a channel evaluation index system is constructed, and the channel evaluation index system is used as an input feature of the channel effect evaluation model, wherein the channel evaluation index system includes conversion rate, next-day retention rate, 7-day retention rate, and average order value;

[0014] In the channel effect evaluation model, the channel input cost and evaluation index are used as independent variables, the channel output income is used as the dependent variable, and the model parameters are obtained by training;

[0015] Through the trained model, for each promotion channel, input its input cost and various indicator values, predict its output revenue, and calculate the input-output ratio.

[0016] More specifically, in the above technical solution, based on the channel effect evaluation model, the operation characteristics of the reserved user are extracted, based on the operation characteristics, a user download willingness prediction model is constructed, and the download probability value of the reserved user is calculated according to the user download willingness prediction model, which specifically includes:

[0017] Based on user behavior data, obtain the device brand, operating system, network environment, geographic location, visit frequency and stay time of the reservation user, convert non-numeric features into numerical representation, and construct a user feature vector;

[0018] Taking the user feature vector as input and whether the user downloads the App as the target variable, a user download intention prediction model is trained;

[0019] The user download willingness prediction model is evaluated using a cross-validation method to predict the download probability value of each reserved user, and the users are divided into three levels: high willingness, medium willingness and low willingness;

[0020] For high-willing users, push the App installation package download link; for medium-willing users, provide App function introductions and preferential incentives; for low-willing users, reduce the push frequency.

[0021] More specifically, in the above technical solution, the crowd attributes, creative themes and delivery time parameters in the competitive product promotion are analyzed, an advertising delivery combination is constructed in combination with the functions of the APP, and the parameters of the advertising delivery combination are optimized, specifically including:

[0022] Collect competitive product advertising data, including the attributes of the audience in the promotion, creative themes, and delivery time parameters;

[0023] According to the functions of the APP and the characteristics of different groups of people, multiple groups of advertising creative themes and copywriting are designed, and an advertising creative library is constructed. For each group of people, materials are selected from the advertising creative library to generate an advertising delivery combination.

[0024] More specifically, in the above technical solution, obtaining the user's evaluation text of the advertisement and optimizing the creative parameters of the advertisement based on the evaluation text specifically include:

[0025] Obtaining the user's evaluation text of the advertisement, and preprocessing the evaluation text using natural language processing technology;

[0026] Dividing the preprocessed text data into a training set and a test set, inputting the training set and the test set into a convolutional neural network model for fine-tuning training, and evaluating the performance of the convolutional neural network model on the test set, obtaining the sentiment tendency of the evaluation text through the convolutional neural network model, and judging whether the user's evaluation of the advertising creativity is positive, neutral or negative according to the sentiment tendency;

[0027] According to the evaluation texts with different emotional tendencies, a vectorized representation reflecting user feedback is constructed, the vector of the user feedback is used as an optimization target, a fitness function of a genetic algorithm is designed so that the fitness function is positively correlated with user satisfaction, the fitness function is input into the genetic algorithm, and the creative parameters of the advertisement are optimized, the creative parameters of the advertisement including copy, pictures and layout.

[0028] More specifically, in the above technical solution, the time data includes holidays, hot events and APP downloads;

[0029] Introducing time data, calculating the correlation coefficient between the time data and the APP download volume, and adjusting the parameters of the advertising delivery combination based on the correlation coefficient, specifically including:

[0030] Taking the holidays and hot events as independent variables and the APP download volume as dependent variables, a vector autoregression model is constructed. The relationship between holidays, hot events and APP download volume is calculated through the vector autoregression model, and the parameters of the advertising delivery combination during certain holidays and hot events are determined based on the relationship.

[0031] In the second aspect, the present application also proposes a promotion system based on APP reservation data, including:

[0032] The first processing module: obtains user behavior data in the promotion channel, builds a channel evaluation index system based on the user behavior data, establishes a channel effect evaluation model, obtains input-output data based on the channel effect evaluation model, and triggers budget adjustment when the input-output ratio is lower than a preset threshold;

[0033] The second processing module: extracts the operation characteristics of the reserved user based on the channel effect evaluation model, builds a user download willingness prediction model based on the operation characteristics, and calculates the download probability value of the reserved user according to the user download willingness prediction model;

[0034] The third processing module: analyzing the crowd attributes, creative themes and delivery time parameters in the competitive product promotion, building an advertising delivery combination based on the functions of the competitive product APP, and optimizing the parameters of the advertising delivery combination;

[0035] Fourth processing module: obtaining the user's evaluation text of the advertisement, and optimizing the creative parameters of the advertisement based on the evaluation text;

[0036] When optimizing the parameters of the advertising delivery combination, time data is introduced, a correlation coefficient between the time data and the APP download volume is calculated, and parameter information of the advertising delivery combination is adjusted based on the correlation coefficient.

[0037] In a third aspect, the present application further proposes a terminal device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements any of the methods described above when executing the computer program.

[0038] In a fourth aspect, the present application further proposes a computer-readable storage medium, wherein the computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the method as described in any one of the above items is implemented.

[0039] Compared with the prior art, the embodiments of the present invention have the following beneficial effects:

[0040] A comprehensive channel evaluation indicator system has been built, and an accurate channel effect evaluation model has been established. Once it falls below the preset threshold, the budget adjustment is triggered, which enables the refined management of promotion resources, effectively avoids resource waste, and improves the overall promotion efficiency.

