An account screening system to improve advertising ROI

Optimizing advertising resource allocation through account data monitoring, user behavior analysis and prediction modules, the problem that advertising strategies in the existing technology cannot adapt to market changes is solved, and efficient allocation of advertising resources and maximum advertising effect is achieved.

CN120198180BActive Publication Date: 2025-08-08BEIJING ZHUANZHUAN SPIRIT TECH CO LTD
View PDF 2 Cites 0 Cited by

Patent Information

Application Number
CN202510254122.1
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2025-03-05
Publication Date
2025-08-08
Estimated Expiration
2045-03-05

AI Technical Summary

Technical Problem

The existing technology lacks real-time analysis and instant response capabilities for user interaction behavior, resulting in advertising strategies that cannot adapt to the immediate changes in the market, the allocation of advertising resources cannot maximize efficiency, and the advertising content cannot accurately match the current needs of users, affecting the advertising return on investment.

Method used

Interactive data is collected through the account data monitoring module, user behavior analysis module analyzes interaction stability and user preferences, marketing direction adjustment module identifies active periods and delivery frameworks, account effect prediction module predicts future interaction trends, high ROI account screening module filters potential accounts, and comprehensively utilizes multi-dimensional interactive data to optimize advertising resource allocation.

Benefits of technology

It has achieved accurate adjustments to the advertising strategy, improved the effective reach rate of advertisements, reduced invalid exposure, enhanced the interactive effect of advertisements, maximized the advertising delivery effect, and reduced advertising costs.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN120198180B_ABST
    Figure CN120198180B_ABST
Patent Text Reader

Abstract

The present invention relates to the field of account screening technology, and specifically to an account screening system for improving the ROI of advertising. The system includes an account data monitoring module, a user behavior analysis module, a marketing direction adjustment module, an account effect prediction module, and a high ROI account screening module. The present invention analyzes multi-dimensional interactive data including click volume, viewing time, and number of comments, and compares the data with the time, frequency, and location of content release, so that advertising strategies can be accurately adjusted to match current market dynamics and user behavior, thereby optimizing the allocation of advertising resources, reducing the exposure of invalid advertisements, and increasing the effective reach of advertisements. By forward-lookingly analyzing future interaction trends and conversion rates and dividing user preferences in detail, advertising content can be more accurately targeted to the target audience, so as to promote high ROI potential accounts, reduce advertising costs, enhance advertising interaction effects, and maximize the advertising delivery effect under the account.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention relates to the technical field of account screening, and in particular to an account screening system for improving ROI of advertising. Background Art

[0002] Account screening technology mainly involves analyzing and selecting users or advertising accounts on digital advertising platforms to improve the effectiveness and efficiency of advertising campaigns. This technology uses data mining and machine learning methods to identify advertising accounts that are most likely to generate the highest return on investment (ROI). By analyzing historical data, user behavior and market trends, it optimizes the allocation of advertising resources and reduces the exposure of invalid advertisements, thereby improving the overall performance and return rate of advertising campaigns.

[0003] Among them, the account screening system that improves the ROI of advertising refers to improving the return on advertising investment by screening and optimizing advertising accounts. The system uses algorithms to analyze various data points (such as click-through rate, conversion rate, interaction rate, etc.) to determine which accounts are most likely to have a positive response to the advertising budget. Its purpose is to help advertisers and marketers target target audiences more accurately, reduce wasted advertising spending, and improve the overall return on advertising through more efficient account management.

[0004] Existing technologies lack the ability to analyze and respond to user interactive behaviors in real time, resulting in advertising strategies that cannot adapt to immediate market changes, reducing the timeliness and accuracy of advertising strategies. The lack of in-depth analysis of different content types and user interactive behaviors makes it impossible to maximize the efficiency of advertising resource allocation, and advertising content cannot accurately match users' current needs, resulting in a waste of resources. In addition, existing technologies have limited capabilities in predicting user behavior and optimizing advertising timing, and fail to fully utilize available data for effective predictions, resulting in inaccurate matching of advertising timing and audience, further affecting the return on investment of advertising. Summary of the Invention

[0005] The purpose of the present invention is to solve the shortcomings of the prior art and to propose an account screening system for improving the ROI of advertising.

[0006] To achieve the above objectives, the present invention adopts the following technical solution: an account screening system for improving advertising ROI, the system comprising:

[0007] The account data monitoring module collects interaction data from influencer accounts, organizes it by time, compares content interaction ratios within the same time period, and filters out accounts with abnormal fluctuations in interaction rates to obtain interaction data stability indicators.

