CN-121981778-A - Advertisement delivery ROI dynamic optimization method and system based on real-time data driving

CN121981778ACN 121981778 ACN121981778 ACN 121981778ACN-121981778-A

Abstract

The application discloses a dynamic optimization method and system for advertisement delivery ROI based on real-time data driving, which relates to the technical field of Internet advertisement service, the proposal carries out calibration on the acquired ROI analysis data and acquires the calibration result, evaluates whether to carry out the filtering operation of the ROI analysis data according to the calibration result, carries out advertisement delivery split evaluation based on the acquired ROI analysis data after the calibration is completed, and acquires the evaluation result, whether attribution contribution evaluation optimization is carried out is assessed according to the evaluation result, after the advertisement putting split evaluation is carried out, linkage analysis is carried out based on the obtained ROI analysis data to assess the contribution degree of conversion of each advertisement platform, the analysis result is obtained, whether ROI dynamic optimization is carried out is assessed according to the analysis result, and the problem that the dynamic optimization flexibility of the whole advertisement putting ROI is limited greatly is solved.

Inventors

Liang Rongda
CHEN HONGLIE

Assignees

广州市启点创意科技有限公司

Dates

Publication Date: 20260505
Application Date: 20260129

Claims (10)

1. The method for dynamically optimizing the advertisement delivery ROI based on real-time data driving is characterized by comprising the following steps of: performing calibration on the acquired ROI analysis data, acquiring a calibration result, and evaluating whether to perform ROI analysis data filtering operation according to the calibration result; After calibration is completed, advertisement delivery split evaluation is executed based on the acquired ROI analysis data, an evaluation result is acquired, and whether attribution contribution evaluation optimization is carried out is evaluated according to the evaluation result; After the advertisement putting splitting evaluation is completed, linkage analysis is performed based on the acquired ROI analysis data to evaluate the contribution degree of conversion of each advertisement platform, analysis results are acquired, and whether ROI dynamic optimization is carried out is evaluated according to the analysis results.
2. The method for dynamic optimization of ROI based on real-time data driven advertising as claimed in claim 1 wherein said calibration of the acquired ROI analysis data is performed as follows: The time difference between the moment when the management platform starts to receive the ROI analysis data and the moment when the weight evaluation analysis is carried out on the ROI analysis data is represented as the ROI analysis data evaluation duration for evaluating the weight instantaneity; Judging whether the ROI analysis data evaluation time length is in the reference range of the ROI analysis data evaluation time length; If the ROI analysis data evaluation time length is larger than the maximum value in the ROI analysis data evaluation time length reference range, performing ROI analysis data filtering operation on the next acquired ROI analysis data; If the evaluation duration of the ROI analysis data is smaller than the minimum value in the reference range of the evaluation duration of the ROI analysis data, sending a prompt of disqualification of the calibration of the ROI analysis data to a preset person; if the ROI analysis data evaluation duration is in the reference range of the ROI analysis data evaluation duration, advertisement delivery split evaluation is executed.
3. The real-time data driven based dynamic optimization method of advertisement delivery ROI as set forth in claim 2, wherein the specific process of the ROI analysis data filtering operation is as follows: filtering invalid data based on the edge computing node; Transmitting the filtered ROI analysis data to a distributed real-time database in real time for aggregation storage and generating an acquisition log so as to acquire the ROI analysis data of the advertisement platform of the full node of the conversion link; performing data cleaning processing based on the edge computing node; after the conversion link full-node priority adjustment is carried out based on the data transmission priority scheduling algorithm so as to improve the transmission speed of the ROI analysis data, the corresponding ROI analysis data is used for carrying out advertisement delivery split evaluation; The conversion link full-node priority adjustment refers to transmitting the ROI analysis data on the conversion link nodes to the management platform on the basis of the preset node priority ordering on the conversion link.