[0041] Based on the operation characteristics of reservation users extracted by the channel effect evaluation model, a user download intention prediction model was constructed, which can predict the download probability value of each reservation user and implement personalized promotion strategies according to users with different willingness levels, thereby improving user conversion rate, enhancing user experience and improving user satisfaction.

[0042] By analyzing the promotion data of competing products and combining it with the functional characteristics of the APP, we have built a rich advertising creative library and designed an advertising combination optimization mechanism based on user feedback. At the same time, we introduced time data and used the vector autoregression model to analyze the relationship between holidays, hot events and APP downloads, dynamically adjusted advertising parameters, realized the intelligence and precision of advertising, and further improved the advertising effect. BRIEF DESCRIPTION OF THE DRAWINGS

[0043] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the drawings required for use in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0044] Figure 1 It is a flowchart of a promotion method based on APP reservation data provided by an embodiment of the present invention;

[0045] Figure 2 It is a structural diagram of a promotion system based on APP reservation data provided by an embodiment of the present invention;

[0046] Figure 3 It is a schematic diagram of the structure of the terminal device provided in an embodiment of the present application. DETAILED DESCRIPTION

[0047] In the following description, specific details such as specific system structures, technologies, etc. are provided for the purpose of illustration rather than limitation, so as to provide a thorough understanding of the embodiments of the present application. However, it should be clear to those skilled in the art that the present application may also be implemented in other embodiments without these specific details. In other cases, detailed descriptions of well-known systems, devices, circuits, and methods are omitted to prevent unnecessary details from obstructing the description of the present application.

[0048] It should be understood that when used in the present specification and the appended claims, the term "comprising" indicates the presence of described features, wholes, steps, operations, elements and / or components, but does not exclude the presence or addition of one or more other features, wholes, steps, operations, elements, components and / or combinations thereof.

[0049] It should also be understood that the term “and / or” used in the specification and appended claims refers to any and all possible combinations of one or more of the associated listed items, and includes these combinations.

[0050] As used in the specification and appended claims of this application, the term "if" can be interpreted as "when" or "uponce" or "in response to determining" or "in response to detecting", depending on the context. Similarly, the phrase "if it is determined" or "if [described condition or event] is detected" can be interpreted as meaning "uponce it is determined" or "in response to determining" or "uponce [described condition or event] is detected" or "in response to detecting [described condition or event]", depending on the context.

[0051] In addition, in the description of the present application specification and the appended claims, the terms "first", "second", "third", etc. are only used to distinguish the descriptions and cannot be understood as indicating or implying relative importance.

[0052] References to "one embodiment" or "some embodiments" etc. described in the specification of this application mean that one or more embodiments of the present application include specific features, structures or characteristics described in conjunction with the embodiment. Therefore, the statements "in one embodiment", "in some embodiments", "in some other embodiments", "in some other embodiments", etc. that appear in different places in this specification do not necessarily refer to the same embodiment, but mean "one or more but not all embodiments", unless otherwise specifically emphasized in other ways. The terms "including", "comprising", "having" and their variations all mean "including but not limited to", unless otherwise specifically emphasized in other ways.

[0053] See also Figure 1, the present application proposes a promotion method based on APP reservation data, including: S101 obtaining user behavior data in the promotion channel, building a channel evaluation index system according to the user behavior data, establishing a channel effect evaluation model, obtaining input-output data according to the channel effect evaluation model, and triggering budget adjustment when the input-output ratio is lower than a preset threshold, S102 extracting the operation characteristics of the reservation user based on the channel effect evaluation model, building a user download intention prediction model based on the operation characteristics, and calculating the download probability value of the reservation user according to the user download intention prediction model, S103 analyzing the crowd attributes, creative themes and delivery period parameters in the promotion of competing products, building an advertising delivery combination in combination with the functions of the competing APP, and optimizing the parameters of the advertising delivery combination, S104 introducing time data when optimizing the parameters of the advertising delivery combination, calculating the correlation coefficient between the time data and the APP download volume, and adjusting the parameter information of the advertising delivery combination based on the correlation coefficient, S105 obtaining the user's evaluation text of the advertisement, and optimizing the creative parameters of the advertisement based on the evaluation text.