[0008] The user behavior analysis module screens accounts with key interaction stability based on the interaction data stability index, calculates the average user stay time and the proportion of secondary interactions, analyzes the content attractiveness of different accounts, and obtains the user preference trend value;

[0009] The marketing direction adjustment module calculates the average content interaction rate of different time periods based on the user preference trend value, identifies the user active period, selects the key secondary interaction rate time period, calculates the advertising exposure and interaction ratio, and obtains the marketing delivery time frame;

[0010] The account effect prediction module selects influencer accounts with both high exposure and high interaction volume based on the marketing delivery time frame, calculates the fluctuation range of growth rate over the past thirty days, and predicts the interaction trend over the next seven days to obtain the future interaction estimation index;

[0011] The high ROI account screening module screens accounts with predicted interaction values higher than the current average based on the future interaction prediction index, compares the predicted interaction values with the future interaction prediction values, calculates the expected conversion growth rate, and obtains a list of preferred ROI accounts.

[0012] The present invention has improvements in that the interaction data stability index includes volatility, stability rating, and abnormal account information; the user preference trend value specifically includes short video preference, live broadcast preference, and picture and text preference; the marketing delivery time frame specifically includes optimization period, interaction peak, and advertising exposure rate; the future interaction prediction index includes interaction growth prediction information, content activity expectation, and user engagement prediction results; the preferred ROI account list includes high-efficiency accounts, potential accounts, and priority sorting results.

[0013] The present invention is improved in that the account data monitoring module includes:

[0014] The interactive data collection submodule collects interactive data from influencer accounts, including clicks, viewing time, number of comments, number of shares, and like ratio, and organizes them according to the time dimension to create a time-series interactive data set;

[0015] The interaction data analysis submodule calls the time series interaction data set and compares the content interaction ratios in the same time period based on the content release time, display frequency, and display position, using the formula:

[0016]

[0017] Calculate interaction rate I r , get the interactive analysis results of each account, where c i represents the number of comments on the i-th content, l i represents the number of likes for the i-th content, v i represents the viewing time of the i-th content, s iIndicates the number of times the i-th content is shared, n o is the amount of content;

[0018] The anomaly detection submodule analyzes the volatility of the interaction rate based on the interaction analysis results of each account, filters out accounts with abnormal fluctuations in interaction rate, and obtains an interaction data stability index.

[0019] The present invention is improved in that the user behavior analysis module includes:

[0020] The interaction stability screening submodule calculates the interaction stability value of each account based on the interaction data stability index, screens key accounts that meet the stability requirements, and obtains a screened stable account list;

[0021] The user stay time calculation submodule calculates the average stay time of each account based on the filtered stable account list using the formula:

[0022]

[0023] Get the calculation result of the stay time, where T avg represents the average length of stay, D i represents the user stay time of the i-th account, W i represents the interaction stability value of the i-th account, and N represents the number of stable accounts after screening;

[0024] The content attractiveness analysis submodule analyzes the proportion of secondary interactions of short videos, live broadcasts, and text and pictures based on the calculation results of the dwell time, calculates the content attractiveness value of each account type, and determines the degree of attractiveness of different account contents to users to obtain the user preference trend value.

[0025] The present invention is improved in that the marketing direction adjustment module includes:

[0026] The expert account screening submodule calculates the ratio of the total interaction amount to the display frequency of each account based on the user preference trend value, screens the expert account with the best display frequency, and obtains the best expert account list;

[0027] The active time period identification submodule collects user interactions in different time periods based on the list of the best influencer accounts, calculates the average content interaction rate in multiple time periods, selects the peak interaction time periods, and counts the key secondary interaction rate time periods to obtain user active time period distribution data;

[0028] The advertisement delivery timing calculation submodule collects advertisement delivery exposures in different time periods based on the user active time period distribution data, using the formula:

[0029]

[0030] Calculate the advertising exposure and interaction ratio R for each period ad , get the marketing delivery time frame, where ES i Represents the advertising exposure in the i-th time period, IS i represents the total amount of interaction in the i-th time period, HS i represents the peak interaction value in the i-th time period, SU i represents the seasonal adjustment coefficient, N S Represents the number of time periods.

[0031] The present invention is improved in that the account effect prediction module includes:

[0032] The data screening submodule screens influencer accounts that meet the key requirements of both exposure and interaction based on the marketing delivery time frame, compares the account interaction and exposure data, and calculates the exposure-to-interaction ratio of each account. Accounts with ratios within a threshold range are retained to obtain the screened key influencer accounts.

[0033] The growth rate calculation submodule obtains the exposure and interaction data of the key influencer accounts after screening in the past 30 days, analyzes the daily interaction growth rate, and uses the formula:

[0034]

[0035] Calculate the growth rate change of each account VX and smooth it to get the smoothed growth rate, where GX i represents the growth rate of interaction volume on day i, GX i-1 represents the growth rate of interaction on day i-1, n X Represents the calculation period, which is 30 days;

[0036] The interaction trend prediction submodule predicts the interaction volume trend in the next seven days based on the smoothed growth rate to obtain a future interaction estimation index.