4. The real-time data driven based dynamic optimization method for advertisement delivery ROI as set forth in claim 3, wherein the specific process of advertisement delivery split evaluation is as follows: Matching the acquired ROI analysis data with the unique identifier of the management platform, specifically: classifying and collecting the collected ROI analysis data of each advertising platform by taking the unique identification of the management platform as a core matching field, and counting the independent ROI analysis data of each advertising platform, wherein the ROI analysis data comprises advertising delivery schedule data, conversion node interaction data and transaction conversion data; Screening and collecting all the independent ROI analysis data of the advertising platforms which are completely matched with the unique identifiers of the management platforms, grouping and marking the independent ROI analysis data as qualified ROI analysis data, and simultaneously, independently classifying and marking part of all the independent ROI analysis data of the advertising platforms which cannot be matched with the unique identifiers of the management platforms as abnormal data to realize consistency check and comparison, and acquiring the split and put times of all the advertising platforms based on the qualified ROI independent analysis data acquired after the complete matching; The absolute value of the difference value between the split putting times of each advertisement platform and the corresponding preset putting times of the advertisement platform is expressed as an advertisement putting splitting error for reflecting the matching degree of the single-platform data and the total data dimension; Judging whether the advertisement putting splitting error is smaller than a preset error threshold value or not; And if the advertisement putting splitting error is greater than or equal to a preset error threshold value, carrying out attribution contribution evaluation optimization, otherwise, executing ROI analysis data linkage analysis.
5. The real-time data driven based dynamic optimization method of advertisement placement ROI as set forth in claim 4 wherein said attribution contribution assessment optimization is performed as follows: Sending a prompt to preset personnel, and increasing the weight ratio of the browsing duration of the advertisement platform on the basis of the weight of the browsing duration of the original advertisement platform based on a preset attribution algorithm; If the advertisement putting splitting error obtained again after the attribution contribution evaluation optimization is smaller than a preset error threshold, judging that the splitting result is qualified, and executing the ROI analysis data linkage analysis; if the advertisement delivery splitting error acquired again after the contribution evaluation optimization is still larger than the preset error threshold, the ROI analysis data of all the nodes of the link are synchronously converted based on the distributed data consistency protocol, and the specific process is as follows: based on a distributed data consistency protocol, synchronously calibrating the ROI analysis data of other distributed nodes by taking a core acquisition node as a reference; Acquiring and summarizing the ROI analysis data of each advertisement platform again, executing the advertisement putting splitting evaluation flow again, acquiring advertisement putting splitting errors again, and executing the ROI analysis data linkage analysis if the advertisement putting splitting errors are smaller than a preset error threshold value; if the advertisement putting splitting error is still greater than or equal to a preset error threshold value, based on the obtained Shapley value, attribution contribution evaluation optimization is realized.
6. The real-time data driven based dynamic optimization method of advertisement placement ROI as set forth in claim 4 wherein said attribution contribution assessment optimization is performed as follows: Acquiring exclusive data dimension required by Shapley values; Acquiring each contact weight based on the Shapley value, summarizing each contact weight to a management platform, and finally verifying a preset attribution algorithm based on the AB test and acquiring an advertisement contribution coefficient; and acquiring optimal advertisement contribution coefficients of each advertisement platform in the full conversion link based on a preset attribution algorithm, and executing ROI analysis data linkage analysis.
7. The real-time data driven based dynamic optimization method of advertisement delivery ROI as set forth in claim 6, wherein the specific process of ROI analysis data linkage analysis is as follows: sending a prompt to preset personnel to set a contribution coefficient threshold in advance so as to distinguish a core platform from a cooperative platform; acquiring weight evaluation accuracy; judging whether the weight evaluation accuracy is greater than or equal to a weight evaluation accuracy reference value; if the weight evaluation accuracy is smaller than the weight evaluation accuracy reference value, a prompt for increasing the weight of the preset exclusive feature dimension is sent to preset personnel, and otherwise, the ROI is dynamically optimized.
8. The real-time data driven based dynamic optimization method of an advertisement delivery ROI as set forth in claim 7, wherein the specific process of the dynamic optimization of the ROI is as follows: Based on advertisement contribution coefficients corresponding to all advertisement platforms, obtaining the number of budget allocation instructions corresponding to all advertisement platforms according to preset budget allocation proportion; The budget allocation instruction is issued to each advertisement platform delivery interface in real time, and a budget allocation log is recorded; Collecting optimized platform throwing data in real time; the optimized delivery data of each platform refers to ROI analysis data generated by each advertisement platform after budget allocation instructions are executed; for the core platform, the data volume of the ROI analysis data of the core platform is acquired, expressed as the core ROI analysis data volume, and the core ROI analysis data transmission analysis is performed.