[0054] Specifically, obtain user behavior data in the promotion channel, such as user click, browse, register, and make reservations in different channels (such as search engines, social media, app stores, etc.). These data can be obtained by integrating the SDK or API interface of each channel. In this way, users' behavior preferences in different channels can be understood, providing a data basis for subsequent channel evaluation and prediction of user download intention. For example, by analyzing the keywords that users search in search engines, users' needs and interests can be understood, thereby optimizing the keyword settings for advertising. Construct a channel evaluation index system based on user behavior data. For example, indicators such as click-through rate, conversion rate, average user stay time, and reservation rate can be counted for each channel. These indicators can reflect the quality and effect of the channel. For example, if a channel has a high click-through rate but a low conversion rate, it may mean that the channel has attracted a large number of non-target users. Establish a channel effect evaluation model, for example, machine learning algorithms such as logistic regression and decision trees can be used to predict the conversion effect of each channel based on the channel evaluation index system. Through the evaluation model, different channels can be quantitatively evaluated, and budget allocation can be performed based on the evaluation results. For example, if the model predicts that the conversion rate of channel A is higher than that of channel B, more budget can be allocated to channel A. According to the channel effect evaluation model, input-output data is obtained. For example, the customer acquisition cost and user lifetime value of each channel can be calculated. When the input-output ratio is lower than the preset threshold, for example, lower than 1:2, the budget adjustment is triggered, such as reducing the advertising budget for the channel or stopping the advertising on the channel. This can avoid wasting budget on inefficient channels and concentrate the budget on efficient channels. For example, if it is found that the customer acquisition cost of a certain channel is too high and the input-output ratio is lower than the preset threshold, the budget investment in the channel can be reduced. Based on the channel effect evaluation model, the operation characteristics of the appointment user are extracted. For example, the number of pages browsed by the user before the appointment, the dwell time, the number of clicks, etc. These characteristics can reflect the user's interest in the product and the willingness to download. For example, if the user browses multiple product pages and stays on each page for a long time, it means that the user is more interested in the product and has a higher willingness to download. Based on the operation characteristics, a user download willingness prediction model is constructed. For example, machine learning algorithms such as logistic regression and support vector machines can be used to predict the download probability values ​​of scheduled users. This model can help identify which scheduled users are more likely to download the APP, thereby conducting precise marketing promotions. For example, if the download probability value of a certain user is predicted to be 90%, coupons or other incentives can be pushed to the user to encourage him or her to download the APP.Analyze the crowd attributes, creative themes, and delivery time parameters in the promotion of competing products. For example, analyze the target user groups, advertising copy, image materials, delivery time periods, etc. of competing apps. This information can help understand the promotion strategies of competing products and learn from their successful experiences. For example, if it is found that competing apps mainly target young female users and that advertising at night is more effective, you can learn from these experiences and optimize your own advertising delivery strategy. Combine the functions of competing apps. For example, if the competing apps have social functions, you can design some advertising ideas that highlight the social functions. Construct advertising delivery combinations. For example, combine different crowd attributes, creative themes, and delivery time periods into different advertising delivery plans. Optimize the parameters of the advertising delivery combination. For example, adjust the advertising copy and image materials according to different crowd attributes, and adjust the advertising delivery time according to different time periods. Introduce time data. For example, count the number of APP downloads in different time periods. Calculate the correlation coefficient between time data and APP downloads. For example, if the calculation finds that the number of downloads on weekends is significantly higher than that on weekdays, the correlation coefficient is high. Adjust the parameters of the advertising delivery combination based on the correlation coefficient. For example, you can increase the advertising delivery budget on weekends and adjust the advertising delivery time on weekends. Get the user's evaluation text of the advertisement, for example, collect the user's comments on the advertisement on the app store, social media and other platforms. Based on the evaluation text, optimize the creative parameters of the advertisement. For example, if the user comments that the advertisement copy is not attractive enough, the advertisement copy can be modified. For example, if the user comments that the advertisement image is not clear enough, a higher quality image can be replaced. Through the above steps and methods, a complete promotion optimization closed loop can be formed to continuously improve the promotion effect and reduce the promotion cost.

[0055] In some embodiments, user behavior data includes clicks, registrations, downloads, and payments; a channel evaluation index system is constructed based on the user behavior data, a channel effect evaluation model is established, input-output data is obtained based on the channel effect evaluation model, and a budget adjustment is triggered when the input-output ratio is lower than a preset threshold, specifically including: cleaning the user behavior data and obtaining a basic data set; performing data preprocessing on the basic data set, including performing numerical normalization, performing unique hot encoding on categorical variables, and converting the categorical variables into numerical variables; based on the preprocessed data, a channel evaluation index system is constructed, and the channel evaluation index system is used as the input feature of the channel effect evaluation model, and the channel evaluation index system includes conversion rate, next-day retention rate, 7-day retention rate, and average order value; in the channel effect evaluation model, the channel input cost and evaluation index are used as independent variables, and the channel output income is used as the dependent variable, and the model parameters are trained; through the trained model, for each promotion channel, its input cost and various indicator values ​​are input, its output income is predicted, and the input-output ratio is calculated.

[0056] Specifically, user behavior data includes clicks, registrations, downloads, and payments. These data are the basis for evaluating channel effects. For example, an educational app called "Happy Learning" is being promoted. First, the user behavior data needs to be cleaned to obtain the basic data set. Cleaning data includes removing duplicate data, invalid data (for example, abnormal click behavior), etc. For example, some users may repeatedly click on advertisements due to network problems. These repeated clicks need to be removed to avoid exaggerating the channel effect. After cleaning, a basic data set containing user clicks, registrations, downloads, and payment behaviors is obtained. Next, the basic data set is preprocessed. Data preprocessing includes numerical normalization and unique hot encoding. Numerical normalization is to eliminate the difference in dimensions between different indicators. For example, the magnitude of the number of clicks and the amount of payment is different, and normalization is required to make them within the same range. Suppose a channel brings 1,000 clicks and another channel brings 10 payments. It makes no sense to compare these two numbers directly. After normalization, the effects of different channels can be compared more fairly. Unique hot encoding is to convert categorical variables into numerical variables. For example, the promotion channels may be "search engine", "social media", "app store", etc. These categorical variables need to be converted into numerical variables before they can be used for model training. Three numerical values ​​can be used to represent these three channels: search engine is represented by "100", social media is represented by "010", and app store is represented by "001". Based on the preprocessed data, a channel evaluation indicator system is constructed. The indicator system includes conversion rate, next-day retention rate, 7-day retention rate and customer unit price. Conversion rate refers to the proportion of users who click on the ad to download the APP, which reflects the attractiveness of the advertisement. Next-day retention rate refers to the proportion of users who still use the APP the next day after downloading the APP, which reflects the retention ability of the APP. The same is true for the 7-day retention rate. Customer unit price refers to the average payment amount of each paying user, which reflects the profitability of the APP. Assuming that the "search engine" channel brings 1,000 clicks, of which 100 people download the APP, then the conversion rate is 10%. Among these 100 downloading users, 50 people continue to use the APP the next day, so the next-day retention rate is 50%. After 7 days, there are still 20 people who continue to use it, so the 7-day retention rate is 20%; assuming that 5 of these 20 people pay, and the total payment amount is 100 yuan, then the average customer price is 20 yuan. The channel evaluation index system is used as the input feature of the channel effect evaluation model. In the model, the channel input cost and evaluation index are used as independent variables, and the channel output income is used as the dependent variable. The model is trained by the machine learning algorithm to obtain the model parameters. Assume that 1,000 yuan is invested in the "search engine" channel and the above indicator data is obtained. Through the trained model, the output income of the channel can be predicted, for example, 2,000 yuan. Through the trained model, for each promotion channel, its input cost and various indicator values ​​are input, its output income is predicted, and the input-output ratio is calculated.For example, the input-output ratio of the "search engine" channel is 2,000 yuan / 1,000 yuan = 2. If the preset threshold is 5, then the input-output ratio of the channel is higher than the threshold, indicating that the effect is good. If the input-output ratio of a channel is lower than the preset threshold, it triggers budget adjustments, such as reducing the budget of the channel or adjusting the delivery strategy. In this way, the budget can be allocated to channels with better effects, improving the overall promotion efficiency.