[0037] The present invention is improved in that the high ROI account screening module includes:

[0038] The interaction volume prediction submodule obtains the historical interaction data of the target account based on the future interaction prediction index, calculates the future interaction volume prediction value of each account, selects accounts with interaction volume prediction values higher than the average, and marks them as potential high-conversion accounts;

[0039] The conversion rate comparison submodule retrieves historical advertising data based on the potential high-conversion accounts, extracts advertising volume, ad click volume, and ad conversion volume, calculates the historical advertising conversion rate of each account, and obtains historical conversion rate data;

[0040] The ROI account screening submodule is based on the historical conversion rate data: using the formula:

[0041]

[0042] Calculate the expected conversion growth rate enhancement value GU and obtain the preferred ROI account list, where P f represents the predicted value of future interaction volume, P h Represents the historical advertising conversion rate, T i Represents the historical conversion rate of multiple accounts, T m Represents the historical average conversion rate of all accounts, n A Represents the total number of accounts.

[0043] Compared with the prior art, the advantages and positive effects of the present invention are:

[0044] In the present invention, by analyzing multi-dimensional interactive data including click volume, viewing time and number of comments, and comparing the data with the time, frequency and location of content release, the advertising strategy can be accurately adjusted to match the current market dynamics and user behavior, thereby optimizing the allocation of advertising resources, reducing the exposure of invalid advertisements, and improving the effective reach of advertisements. By forward-lookingly analyzing future interaction trends and conversion rates, user preferences are divided in detail, so that advertising content can be more accurately targeted to the target audience, so as to promote high ROI potential accounts, reduce advertising costs, enhance advertising interaction effects, and maximize the advertising delivery effect under the account. BRIEF DESCRIPTION OF THE DRAWINGS

[0045] Figure 1 is a system flow chart of the present invention;

[0046] Figure 2 This is a flow chart of the account data monitoring module in the present invention;

[0047] Figure 3 This is a flow chart of the user behavior analysis module in the present invention;

[0048] Figure 4 This is a flow chart of the marketing direction adjustment module in the present invention;

[0049] Figure 5 This is a flow chart of the account effect prediction module in the present invention;

[0050] Figure 6 This is a flow chart of the high ROI account screening module in the present invention. DETAILED DESCRIPTION

[0051] In order to make the purpose, technical solutions and advantages of the present invention more clearly understood, the present invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention.

[0052] In the description of the present invention, it should be understood that the terms "length", "width", "up", "down", "front", "back", "left", "right", "vertical", "horizontal", "top", "bottom", "inside", "outside", etc., indicating directions or positional relationships, are based on the directions or positional relationships shown in the accompanying drawings and are only for the convenience of describing the present invention and simplifying the description. They do not indicate or imply that the devices or elements referred to must have a specific direction, be constructed and operate in a specific direction, and therefore should not be understood as limiting the present invention. In addition, in the description of the present invention, the meaning of "plurality" is two or more, unless otherwise clearly and specifically defined.

[0053] Example

[0054] See also Figure 1 The present invention provides a technical solution: an account screening system for improving advertising ROI, comprising:

[0055] The account data monitoring module collects interactive data from influencer accounts, including clicks, viewing time, number of comments, number of shares, and like ratio, and organizes it by time. It then compares the content interaction ratios within the same time period based on content release time, display frequency, and display location. It then filters out accounts with abnormally fluctuating interaction rates to obtain an interaction data stability indicator.

[0056] The user behavior analysis module screens accounts with key interaction stability based on interaction data stability indicators, calculates the average user stay time and the proportion of secondary interactions (comments after forwarding), and analyzes the content attractiveness of different accounts based on short video, live broadcast, and graphic types to obtain user preference trend values;

[0057] The marketing direction adjustment module selects influencer accounts with the highest display frequency based on user preference trend values, calculates the average content interaction rate for different time periods, identifies user active periods, calculates the distribution time periods of interaction methods (browsing, liking, commenting, and forwarding), selects key secondary interaction rate time periods, and calculates the ad exposure to interaction ratio based on ad exposure to obtain the marketing delivery time frame.

[0058] The account performance prediction module screens influencer accounts with both high exposure and high engagement based on the marketing timeframe. It calculates the fluctuation in growth rate over the past 30 days, uses the account's content type, publishing time, and exposure frequency over the past seven days, and predicts engagement trends over the next seven days to derive an estimated future engagement index.

[0059] The high ROI account screening module uses the future interaction prediction index to screen accounts with predicted interaction values higher than the current average. It obtains the account's historical advertising conversion rate, compares it with the future interaction prediction value, calculates the expected conversion growth rate, and screens accounts with growth rates exceeding the threshold to obtain a list of preferred ROI accounts.

