A real-time income estimation and decision method for mobile APP advertisement conversion reporting

Abstract

The application is suitable for the field of mobile internet advertisement, and provides a real-time income estimation and decision method for mobile APP advertisement conversion reporting, which comprises the following steps: receiving a traffic distribution request from an advertisement business; performing multi-layer filtering and checking on a delivery monitoring link in a mapping relationship to generate a valid candidate delivery monitoring link set under the current request; obtaining real-time conversion rate data and a business value coefficient of each candidate delivery monitoring link in the candidate delivery monitoring link set, and calculating a comprehensive income index of each candidate delivery monitoring link in real time, and sorting the candidate delivery monitoring links according to the comprehensive income index to obtain a sorting result; and generating an intelligent shunting decision by using a hybrid strategy combining main recommendation and exploration recommendation based on the sorting result. The method effectively solves the response lag and scale expansion bottleneck caused by traditional manual static shunting, ensures the optimal path selection of conversion reporting in a complex competitive environment while guaranteeing the maximization of overall income.

Description

Technical Field

[0001] This invention belongs to the field of mobile internet advertising, and in particular relates to a real-time revenue prediction and decision-making method for mobile APP advertising conversion reporting. Background Technology

[0002] In the mobile internet advertising industry, when advertising agencies simultaneously serve multiple advertisers who compete or are substitutes for each other within the same target user group, they typically employ attribution management solutions based on fixed-ratio traffic allocation to achieve ad placement and performance evaluation on the same media platform. In current practice, agency operations staff need to manually configure a static traffic allocation rule for all advertisers in advance, assigning each advertiser a fixed percentage of traffic or conversion reporting. When adding new advertisers or launching new campaigns, they often rely on long-term offline manual testing, manually calculating and solidifying an optimal fixed traffic allocation ratio based on the average performance during the testing period. The majority of traffic is then directed to this optimal advertiser, while a fixed percentage is reserved for other advertisers to maintain client relationships.

[0003] However, the aforementioned existing technical solutions have significant drawbacks: First, management complexity increases explosively and scalability is poor. As the number of competing advertisers handled increases, the traffic allocation combinations and proportions that rely on manual definition and maintenance grow exponentially. This makes them prone to errors and difficult to scale when the business expands, severely restricting operational efficiency. Second, poor real-time decision-making leads to overall revenue loss. Existing static traffic allocation decisions are based on historical offline data and cannot respond to real-time dynamic changes during ad delivery. This prevents the system from allocating every conversion opportunity in real time to the advertiser with the highest bid or the highest revenue, thus preventing the advertising agency from maximizing the overall revenue of the advertiser portfolio managed by the agency. Summary of the Invention

[0004] The purpose of this invention is to provide a real-time revenue prediction and decision-making method for mobile APP advertising conversion reporting, aiming to solve the problems mentioned in the background art.

[0005] This invention is implemented as follows: a real-time revenue prediction and decision-making method for mobile app advertising conversion reporting, the method comprising: Step 1: Receive a traffic allocation request from the advertising business. The request shall at least include the advertiser link identifier and user identifier of the traffic to be allocated, and obtain the pre-cached mapping relationship between the advertiser link identifier and the product, port and delivery monitoring link. Step 2: Perform multi-layer filtering and verification on the delivery monitoring links in the mapping relationship to generate a set of valid candidate delivery monitoring links for the current request; Step 3: Obtain the real-time conversion rate data and business value coefficient of each candidate ad placement monitoring link in the candidate ad placement monitoring link set, and calculate the comprehensive revenue index of each candidate ad placement monitoring link in real time. and according to the aforementioned comprehensive benefit index Perform sorting to obtain the sorted result; Step 4: Based on the ranking results, a hybrid strategy combining primary recommendation and exploratory recommendation is adopted to generate intelligent traffic diversion decisions, and the diversion ratio is dynamically adjusted for each time in combination with the data indicator requirements customized by the business. Step 5: Return the intelligent traffic diversion decision results to the advertising business system to execute the final ad placement or conversion reporting.