[0057] In some embodiments, based on the channel effect evaluation model, the operation characteristics of the reserved users are extracted, based on the operation characteristics, a user download willingness prediction model is constructed, and the download probability value of the reserved users is calculated according to the user download willingness prediction model, specifically including: according to the user behavior data, the device brand, operating system, network environment, geographical location, access frequency and stay time of the reserved users are obtained, non-numerical features are converted into numerical representations, and a user feature vector is constructed; the user download willingness prediction model is trained with the user feature vector as input and whether the user downloads the App as the target variable; the user download willingness prediction model is evaluated using a cross-validation method to predict the download probability value of each reserved user, and the users are divided into three levels: high willingness, medium willingness and low willingness; for high willingness users, the App installation package download link is pushed; for medium willingness users, App function introduction and preferential incentives are provided; for low willingness users, the push frequency is reduced.

[0058] Specifically, the channel effect evaluation model can not only evaluate the quality of the channel, but also be used to predict user behavior, such as predicting the download intention of the scheduled users. By analyzing the user's operation characteristics, a user download intention prediction model can be constructed to achieve precision marketing. First, it is necessary to extract the operation characteristics of the scheduled users. Information such as the user's device brand, operating system, network environment, geographic location, access frequency and length of stay can all be used as features. For example, users who use Apple phones may be more likely to download apps than users who use Android phones, because Apple phone users are generally considered to have a higher willingness to pay; for example, users in a 4G network environment are more likely to download apps than users in a 2G network environment because the download speed is faster and the user experience is better. The user's geographic location may also affect the willingness to download. For example, users in first-tier cities may have a higher acceptance of new apps. User access frequency and length of stay are also important features. Users who visit frequently and stay for a long time indicate that they have a higher interest in the app and are more likely to download. For non-numerical features, they need to be converted into numerical representations. For example, the device brand can be converted using a one-hot encoding method, and each brand such as Apple, Huawei, and Xiaomi corresponds to a separate dimension. The operating system can also be processed in a similar way. The network environment can be assigned a value, for example, 2G is assigned a value of 1, 3G is assigned a value of 2, 4G is assigned a value of 3, 5G is assigned a value of 4, and Wi-Fi is assigned a value of 5. The geographical location can be assigned a value according to the city level, for example, first-tier cities are assigned a value of 3, second-tier cities are assigned a value of 2, and third-tier cities are assigned a value of 1. By combining these features together, a user feature vector can be constructed. For example, a user's feature vector can be expressed as: [Apple, iOS, 4G, Beijing, 5 times, 30 minutes], and after numerical processing, it can be expressed as: [1,0,0,0,3,5,3,5,30], where the first four digits represent the device brand (assuming that there are only Apple, Huawei, Xiaomi, and other four brands), the next digit represents the operating system, the next digit represents the network environment, and then a digit represents the geographical location, and the last two digits represent the frequency of visits and the length of stay, respectively. Next, with the user feature vector as input and whether the user downloads the App as the target variable, a user download intention prediction model is trained. Machine learning algorithms such as logistic regression and support vector machines can be used for training. For example, you can collect feature vectors of 1,000 reservation users and data on whether they have downloaded the App, of which 800 data are used to train the model and 200 data are used to test the performance of the model. After the model training is completed, the model can be evaluated using the cross-validation method. For example, you can use a five-fold cross-validation method to divide the training data into five parts, use four of them to train the model each time, and use the remaining one for validation. The final evaluation result is the average of the five validation results.Finally, the trained model can be used to predict the download probability value of each pre-booked user, and the users can be divided into three levels: high willingness, medium willingness, and low willingness. For example, users with a download probability greater than 8 are classified as high willingness users, users between 5 and 8 are classified as medium willingness users, and users less than 5 are classified as low willingness users. Different marketing strategies are adopted for users of different levels. For example, for high willingness users, the App installation package download link can be pushed directly. For medium willingness users, App function introductions and preferential incentives can be provided, such as coupons, membership trials, etc., to encourage them to download. For low willingness users, the push frequency can be reduced to avoid excessive interruptions, or other promotion methods can be tried. This can improve marketing efficiency and reduce promotion costs.