[0060] Interaction data stability indicators include volatility, stability rating, and abnormal account information. User preference trend values include short video preference, live broadcast preference, and image and text preference. The marketing delivery time frame includes optimization period, interaction peak, and advertising exposure rate. The future interaction estimation index includes interaction growth forecast information, content activity expectations, and user engagement estimation results. The preferred ROI account list includes high-efficiency accounts, potential accounts, and priority sorting results.

[0061] See also Figure 2 , the account data monitoring module includes:

[0062] The interactive data collection submodule collects interactive data from influencer accounts, including clicks, viewing time, number of comments, number of shares, and like ratio, and organizes them according to the time dimension to create a time-series interactive data set;

[0063] Collect interactive data of expert accounts, including clicks, viewing time, number of comments, number of shares and like ratio. The data comes from the actual interactive data records of expert accounts on various platforms. Specifically, the clicks of each account are obtained by counting the number of users' clicks on a certain content, and the viewing time is calculated by recording the time each user stays on the content. The number of comments, number of shares and like ratio are all derived from the social media interaction log of the account and sorted by time dimension. In order to ensure the integrity and time consistency of the data, all data are timestamp matched and the data in the same time period are aggregated so that the interactive data of each time period can be compared with each other. For example, in the time period of 10:00-10:30, the clicks of an expert account are The total number of views is 5000, the total viewing time is 12000 seconds, the number of comments is 300, the number of shares is 200, and the number of likes is 1000. Then the interaction data set for this time period should be [5000, 12000, 300, 200, 1000]. The time series data set is arranged in chronological order to ensure that the data of each time period can be accurately classified to the correct time point. For example, if an account records different interaction data at different time periods of the same day (such as 10:00-10:30, 10:30-11:00, and 11:00-11:30), the data needs to be recorded and sorted one by one according to the timeline so that the data of different time periods of the same account form a complete time series, thus obtaining a time series interaction data set.

[0064] The interaction data analysis submodule calls the time series interaction dataset and compares the content interaction ratios in the same time period based on the content release time, display frequency, and display position. The formula is:

[0065]

[0066] Calculate interaction rate I r , which means measuring the intensity of user interaction by the ratio of comments and likes to views and shares, and obtaining the interaction analysis results of each account, where c i represents the number of comments on the i-th content, l i represents the number of likes for the i-th content, v i represents the viewing time of the i-th content, s i Indicates the number of times the i-th content is shared, n o is the amount of content;

[0067] Call the time series interaction dataset and compare the content interaction ratios in the same time period based on the content release time, display frequency, and display position. The content release time is extracted from the timestamp field in the data record, the display frequency is obtained by calculating the number of times the content appears within the specified time range, and the display position is classified according to the display area of the content recorded in the platform log (such as the homepage, recommendation page, search results page), etc. In the same time period, all content is classified according to the release time, display frequency, and display position, and then the interaction of each content is calculated. In the calculation process, for the data of each time period, the number of comments and likes of all contents are summed up, and the viewing time and number of shares of all contents are summed up. For example, in a certain time period, there are three contents with data of (c1=100, l1=200, v1=4000, s1=100), (c2=150, l2=250, v2=5000, s2=120), and (c3=200, l3=300, v3=6000, s3=150). Substitute them into the formula to calculate the interaction rate:

[0068]

[0069] This result indicates that the engagement rate is lower than the common engagement benchmark, meaning that during this time period, although there were a moderate amount of comments and likes, user interaction was not very active relative to the total amount of views and shares.

[0070] The anomaly detection submodule analyzes the volatility of the interaction rate based on the interaction analysis results of each account, filters out accounts with abnormal interaction rate fluctuations, and obtains the interaction data stability index;

[0071] Based on the interaction analysis results of each account, analyze the volatility of the interaction rate and screen out accounts with abnormal fluctuations in the interaction rate. Volatility analysis uses statistical methods to measure the changes in the interaction rate. First, calculate the standard deviation of the interaction rate over a period of time. The standard deviation formula is as follows Among them, N R is the total number of time periods, I r,i is the interaction rate in the i-th time period, is the average interaction rate of the account. For example, if the interaction rates of an account in 5 time periods are (0.05, 0.07, 0.06, 0.08, 0.04), then its average interaction rate is Calculating standard deviation If the standard deviation of an account's interaction rate is higher than the set threshold, for example, 0.1, its interaction data is highly volatile and is defined as an abnormal account. Otherwise, it is defined as a stable account. In this example, 0.1414 is higher than 0.1, so the account is classified as an account with abnormal interaction rate fluctuations, and the interaction data stability index is obtained.