[0006] This invention provides a real-time revenue prediction and decision-making method for mobile app advertising conversion reporting. By establishing a multi-layered dynamic filtering mechanism and a revenue prediction model based on real-time conversion rates, it achieves minute-level accurate decision-making for traffic allocation in advertising campaigns. It effectively solves the response lag and scalability bottlenecks caused by traditional manual static traffic allocation. Through a hybrid strategy of primary recommendation and exploratory recommendation, along with a cold start optimization algorithm, it maximizes overall revenue while improving the efficiency of new business integration and system resilience, ensuring the optimal path selection for conversion reporting in complex competitive environments. Attached Figure Description

[0007] Figure 1 This is the main flowchart of a real-time revenue forecasting and decision-making method for mobile app advertising conversion reporting. Detailed Implementation

[0008] To make the objectives, technical solutions, and advantages of this invention clearer, the 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 merely illustrative and not intended to limit the invention.

[0009] The specific implementation of the present invention will be described in detail below with reference to specific embodiments.

[0010] This invention provides a real-time revenue prediction and decision-making method for mobile APP advertising conversion reporting, which solves the technical problems in the background art.

[0011] like Figure 1 The diagram shown is a main flowchart of a real-time revenue estimation and decision-making method for mobile app advertising conversion reporting, provided by an embodiment of the present invention. The real-time revenue estimation and decision-making method for mobile app advertising conversion reporting includes: Step 1: Receive a traffic allocation request from the advertising business. The request shall at least include the advertiser link identifier and user identifier of the traffic to be allocated, and obtain the pre-cached mapping relationship between the advertiser link identifier and the product, port and delivery monitoring link. Step 2: Perform multi-layer filtering and verification on the delivery monitoring links in the mapping relationship to generate a set of valid candidate delivery monitoring links for the current request; Step 3: Obtain the real-time conversion rate data and business value coefficient of each candidate ad placement monitoring link in the candidate ad placement monitoring link set, and calculate the comprehensive revenue index of each candidate ad placement monitoring link in real time. and according to the aforementioned comprehensive benefit index Perform sorting to obtain the sorted result; Step 4: Based on the ranking results, a hybrid strategy combining primary recommendation and exploratory recommendation is adopted to generate intelligent traffic diversion decisions, and the diversion ratio is dynamically adjusted for each time in combination with the data indicator requirements customized by the business. Step 5: Return the intelligent traffic diversion decision results to the advertising business system to execute the final ad placement or conversion reporting.

[0012] In this embodiment, when the advertising system detects a user click or activation triggering a conversion reporting request, step one involves receiving a traffic allocation request in real time via a high-concurrency API interface. The advertiser link identifier and user identifier carried in this request are the core raw data for decision-making. By accessing a pre-stored mapping table in the in-memory database, a topological relationship is instantly established between the advertiser link and the specific backend product, logical port, and several underlying candidate placement monitoring links. In step S2, the system performs multi-dimensional filtering on these links within milliseconds, eliminating invalid paths that are closed or do not match the current user profile. Step three calls the real-time streaming computing unit, combining the historical conversion rate of each link with the commercial value coefficient calculated from the advertiser's real-time bid, assigning a quantified revenue score to each link and prioritizing them. In step four, instead of simply adopting a linear logic of highest bidder wins, an exploration mechanism is introduced. While ensuring the main link receives a large amount of traffic, a small proportion of traffic is allocated to high-potential links to detect market fluctuations. Finally, in step five, the decision result is sent back to the business system via a JSON message, completing a closed-loop conversion guidance process.

[0013] In a preferred embodiment of the present invention, the step of performing multi-layer filtering and verification on all delivery monitoring links to generate a set of valid candidate delivery monitoring links for the current request specifically includes: Based on the advertiser link identifier, query the product to which it belongs, and filter out the list of all configured ports under that product; The monitoring links under the port list are queried, and only those monitoring links that are in the listed state and whose daily traffic and consumption have not reached the preset threshold are retained to form a preliminary candidate set. The real-time query interface of the business system is invoked, and the advertiser link identifier and user identifier are used as parameters to match and verify the delivery monitoring links in the preliminary candidate set. Only delivery monitoring links that meet the targeting conditions are retained to form the final valid candidate delivery monitoring link set.