[0059] In some embodiments, the crowd attributes, creative themes, and delivery period parameters in the promotion of competing products are analyzed, an advertising delivery combination is constructed in combination with the functions of the APP, and the parameters of the advertising delivery combination are optimized, specifically including: collecting competing product advertising data, the competing product advertising data includes the crowd attributes, creative themes, and delivery period parameters in the promotion; according to the functions of the APP, combined with the characteristics of different crowd attributes, multiple groups of advertising creative themes and copywriting are designed, an advertising creative library is constructed, and for each crowd attribute, materials are selected from the advertising creative library to generate an advertising delivery combination.

[0060] Specifically, analyze the crowd attributes, creative themes and delivery time parameters in the promotion of competing products, and build an advertising delivery combination in combination with the functions of your own APP. The ultimate goal is to optimize the advertising delivery effect and improve the conversion rate. This requires in-depth analysis of competing products and differentiated positioning based on the characteristics of your own APP. First, collect competitive product advertising data. It is necessary to understand who the target user group of the competing products is, what advertising creative themes they use, and in what time period they are delivered. For example, assuming that a fitness APP is developed, you can analyze the advertising delivery strategies of competing APPs such as Keep and Mint Health. You can understand their crowd attributes (such as age, gender, hobbies, etc.), creative themes (such as weight loss, muscle gain, body shaping, etc.) and delivery time periods (such as weekday evenings, weekends, etc.) by viewing their ads. You can also use some third-party tools to monitor competitive product advertising data, such as AppGrowing, SensorTower, etc. Secondly, according to the functions of the APP and the characteristics of different crowd attributes, design multiple groups of advertising creative themes and copywriting, and build an advertising creative library. For example, the main function of the fitness APP is "personalized fitness plan". Different advertising creative themes can be designed for different crowd attributes. For example, for young female users, you can emphasize "slimming and shaping"; for middle-aged male users, you can emphasize "strengthening physical fitness"; for elderly users, you can emphasize "healthy life". Different advertising copy can be designed for each group of users. For example, for young female users, you can use some lively and fashionable copy, such as "Summer is coming, show off your waistline!"; for middle-aged male users, you can use some stable and professional copy, such as "scientific fitness, achieve a better self!"; for elderly users, you can use some warm and caring copy, such as "exercise makes life healthier!". Organize these creative themes and copy into an advertising creative library for subsequent use. Finally, for each group of users, select appropriate materials from the advertising creative library to generate an advertising delivery combination. For example, for young female users, you can choose the creative theme of "slimming and shaping", and match it with lively and fashionable copy and pictures, and deliver it in leisure time such as evenings or weekends. For middle-aged male users, you can choose the creative theme of "strengthening physical fitness", and match it with stable and professional copy and pictures, and deliver it on weekday evenings. For elderly users, you can choose the creative theme of "healthy life" and add warm and caring copy and pictures to run ads during the day. The advantage of doing this is that you can accurately run ads based on the preferences and habits of different groups of people, thereby increasing the click-through rate and conversion rate of ads. For example, young female users are more interested in "slimming and shaping", so it is easier to attract their attention by running such ads. Middle-aged male users are more concerned about "strengthening their physical fitness", so it is easier to resonate with them by running such ads. Elderly users pay more attention to "healthy life", so it is easier to gain their trust by running such ads.In this way, advertisements can be accurately delivered to target user groups, avoiding wasting advertising budgets and ultimately improving advertising effectiveness.

[0061] In some embodiments, obtaining a user's evaluation text of an advertisement, and optimizing the creative parameters of the advertisement based on the evaluation text, specifically includes: obtaining a user's evaluation text of the advertisement, and preprocessing the evaluation text using natural language processing technology; dividing the preprocessed text data into a training set and a test set, inputting the training set and the test set into a convolutional neural network model for fine-tuning training, and evaluating the performance of the convolutional neural network model on the test set, obtaining the emotional tendency of the evaluation text through the convolutional neural network model, and judging whether the user's evaluation of the creative of the advertisement is positive, neutral or negative according to the emotional tendency; constructing a vectorized representation reflecting user feedback opinions based on the evaluation texts with different emotional tendencies, taking the vector of the user feedback opinions as an optimization target, designing a fitness function of a genetic algorithm so that the fitness function is positively correlated with user satisfaction, inputting the fitness function into the genetic algorithm, and optimizing the creative parameters of the advertisement, wherein the creative parameters of the advertisement include copywriting, pictures and layout.