[0072] See also Figure 3 , the user behavior analysis module includes:

[0073] The interaction stability screening submodule calculates the interaction stability value of each account based on the interaction data stability index, screens key accounts that meet the stability requirements, and obtains a list of screened stable accounts;

[0074] Collect user interaction data in different content formats such as short videos, live broadcasts, and pictures and texts, including the number and time distribution of likes, comments, and reposts. Then calculate the standard deviation of each account's interaction within a certain period of time to measure the degree of interaction fluctuation. Assuming that the daily interaction volume of an account in 7 days is 120, 150, 180, 140, 170, 200, and 160 respectively, its standard deviation can be calculated. Calculate σ=24.49, then set the stability threshold. If σ≤30, the account belongs to the key account of interactive stability, otherwise it is eliminated. Then calculate its interactive frequency and use Among them, I i It represents the interaction volume on the i-th day, T is the time span, and N is the number of days. For example, if the total interaction number of an account in 7 days is 1120, then F = 160. Set the benchmark value to 150 and compare it with it. If F is greater than or equal to the benchmark value, it is determined to be an interactive stable account, and a list of stable accounts after screening is obtained.

[0075] The user stay time calculation submodule calculates the average user stay time of each account based on the filtered stable account list using the formula:

[0076]

[0077] Get the calculation result of the stay time, where T avg represents the average length of stay, D i represents the user stay time of the i-th account, W i represents the interaction stability value of the i-th account, and N represents the number of stable accounts after screening;

[0078] Based on the filtered stable account list, we need to first obtain the viewing time D of each user. i and interactive stability value W i , if an account has five users, whose viewing time is 30, 45, 50, 40 and 35 seconds respectively, and the corresponding interaction stability values are 0.9, 1.0, 0.8, 0.95 and 0.85, then the weighted average stay time is calculated as:

[0079]

[0080] Calculate T avg = 39.94 seconds, determine whether it exceeds the set benchmark value of 35 seconds. If it does, the account has a high user retention ability. Otherwise, it needs to be re-evaluated to obtain the retention time calculation result.

[0081] The content attraction analysis submodule analyzes the proportion of secondary interactions among short videos, live broadcasts, and text and images based on the dwell time calculation results. It calculates the content attraction value of each account type and determines the degree of attraction of different account content to users, thereby obtaining the user preference trend value.

[0082] First, we analyze the proportion of secondary interactions for three content types: short videos, live broadcasts, and pictures and texts. Secondary interactions are defined as the behavior of users commenting after forwarding. If the total number of interactions for a short video is 1000, of which the number of secondary interactions is 250, the proportion of secondary interactions is calculated as follows: Calculate the secondary interaction ratio of live broadcast and pictures and text, and compare them. If the secondary interaction ratio of live broadcast and pictures and text is 0.18 and 0.30 respectively, calculate the content attractiveness value of different types of content. If the interaction frequency F is 0.2, then After calculation, the differences between the three are compared to obtain the user preference trend value.

[0083] See also Figure 4 , the marketing direction adjustment module includes:

[0084] The expert account screening submodule calculates the ratio of the total interaction volume and display frequency of each account based on the user preference trend value, screens the expert accounts with the best display frequency, and obtains the best expert account list;

[0085] Get the display frequency of each account, extract the number of exposures of each account in different content types such as short videos, live broadcasts, and pictures and texts within a specific time period, count the daily exposure data, and calculate the average display frequency within a week or a month. Assuming that the display frequencies of an account within 7 days are 500, 520, 510, 495, 530, 515, and 505 respectively, then its average display frequency F is avg The calculation is as follows Then calculate the total number of interactions for each account, including interaction data such as likes, comments, shares, and favorites, and obtain the total number of interactions within the same time period. For example, the total number of interactions for this account within 7 days is 150, 160, 170, 155, 165, 175, and 180, and its total interaction volume I sum For I sum =150+160+170+155+165+175+180=1155, then calculate the ratio of total interactions to display frequency R int for Assume the screening threshold is 0.3, if R int If it is greater than or equal to 0.3, the account will be included in the screening range, otherwise it will be eliminated, thereby screening the expert accounts with the best display frequency and obtaining the best expert account list.

[0086] The active time period identification submodule collects user interactions in different time periods based on the list of top influencer accounts, including browsing, liking, commenting, and forwarding operations. It calculates the average content interaction rate in multiple time periods, filters out peak interaction periods, and counts key secondary interaction rate time periods to obtain user active time period distribution data.

[0087] Based on the list of the best influencer accounts, we collect user interactions in different time periods, including browsing, liking, commenting, and forwarding operations. We set the statistical interval to the hourly level and extract interaction data for each time period within 24 hours every day. For example, the interaction data of an account in different time periods such as 0-1, 1-2, and 2-3 is [20, 25, 18, 30, 50, 65, 90, 110, 140, 180, 230, 250, 270, 300, 280, 260, 220, 190, 170, 150, 130, 110, 85, 50]. Then, we calculate the average content interaction rate for each time period and define the interaction rate calculation formula. Among them, I h is the total amount of interaction in a certain hour, V h is the number of content views in that hour. Assuming the number of interactions at 0-1 is 20 and the number of views is 500, then Calculate the interaction rate for all time periods and filter out the peak interaction period, that is, the time period with significantly higher interaction rate within 24 hours, and further calculate the key secondary interaction rate time period to define the secondary interaction rate. Among them, I sec is the secondary interaction volume in that hour period. Assuming the total interaction volume in a certain period is 200 and the secondary interaction volume is 60, then Assume the baseline is 0.25, if R sec If it is greater than or equal to 0.25, the period is marked as the key secondary interaction period, and the user active period distribution data is obtained.