[0014] In this embodiment, product and rule filtering is performed first. Based on the mapping relationship, the business group to which the advertiser belongs is determined. For example, if the traffic is determined to belong to the "Fantasy / Martial Arts Games" product line, the list of all online ports under that product line is automatically retrieved. Next, the status filtering stage begins, querying the survival status and daily budget consumption metrics of each ad placement monitoring link in real time. If the advertiser's account balance corresponding to a link is lower than the warning value, or if the number of conversions it has received that day has reached the set threshold, the link will be blocked in real time to prevent wasted resources. Furthermore, the real-time interface filtering layer demonstrates dynamism. The system uses the user identifier in the request as the key and transmits it to the business risk control interface to verify whether the user is a long-time user of the product or a blacklisted user. If the interface reports that the user does not meet the targeting conditions of a specific ad placement monitoring link, the link is removed. Through this progressive filtering process, it is ensured that every member in the candidate ad placement monitoring link set is legally valid and business-accessible.

[0015] As a preferred embodiment of the present invention, the comprehensive benefit index The calculation formula is: ; In the formula, This is the commercial value coefficient. Let these be the first preset time period, the second preset time period, ..., the nth preset time period, respectively. These are the corresponding weighting coefficients; In application, the commercial value coefficient of this embodiment... This is typically calculated by combining the advertiser's base bid (CPA / CPS) with the current conversion weight. The first preset time period uses the real-time conversion rate of the most recent 5 minutes, with weighting... It can be set to 0.5; the second preset time period takes the conversion rate of the most recent hour, with a weight. It can be set to 0.3; the third preset time period takes the cumulative conversion rate of the day, with a weight. It can be set to 0.2. The internal scheduling task pulls the latest conversion feedback from the OLAP database every minute or less, dynamically recalculating the overall revenue metrics for all valid candidate links. The sorting process is completed in a distributed cache, ensuring that when a request arrives, the pre-sorted descending sequence can be read directly, greatly reducing computational latency and thus supporting real-time decision-making responses at the second level.

[0016] In a preferred embodiment of the present invention, the step of generating intelligent traffic allocation decisions based on the ranking results using a hybrid strategy combining primary recommendation and exploratory recommendation, and dynamically adjusting the traffic allocation ratio for each allocation based on business-defined data metrics, specifically includes: Based on the ranking results, select the comprehensive benefit index. The top-ranked candidate ad placement monitoring link is selected as the primary recommendation, and a pre-defined high proportion of traffic is allocated to this primary recommendation. Based on the ranking results, for each port other than the port to which the main recommendation item belongs, their comprehensive revenue index is selected respectively. The highest-ranking candidate delivery monitoring link is selected as the exploration recommendation, and the remaining low-proportion preset traffic allocation is directed to the exploration recommendation.

[0017] It should be understood that the preset primary recommendation traffic ratio is 90%, and the exploration traffic ratio is 10%. When determining the primary recommendation item, the comprehensive revenue indicator in the ranking results is directly locked. The highest-ranking candidate ad placement monitoring link is designated as the absolute mainstay of the current business cycle, undertaking the vast majority of conversion reporting tasks. When handling the remaining 10% of exploratory traffic, a port avoidance principle is adopted: to avoid putting all eggs in one basket, the port associated with the main recommendation is skipped, and instead, comprehensive revenue metrics are sought across different ports. The optimal link is selected as the exploration item. If the main recommendation comes from Tencent's channel, the exploration item will be selected from Toutiao or Kuaishou channels. This diversified exploration method can effectively prevent the model from getting stuck in local optima. When the advertising environment changes drastically, it can quickly detect newly launched high-quality links by exploring traffic, and achieve a smooth shift in the center of revenue.

[0018] In a preferred embodiment of the present invention, the specific logic for dynamically adjusting the diversion ratio for each instance is as follows: Based on the aforementioned data indicator requirements, the diversion ratio is weighted or deweighted by comprehensively analyzing the relevant indicator prediction data of all products. If the comprehensive return index If the highest-ranking candidate ad monitoring link causes the overall conversion rate to fail to meet the data metric requirements, the actual reporting traffic allocation ratio assigned to that candidate ad monitoring link will be reduced in real time.

[0019] In this embodiment, the average overall conversion rate of all products is monitored in real time during application. If the selected top-ranked candidate link has a high overall revenue index... If the conversion rate is extremely high, but the estimated conversion rate is too low, and this would cause the overall conversion rate of the product to fall below the advertiser's target range, then a demotion adjustment will be triggered. The adjustment logic is achieved by multiplying the traffic allocation ratio by a penalty coefficient, redistributing the excess traffic to the next-best links with more stable conversion rates. In addition, it will also take into account the daily progress requirements of each product group. If a product has completed 95% of its task for the day and its conversion cost exceeds the target, the overall traffic allocation weight of all links under that product will be automatically reduced, achieving optimal global returns and risk control across products and links.