[0062] Specifically, user evaluations are crucial for ad optimization. By analyzing users' evaluations of ad creativity (copywriting, images, layout), ad strategies can be adjusted to enhance ad effectiveness. Here, natural language processing techniques and genetic algorithms are adopted to achieve the optimization of ad creativity parameters based on user evaluations. First, it is necessary to obtain the evaluation texts of users on ads. These texts can come from app store reviews, social media messages, feedback on ad platforms, etc. For example, for an ad of a fitness APP, the following user evaluations are collected: "The ad image is very attractive, and the copywriting is also very powerful.", "The copywriting is too long and I can't read it.", "The model in the picture has such a great figure that it makes me feel a lot of pressure." Next, preprocess the collected evaluation texts. This includes removing special characters, unifying case, removing stop words (such as "de", "shi", "le", etc.) and performing word segmentation, etc. For example, after preprocessing the evaluation "The copywriting is too long and I can't read it.", we get "copywriting, long, can't read it". The preprocessed text data is more standardized and convenient for subsequent analysis. The preprocessed text data needs to be divided into a training set and a test set. For example, 80% of the data is used as the training set to train the convolutional neural network model, and the remaining 20% of the data is used as the test set to evaluate the performance of the model. The convolutional neural network model is a deep learning model that performs well in the field of natural language processing. It can learn the features in the text and use them for sentiment classification. Input the training set and the test set into the convolutional neural network model for fine-tuning training. Fine-tuning training means further training using specific task data based on an existing pre-trained model, so as to improve the performance of the model on this task. Evaluate the performance of the model on the test set, for example, evaluate metrics such as the accuracy, precision, and recall of the model. Through the trained convolutional neural network model, the sentiment tendency of the evaluation text can be obtained. The model will classify each evaluation text as positive, neutral, or negative. For example, "The ad image is very attractive, and the copywriting is also very powerful." will be judged as a positive evaluation; "The copywriting is too long and I can't read it." will be judged as a negative evaluation; "It feels okay" will be judged as a neutral evaluation. Then, it is necessary to construct a vectorized representation of the evaluation texts with different sentiment tendencies. For example, using word vector technology, each word is converted into a vector, and then all word vectors are combined into a sentence vector to represent the user's feedback. Positive evaluations, neutral evaluations, and negative evaluations will be converted into different vectors, reflecting different feedback directions of users. To optimize the ad creativity parameters, a genetic algorithm is adopted. The genetic algorithm is an optimization algorithm that simulates the process of natural selection. It continuously iterates by simulating operations such as gene crossover and mutation, and finally finds the optimal solution. First, it is necessary to design the fitness function of the genetic algorithm. The fitness function is used to evaluate the quality of each individual (i.e., a set of ad creativity parameters). Here, make the fitness function positively correlated with user satisfaction.For example, the number of positive reviews can be used as part of the fitness function, and the number of negative reviews can be taken as a negative value as the other part of the fitness function. By inputting the fitness function into the genetic algorithm, you can start optimizing the creative parameters of the advertisement. The genetic algorithm will continuously iterate and adjust parameters such as copy, pictures, and layout to maximize the value of the fitness function, thereby improving user satisfaction. For example, if many users say that the copy is too long, the genetic algorithm may try to shorten the copy; if users say that the picture is not attractive enough, the genetic algorithm may try to replace the picture or adjust the layout of the picture. Through continuous optimization, a set of advertising creative parameters with the best user evaluation is finally obtained.

[0063] In some embodiments, the time data includes holidays, hot events and APP downloads; the time data is introduced, the correlation coefficient between the time data and the APP downloads is calculated, and the parameters of the advertising delivery combination are adjusted based on the correlation coefficient, specifically including: taking the holidays and hot events as independent variables, the APP downloads as dependent variables, constructing a vector autoregression model, calculating the relationship between holidays, hot events and APP downloads through the vector autoregression model, and determining the parameters of the advertising delivery combination during certain holidays and hot events according to the relationship.

[0064] Specifically, in advertising strategies, it is crucial to consider the relationship between holidays, hot events, and app downloads. For example, if a game app wants to increase downloads during the National Day, the advertising strategy team can use the vector autoregression model to analyze data from past National Days. First, the team will collect app download data during the National Day and advertising data during the same period, including the types, frequency, and delivery platforms of ads. Through the vector autoregression model, the team can analyze the impact of specific dates of holidays and hot events on app downloads. The model may show that around the National Day, the download volume of game apps increased significantly due to the increase in users' leisure time. In addition, if there are related hot events during this period, such as a large-scale e-sports competition, it may further promote the increase in downloads. Based on these analysis results, the advertising team can decide to increase the amount of advertising around the National Day and choose to place competition-related ads on relevant channels and social media during the e-sports competition. Such strategy adjustments are not only based on historical data analysis, but also enable precision marketing for specific holidays and hot events, thereby maximizing advertising effects and app downloads. In addition, by continuously monitoring the impact of holidays and hot events, the advertising team can continuously adjust and optimize advertising strategies. For example, if it is found that certain advertising has not increased downloads as expected, the team can quickly adjust the advertising content or delivery channels to seek better market response. In this way, advertising is not only more targeted, but also more flexible to respond to market changes, ensuring that advertising resources are used most effectively and ultimately maximizing APP downloads.

[0065] See also Figure 2 , the application also proposes a promotion system based on APP reservation data, including:

[0066] The first processing module 201: obtains user behavior data in the promotion channel, constructs a channel evaluation index system based on the user behavior data, establishes a channel effect evaluation model, obtains input-output data based on the channel effect evaluation model, and triggers budget adjustment when the input-output ratio is lower than a preset threshold;

[0067] The second processing module 202: extracting the operation characteristics of the reserved user based on the channel effect evaluation model, constructing a user download intention prediction model based on the operation characteristics, and calculating the download probability value of the reserved user according to the user download intention prediction model;

[0068] The third processing module 203: analyzing the crowd attributes, creative themes and delivery time parameters in the competitive product promotion, building an advertising delivery combination based on the functions of the competitive product APP, and optimizing the parameters of the advertising delivery combination;

[0069] Fourth processing module 204: obtaining the user's evaluation text of the advertisement, and optimizing the creative parameters of the advertisement based on the evaluation text;

[0070] When optimizing the parameters of the advertising delivery combination, time data is introduced, a correlation coefficient between the time data and the APP download volume is calculated, and parameter information of the advertising delivery combination is adjusted based on the correlation coefficient.