[0088] The advertisement delivery timing calculation submodule collects advertisement delivery exposures in different time periods based on the user active period distribution data, using the formula:

[0089]

[0090] Calculate the advertising exposure and interaction ratio R for each period ad , get the marketing delivery time frame, where ES i Represents the advertising exposure in the i-th time period, IS i represents the total amount of interaction in the i-th time period, HS i Represents the peak interaction value in the i-th time period, which is used to calculate the interaction peak influence in the time period. i represents the seasonal adjustment coefficient, which takes into account the influence of different seasons or special dates, N S Represents the number of time periods;

[0091] If the ad exposures in different time periods are [500, 600, 700, 800, 900, 1000, 1200], and the corresponding total interactions are [50, 80, 100, 120, 140, 160, 180], calculate the ad exposure to interaction ratio:

[0092]

[0093] Calculated:

[0094]

[0095]

[0096] The marketing delivery time frame was obtained. The result showed that the average advertising interaction ratio was 796.8. The high or low value of the ratio is directly related to the efficiency of advertising delivery. A high value means that advertising during these time periods can obtain a higher user interaction rate. Therefore, it can be used to determine the key data indicators of the most effective advertising delivery time frame.

[0097] See also Figure 5 , the account effect prediction module includes:

[0098] The data screening submodule selects influencer accounts that meet the key requirements of both exposure and engagement based on the marketing delivery timeframe. It then compares the account engagement and exposure data and calculates the exposure-to-engagement ratio for each account. Accounts with ratios within the threshold are retained to obtain the filtered key influencer accounts.

[0099] Based on the marketing delivery time frame, determine the expert account data to be screened, and obtain the exposure and interaction data of the past thirty days. Exposure refers to the total number of times the content published by the expert account is seen by users, and interaction includes the sum of user participation behaviors such as likes, comments, and shares. Next, compare the exposure and interaction data of all expert accounts, and screen out accounts with high exposure and interaction. Set the threshold range of exposure and interaction, for example, the exposure must exceed 50,000 times and the interaction must exceed 5,000 times. Screen the expert accounts that meet the conditions, calculate the exposure and interaction ratio of each expert account, and use the formula Among them, RG represents the ratio of exposure to interaction, EG represents account exposure, and IG represents account interaction. The ratio of each account is calculated in this way, and a threshold range is set. For example, only accounts with a ratio between 10 and 30 will be retained. If an account has an exposure of 60,000 and an interaction of 5,000, its ratio is 12, which meets the screening conditions. If another account has an exposure of 80,000 and an interaction of 2,000, and a ratio of 40, which does not meet the conditions, it will be eliminated and the filtered key influencer accounts will be obtained.

[0100] The growth rate calculation submodule obtains the exposure and interaction data of the past 30 days based on the filtered key influencer accounts, analyzes the daily interaction growth rate, and uses the formula:

[0101]

[0102] Calculate the growth rate change of each account VX and smooth it to get the smoothed growth rate, where GX i represents the growth rate of interaction volume on day i, GX i-1 represents the growth rate of interaction volume on day i-1, n X Represents the calculation period, which is 30 days;

[0103] Based on the filtered key influencer accounts, we obtain their exposure and interaction data for the past thirty days and analyze the daily interaction growth rate. The growth rate is calculated as follows: Among them, GX i Represents the growth rate of interaction on day i, I i Represents the interaction volume on day i, I i-1Represents the interaction volume on the i-1th day to calculate the change in growth rate. For example, if the interaction volume of a certain expert account on the 10th day is 6000 and the interaction volume on the 9th day is 5500, then the interaction growth rate on the 10th day is That is 9.09%. Next, calculate the daily growth rate fluctuation. If the 30-day growth rate data of a certain expert account changes as follows, with 5%, 7%, and 6% on the 1st to 3rd day respectively, then:

[0104]

[0105] The calculation results represent the average rate of change in the daily interaction growth rate of the screened key influencer accounts in the past thirty days, reflecting the stability and volatility of account interactions.

[0106] The interaction trend prediction submodule predicts the interaction volume trend for the next seven days based on the smoothed growth rate and obtains the future interaction estimation index;

[0107] Based on the smoothed growth rate, we obtain the interaction data of key influencer accounts in the past seven days, calculate the interaction trend in the next seven days, and use the weighted average method to predict the future trend. t+1 =I t +(G t ×I t ), where I t+1 represents the predicted interaction volume on day t+1, I t represents the interaction volume on day t, G t represents the growth rate on the tth day. For example, if the interaction volume of a certain expert account on the 30th day is 7,000 and the growth rate is 8%, then I 31 =7000+(0.08×7000)=7560, and so on. Calculate the interaction trend for the next seven days, make deviation adjustments, and obtain the future interaction forecast index.