[0020] As a preferred embodiment of the present invention, it further includes a cold start processing step for newly added campaign monitoring links with insufficient historical conversion data, wherein the cold start processing step calculates the comprehensive revenue index in step three. Previous execution: For newly added campaign monitoring links in the candidate campaign monitoring link set that lack sufficient historical conversion data, a cold start optimization algorithm is used to estimate their initial comprehensive revenue index. This allows it to participate in the sorting process in step three.

[0021] In this embodiment, after reading the candidate campaign monitoring link set, the lifecycle attribute of each link is first checked: if the activation duration of the link is less than 24 hours, or the cumulative number of click samples received is lower than a preset significance threshold (e.g., 500 clicks), it is marked as a new campaign monitoring link. At this time, the reading of its real metrics is temporarily skipped, and the cold start optimization algorithm is forcibly executed. This allows it to enter the sorting stage with extremely high priority or protected priority, ensuring that the new link can obtain the necessary raw test traffic, thereby accumulating real data in a short period of time and quickly passing through the data vacuum period.

[0022] In a preferred embodiment of the present invention, the cold start optimization algorithm includes the following steps: Calculate the similarity between the newly added campaign monitoring link and the existing mature campaign monitoring link in terms of multi-dimensional features, wherein the multi-dimensional features include at least product type, target audience and bidding mode; Based on the similarity score, several mature deployment monitoring links that are most similar were selected; Using the Bayesian averaging algorithm, the initial prior value of the newly added monitoring links is smoothly fused with the historical performance data of the selected mature monitoring links to calculate the initial comprehensive revenue index of the newly added monitoring links. .

[0023] In this embodiment, a feature vector is constructed for each link, covering dimensions such as product type, target audience, and bidding mode. By calculating cosine similarity, 5-10 mature links with the closest features are retrieved from a massive historical link database. Next, the algorithm does not directly borrow the values ​​from mature links, but instead uses a Bayesian averaging algorithm for smoothing. The specific formula logic is as follows: the initial prior value of the new link is used as a baseline and assigned a large confidence weight; as the number of real samples of the new link increases, the weight of the real data gradually increases in the calculation, while the weight of the prior value gradually decreases. The resulting initial comprehensive revenue index... It retains the accuracy of industry experience while also possessing the ability to self-correct.

[0024] The above embodiments of the present invention provide a real-time revenue estimation and decision-making method for mobile app advertising conversion reporting. By establishing a multi-layered dynamic filtering mechanism and a revenue estimation model based on real-time conversion rates, it achieves minute-level accurate decision-making for traffic allocation in advertising campaigns. This solution effectively solves the response lag and scalability bottlenecks caused by traditional manual static traffic allocation. Through a hybrid strategy of primary recommendation and exploratory recommendation, and a cold start optimization algorithm, it maximizes overall revenue while improving the efficiency of new business integration and the system's resilience, ensuring the optimal path selection for conversion reporting in complex competitive environments.

[0025] In order for the above methods and systems to operate smoothly, the system may include more or fewer components than those described above, or combine certain components, or different components, in addition to the various modules mentioned above. For example, it may include input / output devices, network access devices, buses, processors, and memory.

[0026] The processor can be a central processing unit, or other general-purpose processors, digital signal processors, application-specific integrated circuits (ASICs), off-the-shelf programmable gate arrays (OPGs) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. A general-purpose processor can be a microprocessor or any conventional processor. This processor is the control center of the system, connecting various parts via various interfaces and lines.

[0027] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0028] The embodiments described above are merely illustrative of several implementations of the present invention, and while the descriptions are specific and detailed, they should not be construed as limiting the scope of the present invention. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of the present invention, and these modifications and improvements all fall within the scope of protection of the present invention. Therefore, the scope of protection of this patent should be determined by the appended claims.