[0071] It is understandable that if Figure 1 The contents of the promotion method embodiment based on APP reservation data shown in the figure are applicable to the promotion system embodiment based on APP reservation data. The functions specifically implemented by the promotion system embodiment based on APP reservation data are similar to those in the above-mentioned Figure 1 The promotion method embodiment based on APP reservation data shown is the same as that shown in the embodiment, and the beneficial effects achieved are the same as those shown in the embodiment. Figure 1 The beneficial effects achieved by the promotion method embodiment based on APP reservation data shown are also the same.

[0072] It should be noted that the information interaction, execution process and other contents between the above-mentioned systems are based on the same concept as the embodiment of the method of the present invention. Their specific functions and technical effects can be found in the method embodiment part and will not be repeated here.

[0073] Those skilled in the art can clearly understand that, for the convenience and simplicity of description, only the division of the above-mentioned functional units and modules is used as an example for illustration. In practical applications, the above-mentioned function allocation can be completed by different functional units and modules as needed, that is, the internal structure of the system can be divided into different functional units or modules to complete all or part of the functions described above. The functional units and modules in the embodiment can be integrated in a processing unit, or each unit can exist physically separately, or two or more units can be integrated in one unit. The above-mentioned integrated unit can be implemented in the form of hardware or in the form of software functional units. In addition, the specific names of the functional units and modules are only for the convenience of distinguishing each other, and are not used to limit the scope of protection of this application. The specific working process of the units and modules in the above-mentioned system can refer to the corresponding process in the aforementioned method embodiment, which will not be repeated here.

[0074] See also Figure 3 The embodiment of the present invention further provides a computer device 3, comprising: a memory 302 and a processor 301 and a computer program 303 stored in the memory 302. When the computer program 303 is executed on the processor 301, a promotion method based on APP reservation data as described in any one of the above methods is implemented.

[0075] The computer device 3 may be a computing device such as a desktop computer, a notebook, a PDA, or a cloud server. The computer device 3 may include, but is not limited to, a processor 301 and a memory 302. Those skilled in the art will appreciate that Figure 3 It is only an example of computer device 3 and does not constitute a limitation on computer device 3. It may include more or fewer components than shown in the figure, or a combination of certain components, or different components, for example, it may also include input and output devices, network access devices, etc.

[0076] The processor 301 may be a central processing unit (CPU), or other general-purpose processors, digital signal processors (DSP), application-specific integrated circuits (ASIC), field-programmable gate arrays (FPGA), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. A general-purpose processor may be a microprocessor or any conventional processor, etc.

[0077] In some embodiments, the memory 302 may be an internal storage unit of the computer device 3, such as a hard disk or memory of the computer device 3. In other embodiments, the memory 302 may also be an external storage device of the computer device 3, such as a plug-in hard disk, a smart memory card (Smart Med iaCard, SMC), a secure digital (Secure Digital, SD) card, a flash card (Flash Card), etc. equipped on the computer device 3. Further, the memory 302 may also include both an internal storage unit of the computer device 3 and an external storage device. The memory 302 is used to store an operating system, an application program, a boot loader (BootLoader), data, and other programs, such as the program code of the computer program, etc. The memory 302 may also be used to temporarily store data that has been output or is to be output.

[0078] An embodiment of the present invention further provides a computer-readable storage medium having a computer program stored thereon. When the computer program is executed by a processor, the promotion method based on APP reservation data as described in any one of the above methods is implemented.

[0079] In this embodiment, if the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the present application implements all or part of the processes in the above-mentioned embodiment method, which can be completed by instructing the relevant hardware through a computer program, and the computer program can be stored in a computer-readable storage medium. When the computer program is executed by the processor, the steps of the above-mentioned various method embodiments can be implemented. Among them, the computer program includes computer program code, and the computer program code can be in source code form, object code form, executable file or some intermediate form. The computer-readable medium can at least include: any entity or device that can carry the computer program code to the camera / terminal device, recording medium, computer memory, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), electric carrier signal, telecommunication signal and software distribution medium. For example, a USB flash drive, a mobile hard disk, a magnetic disk or an optical disk. In some jurisdictions, according to legislation and patent practice, computer-readable media cannot be electric carrier signals and telecommunication signals.

[0080] The embodiments described above are only used to illustrate the technical solutions of the present application, rather than to limit them. Although the present application has been described in detail with reference to the aforementioned embodiments, a person skilled in the art should understand that the technical solutions described in the aforementioned embodiments may still be modified, or some of the technical features may be replaced by equivalents. Such modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present application, and should all be included in the protection scope of the present application.

Claims

1. A promotion method based on APP reservation data, characterized in that: include: Obtain user behavior data in the promotion channel, build a channel evaluation index system based on the user behavior data, establish a channel effect evaluation model, obtain input-output data based on the channel effect evaluation model, and trigger budget adjustment when the input-output ratio is lower than a preset threshold; Based on the channel effect evaluation model, extract the operation characteristics of the reserved user, build a user download willingness prediction model based on the operation characteristics, and calculate the download probability value of the reserved user according to the user download willingness prediction model; Analyze the crowd attributes, creative themes and delivery time parameters in the competitive product promotion, build an advertising delivery combination based on the functions of the competitive product APP, and optimize the parameters of the advertising delivery combination; Obtaining the user's evaluation text of the advertisement, and optimizing the creative parameters of the advertisement based on the evaluation text; When optimizing the parameters of the advertising delivery combination, time data is introduced, a correlation coefficient between the time data and the APP download volume is calculated, and the parameters of the advertising delivery combination are adjusted based on the correlation coefficient.