[0108] See also Figure 6 , high ROI account screening module includes:

[0109] The interaction volume prediction submodule obtains the historical interaction data of the target account based on the future interaction prediction index, calculates the future interaction volume prediction value of each account, and selects accounts with interaction volume prediction values higher than the average and marks them as potential high-conversion accounts;

[0110] Obtain the historical interaction data of the target account, including interaction indicators such as likes, comments, and reposts, and distribute and organize them by day, week, or month to calculate its trend over time. For each account, use regression analysis or time series analysis methods to calculate its future interaction volume forecast value. The forecast calculation can use a sliding average or exponential smoothing model. For example, perform a weighted calculation on the average daily interaction volume of an account in the past 30 days. If the average interaction volume of an account in the past 7 days is 1200 times, and the average in the past 30 days is 800 times, then the future interaction volume forecast value based on exponential smoothing calculation can be P f =α×1200+(1-α)×800, where α is the smoothing coefficient. If α=0.7, then P f =0.7×1200+0.3×800=1080. After calculating the predicted future interaction volume of all accounts, find the average of all accounts. Filter out accounts with predicted interaction values higher than the average. For example, if the average is 950, then accounts with 1080 meet the criteria. Mark the account as a potential high-conversion account and obtain a list of potential high-conversion accounts.

[0111] The conversion rate comparison submodule retrieves historical advertising data based on potential high-conversion accounts, extracts advertising volume, ad clicks, and ad conversions, calculates the historical advertising conversion rate of each account, and obtains historical conversion rate data;

[0112] Based on the list of potential high-conversion accounts, retrieve the historical advertising data of each account, including its advertising volume A, ad click volume C, and ad conversion volume T, and calculate the historical advertising conversion rate. The conversion rate calculation formula is as follows For example, if an account has 50,000 ads served, 5,000 clicks, and 400 conversions in the past month, then Organize the conversion rate data of all potential high-conversion accounts to generate historical conversion rate data.

[0113] The ROI account screening submodule is based on historical conversion rate data: using the formula:

[0114]

[0115] Calculate the expected conversion growth rate enhancement value GU and obtain the preferred ROI account list, where P f represents the predicted value of future interaction volume, P h Represents the historical advertising conversion rate, T i Represents the historical conversion rate of multiple accounts, T m Represents the historical average conversion rate of all accounts, n A Represents the total number of accounts;

[0116] Take a certain account as an example, let its Pf =5.5%, P h =3%, historical delivery conversion rate data is {2.5%, 3.0%, 3.5%, 3.8%, 2.9%}, calculation:

[0117]

[0118]

[0119]

[0120] Filter out accounts whose GU is greater than the set threshold. Set the threshold to 50%, then this account meets the conditions and obtain a list of preferred ROI accounts.

[0121] The above are merely preferred embodiments of the present invention and do not limit the present invention in any other form. Any technician familiar with the profession may use the technical content disclosed above to change or modify it into an equivalent embodiment with equivalent changes and apply it to other fields. However, any simple modification, equivalent change and modification made to the above embodiment based on the technical essence of the present invention without departing from the content of the technical solution of the present invention shall still fall within the scope of protection of the technical solution of the present invention.

Claims

1. An account screening system for improving advertising ROI, characterized by: The system comprises: The account data monitoring module collects interaction data from influencer accounts, organizes it by time, compares content interaction ratios within the same time period, and filters out accounts with abnormal fluctuations in interaction rates to obtain interaction data stability indicators. The account data monitoring module includes: The interactive data collection submodule collects interactive data from influencer accounts, including clicks, viewing time, number of comments, number of shares, and like ratio, and organizes them according to the time dimension to create a time-series interactive data set; The interaction data analysis submodule calls the time series interaction data set and compares the content interaction ratios in the same time period based on the content release time, display frequency, and display position, using the formula: ; Calculating interaction rate , get the interactive analysis results of each account, among which, Indicates the Number of comments per content, Indicates the The number of likes for each piece of content, Indicates the The viewing time of each content, Indicates the Number of times content is shared, is the amount of content; The anomaly detection submodule analyzes the volatility of the interaction rate based on the interaction analysis results of each account, screens accounts with abnormal interaction rate fluctuations, and obtains an interaction data stability index; The user behavior analysis module screens accounts with key interaction stability based on the interaction data stability index, calculates the average user stay time and the proportion of secondary interactions, analyzes the content attractiveness of different accounts, and obtains the user preference trend value; The marketing direction adjustment module calculates the average content interaction rate of different time periods based on the user preference trend value, identifies the user active period, selects the key secondary interaction rate time period, calculates the advertising exposure and interaction ratio, and obtains the marketing delivery time frame; The account effect prediction module selects influencer accounts with both high exposure and high interaction volume based on the marketing delivery time frame, calculates the fluctuation range of growth rate over the past thirty days, and predicts the interaction trend over the next seven days to obtain the future interaction estimation index; The high ROI account screening module screens accounts whose predicted interaction values are higher than the current average based on the future interaction prediction index, compares the predicted interaction values with the future interaction prediction values, calculates the expected conversion growth rate, and obtains an ROI account list.