[0029] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A method for real-time revenue estimation and decision-making in mobile app advertising conversion reporting, characterized in that, The method includes: Step 1: Receive a traffic allocation request from the advertising business. The request shall at least include the advertiser link identifier and user identifier of the traffic to be allocated, and obtain the pre-cached mapping relationship between the advertiser link identifier and the product, port and delivery monitoring link. Step 2: Perform multi-layer filtering and verification on the delivery monitoring links in the mapping relationship to generate a set of valid candidate delivery monitoring links for the current request; Step 3: Obtain the real-time conversion rate data and business value coefficient of each candidate ad placement monitoring link in the candidate ad placement monitoring link set, and calculate the comprehensive revenue index of each candidate ad placement monitoring link in real time. and according to the aforementioned comprehensive benefit index Perform sorting to obtain the sorted result; Step 4: Based on the ranking results, a hybrid strategy combining primary recommendation and exploratory recommendation is adopted to generate intelligent traffic diversion decisions, and the diversion ratio is dynamically adjusted for each time in combination with the data indicator requirements customized by the business. Step 5: Return the intelligent traffic diversion decision results to the advertising business system to execute the final ad placement or conversion reporting.

2. The real-time revenue estimation and decision-making method for mobile APP advertising conversion reporting according to claim 1, characterized in that, The step of performing multi-layer filtering and verification on all delivery monitoring links to generate a set of valid candidate delivery monitoring links for the current request specifically includes: Based on the advertiser link identifier, query the product to which it belongs, and filter out the list of all configured ports under that product; The monitoring links under the port list are queried, and only those monitoring links that are in the listed state and whose daily traffic and consumption have not reached the preset threshold are retained to form a preliminary candidate set. The real-time query interface of the business system is invoked, and the advertiser link identifier and user identifier are used as parameters to match and verify the delivery monitoring links in the preliminary candidate set. Only delivery monitoring links that meet the targeting conditions are retained to form the final valid candidate delivery monitoring link set.

3. The real-time revenue estimation and decision-making method for mobile APP advertising conversion reporting according to claim 1, characterized in that, The comprehensive income indicator The calculation formula is: ； In the formula, This is the commercial value coefficient. Let these be the first preset time period, the second preset time period, ..., the nth preset time period, respectively. These are the corresponding weighting coefficients.

4. The real-time revenue estimation and decision-making method for mobile APP advertising conversion reporting according to claim 1, characterized in that, The process of generating intelligent traffic allocation decisions based on the ranking results using a hybrid strategy combining primary recommendation and exploratory recommendation, and dynamically adjusting the allocation ratio for each traffic allocation based on business-defined data metrics, specifically includes: Based on the ranking results, select the comprehensive benefit index. The top-ranked candidate ad placement monitoring link is selected as the primary recommendation, and a pre-defined high proportion of traffic is allocated to this primary recommendation. Based on the ranking results, for each port other than the port to which the main recommendation item belongs, their comprehensive revenue index is selected respectively. The highest-ranking candidate delivery monitoring link is selected as the exploration recommendation, and the remaining low-proportion preset traffic allocation is directed to the exploration recommendation.

5. The real-time revenue estimation and decision-making method for mobile APP advertising conversion reporting according to claim 1, characterized in that, The specific logic for dynamically adjusting the diversion ratio for each instance is as follows: Based on the aforementioned data indicator requirements, the diversion ratio is weighted or deweighted by comprehensively analyzing the relevant indicator prediction data of all products. If the comprehensive return index If the highest-ranking candidate ad monitoring link causes the overall conversion rate to fail to meet the data metric requirements, the actual reporting traffic allocation ratio assigned to that candidate ad monitoring link will be reduced in real time.

6. The real-time revenue estimation and decision-making method for mobile APP advertising conversion reporting according to claim 1, characterized in that, It also includes a cold start processing step for new ad monitoring links with insufficient historical conversion data, wherein the cold start processing step calculates the comprehensive revenue index in step three. Previous execution: For newly added campaign monitoring links in the candidate campaign monitoring link set that lack sufficient historical conversion data, a cold start optimization algorithm is used to estimate their initial comprehensive revenue index. This allows it to participate in the sorting process in step three.

7. The real-time revenue estimation and decision-making method for mobile APP advertising conversion reporting according to claim 6, characterized in that, The process of the cold start optimization algorithm includes: Calculate the similarity between the newly added campaign monitoring link and the existing mature campaign monitoring link in terms of multi-dimensional features, wherein the multi-dimensional features include at least product type, target audience and bidding mode; Based on the similarity score, several mature deployment monitoring links that are most similar were selected; Using the Bayesian averaging algorithm, the initial prior value of the newly added monitoring links is smoothly fused with the historical performance data of the selected mature monitoring links to calculate the initial comprehensive revenue index of the newly added monitoring links. .

Images