2. The promotion method based on APP reservation data according to claim 1 is characterized in that: User behavior data includes clicks, registrations, downloads and payments; a channel evaluation index system is constructed based on the user behavior data, a channel effect evaluation model is established, input-output data is obtained based on the channel effect evaluation model, and budget adjustment is triggered when the input-output ratio is lower than a preset threshold, specifically including: Cleaning the user behavior data to obtain a basic data set; Performing data preprocessing on the basic data set, including performing numerical normalization, performing one-hot encoding on categorical variables, and converting the categorical variables into numerical variables; Based on the preprocessed data, a channel evaluation index system is constructed, and the channel evaluation index system is used as an input feature of the channel effect evaluation model, wherein the channel evaluation index system includes conversion rate, next-day retention rate, 7-day retention rate, and average order value; In the channel effect evaluation model, the channel input cost and evaluation index are used as independent variables, the channel output income is used as the dependent variable, and the model parameters are obtained by training; Through the trained model, for each promotion channel, input its input cost and various indicator values, predict its output revenue, and calculate the input-output ratio.

3. The promotion method based on APP reservation data according to claim 1 is characterized in that: Based on the channel effect evaluation model, extract the operation characteristics of the reserved user, build a user download willingness prediction model based on the operation characteristics, and calculate the download probability value of the reserved user according to the user download willingness prediction model, specifically including: Based on user behavior data, obtain the device brand, operating system, network environment, geographic location, visit frequency and stay time of the reservation user, convert non-numeric features into numerical representation, and construct a user feature vector; Taking the user feature vector as input and whether the user downloads the App as the target variable, a user download intention prediction model is trained; The user download willingness prediction model is evaluated using a cross-validation method to predict the download probability value of each reserved user, and the users are divided into three levels: high willingness, medium willingness and low willingness; Push the App installation package download link to high-willing users; For medium-willing users, provide App function introductions and preferential incentives; for low-willing users, reduce the push frequency.

4. The promotion method based on APP reservation data according to claim 1 is characterized in that: Analyze the crowd attributes, creative themes and delivery time parameters in the competitive product promotion, build an advertising delivery combination based on the functions of the APP, and optimize the parameters of the advertising delivery combination, including: Collect competitive product advertising data, including the attributes of the audience in the promotion, creative themes, and delivery time parameters; According to the functions of the APP and the characteristics of different groups of people, multiple groups of advertising creative themes and copywriting are designed, and an advertising creative library is constructed. For each group of people, materials are selected from the advertising creative library to generate an advertising delivery combination.

5. The promotion method based on APP reservation data according to claim 1 is characterized in that: Obtaining the user's evaluation text of the advertisement, and optimizing the creative parameters of the advertisement based on the evaluation text, specifically including: Obtaining the user's evaluation text of the advertisement, and preprocessing the evaluation text using natural language processing technology; Dividing the preprocessed text data into a training set and a test set, inputting the training set and the test set into a convolutional neural network model for fine-tuning training, and evaluating the performance of the convolutional neural network model on the test set, obtaining the sentiment tendency of the evaluation text through the convolutional neural network model, and judging whether the user's evaluation of the advertising creativity is positive, neutral or negative according to the sentiment tendency; According to the evaluation texts with different emotional tendencies, a vectorized representation reflecting user feedback is constructed, the vector of the user feedback is used as an optimization target, a fitness function of a genetic algorithm is designed so that the fitness function is positively correlated with user satisfaction, the fitness function is input into the genetic algorithm, and the creative parameters of the advertisement are optimized, the creative parameters of the advertisement including copy, pictures and layout.

6. The promotion method based on APP reservation data according to claim 1 is characterized in that: The time data includes holidays, hot events and APP downloads; Introducing time data, calculating the correlation coefficient between the time data and the APP download volume, and adjusting the parameters of the advertising delivery combination based on the correlation coefficient, specifically including: Taking the holidays and hot events as independent variables and the APP download volume as dependent variables, a vector autoregression model is constructed. The relationship between holidays, hot events and APP download volume is calculated through the vector autoregression model, and the parameters of the advertising delivery combination during certain holidays and hot events are determined based on the relationship.

7. A promotion system based on APP reservation data, characterized in that: include: The first processing module: obtains user behavior data in the promotion channel, builds a channel evaluation index system based on the user behavior data, establishes a channel effect evaluation model, obtains input-output data based on the channel effect evaluation model, and triggers budget adjustment when the input-output ratio is lower than a preset threshold; The second processing module: extracts the operation characteristics of the reserved user based on the channel effect evaluation model, builds a user download willingness prediction model based on the operation characteristics, and calculates the download probability value of the reserved user according to the user download willingness prediction model; The third processing module: analyzing the crowd attributes, creative themes and delivery time parameters in the competitive product promotion, building an advertising delivery combination based on the functions of the competitive product APP, and optimizing the parameters of the advertising delivery combination; Fourth processing module: obtaining the user's evaluation text of the advertisement, and optimizing the creative parameters of the advertisement based on the evaluation text; When optimizing the parameters of the advertising delivery combination, time data is introduced, a correlation coefficient between the time data and the APP download volume is calculated, and parameter information of the advertising delivery combination is adjusted based on the correlation coefficient.

8. A terminal device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that: When the processor executes the computer program, the method according to any one of claims 1 to 7 is implemented.

9. A computer-readable storage medium storing a computer program, characterized in that: When the computer program is executed by a processor, the method according to any one of claims 1 to 7 is implemented.

Images