2. The account screening system for improving advertising ROI according to claim 1 is characterized in that: The interaction data stability indicators include volatility, stability rating, and abnormal account information. The user preference trend values specifically include short video preference, live broadcast preference, and picture and text preference. The marketing delivery time frame specifically includes optimization period, interaction peak, and advertising exposure rate. The future interaction prediction index includes interaction growth forecast information, content activity expectations, and user engagement prediction results. The ROI account list includes efficient accounts, potential accounts, and priority sorting results.

3. The account screening system for improving advertising ROI according to claim 1 is characterized in that: The user behavior analysis module includes: The interaction stability screening submodule calculates the interaction stability value of each account based on the interaction data stability index, screens key accounts that meet the stability requirements, and obtains a screened stable account list; The user stay time calculation submodule calculates the average user stay time of each account based on the screened stable account list using the formula: ; Get the calculation result of the stay time, where represents the average length of stay, Represents the user stay time of the i-th account, Represents the interactive stability value of the i-th account, Represents the number of stable accounts after screening; The content attractiveness analysis submodule analyzes the proportion of secondary interactions of short videos, live broadcasts, and text and pictures based on the calculation results of the dwell time, calculates the content attractiveness value of each account type, and determines the degree of attractiveness of different account contents to users to obtain the user preference trend value.

4. The account screening system for improving advertising ROI according to claim 1 is characterized in that: The marketing direction adjustment module includes: The expert account screening submodule calculates the ratio of the total interaction amount to the display frequency of each account based on the user preference trend value, screens the expert account with the best display frequency, and obtains the best expert account list; The active time period identification submodule collects user interactions in different time periods based on the list of the best influencer accounts, calculates the average content interaction rate in multiple time periods, selects the peak interaction time periods, and counts the key secondary interaction rate time periods to obtain user active time period distribution data; The advertisement delivery timing calculation submodule collects advertisement delivery exposures in different time periods based on the user active time period distribution data, using the formula: ; Calculate the ad exposure to interaction ratio for each time period , get the marketing delivery time frame, where Represents the advertising exposure in the i-th time period, represents the total amount of interaction in the i-th time period, represents the peak value of interaction in the i-th time period, represents the seasonal adjustment coefficient, Represents the number of time periods.

5. The account screening system for improving advertising ROI according to claim 1 is characterized in that: The account effect prediction module includes: The data screening submodule screens influencer accounts that meet the key requirements of both exposure and interaction based on the marketing delivery time frame, compares the account interaction and exposure data, and calculates the exposure-to-interaction ratio of each account. Accounts with ratios within a threshold range are retained to obtain the screened key influencer accounts. The growth rate calculation submodule obtains the exposure and interaction data of the key influencer accounts after screening in the past 30 days, analyzes the daily interaction growth rate, and uses the formula: ; Calculate the change in growth rate for each account , and smoothing is performed to obtain the smoothed growth rate, where represents the growth rate of interaction volume on day i, Represents the growth rate of interaction volume on the i-1th day, Represents the calculation period, which is 30 days; The interaction trend prediction submodule predicts the interaction volume trend in the next seven days based on the smoothed growth rate to obtain a future interaction estimation index.

6. The account screening system for improving advertising ROI according to claim 1 is characterized in that: The high ROI account screening module includes: The interaction volume prediction submodule obtains the historical interaction data of the target account based on the future interaction prediction index, calculates the future interaction volume prediction value of each account, selects accounts with interaction volume prediction values higher than the average, and marks them as potential high-conversion accounts; The conversion rate comparison submodule retrieves historical advertising data based on the potential high-conversion accounts, extracts advertising volume, ad click volume, and ad conversion volume, calculates the historical advertising conversion rate of each account, and obtains historical conversion rate data; The ROI account screening submodule is based on the historical conversion rate data: using the formula: ; Calculate expected conversion growth rate enhancement value , get the ROI account list, where Represents the predicted value of future interaction volume, Represents the historical advertising conversion rate, Represents the historical conversion rate of multiple accounts. Represents the average historical conversion rate of all accounts. Represents the total number of accounts.

Citation Information

Patent Citations

  • Social media platform account number digital marketing release evaluation method

    CN115018661A

  • Advertisement self-adaptive optimization putting monitoring method and device, equipment and medium

    CN118469645A