9. The real-time data driven based dynamic optimization method of advertisement delivery ROI as set forth in claim 8, wherein the specific process of the core ROI analysis data transmission analysis is as follows: The time mark required by the core platform to start transmitting the core ROI analysis data to the management platform to receive the core ROI analysis data is used as the core platform transmission time length; the core ROI analysis data refers to the ROI analysis data corresponding to the core platform transmitted to the management platform; judging whether the transmission time length of the core platform is smaller than the transmission time length of a preset core platform, and whether the analysis data size of the core ROI is larger than or equal to the analysis data size of the preset core ROI; if the transmission time length of the core platform is smaller than the transmission time length of the preset core platform and the core ROI analysis data volume is smaller than the preset core ROI analysis data volume, marking the corresponding core ROI analysis data as medium priority ROI analysis data; If the transmission time length of the core platform is longer than or equal to the transmission time length of the preset core platform and the core ROI analysis data volume is longer than or equal to the preset core ROI analysis data volume, marking the corresponding core ROI analysis data as medium priority ROI analysis data; Caching the corresponding medium-priority ROI analysis data to an edge node, and storing the medium-priority ROI analysis data after the high-priority ROI analysis data is stored; If the transmission time length of the core platform is smaller than the transmission time length of the preset core platform and the core ROI analysis data volume is larger than or equal to the preset core ROI analysis data volume, marking the corresponding core ROI analysis data as high-priority ROI analysis data, and sending an ROI dynamic optimization qualified prompt to preset personnel; If the transmission time length of the core platform is not less than the preset transmission time length of the core platform and the core ROI analysis data volume is less than the preset core ROI analysis data volume, marking the corresponding core ROI analysis data as low-priority ROI analysis data, and sending an ROI dynamic optimization abnormality prompt to preset personnel; aiming at the cooperative platform, acquiring the data volume of the ROI analysis data of the cooperative advertisement platform, and representing the data volume as the cooperative ROI analysis data volume; If the collaborative ROI analysis data volume is larger than or equal to the preset collaborative ROI analysis data volume, marking the corresponding data as medium-priority ROI analysis data, caching the collaborative ROI analysis data to an edge node, and transmitting after the transmission of the high-priority ROI analysis data is completed; And otherwise, sending a cooperative ROI analysis data volume abnormality prompt to a preset person.
10. The real-time data-driven advertisement delivery ROI dynamic optimization system is used for realizing the real-time data-driven advertisement delivery ROI dynamic optimization method according to any one of claims 1-9, and is characterized by comprising an ROI analysis data calibration module, an advertisement delivery split evaluation module and an ROI analysis data linkage analysis module; the ROI analysis data calibration module is used for executing calibration on the acquired ROI analysis data and acquiring a calibration result, and evaluating whether to carry out the ROI analysis data filtering operation according to the calibration result; The advertisement putting splitting evaluation module is used for executing advertisement putting splitting evaluation based on the acquired ROI analysis data after calibration is completed, acquiring an evaluation result and evaluating whether attribution contribution evaluation optimization is carried out according to the evaluation result; And the ROI analysis data linkage analysis module is used for executing linkage analysis based on the acquired ROI analysis data to evaluate the contribution degree of conversion of each advertisement platform after the advertisement delivery splitting evaluation is completed, acquiring an analysis result, and evaluating whether the ROI dynamic optimization is carried out according to the analysis result.

Description

Advertisement delivery ROI dynamic optimization method and system based on real-time data driving Technical Field The invention relates to the technical internet advertisement service field, in particular to a real-time data-driven advertisement delivery ROI dynamic optimization method and system. Background With the deep popularization of internet technology, the full coverage of a 5G network and the wide penetration of intelligent terminal equipment, the digital advertising industry has entered a key stage of coexistence of scale expansion and quality upgrading, and has become brand marketing, especially has become a core matrix for reaching users and realizing business transformation by financial brands, under the background, the requirements of advertisers in the financial industry on quantitative evaluation of the delivery effect are increasingly urgent, and ROI (Return on Investment ) has become a core index for measuring the value of financial marketing activities. However, the traditional advertisement putting mode is difficult to adapt to the current fragmented and personalized propagation environment, various structural pain points obviously restrict the promotion of the ROI, a core driving force is provided for technical innovation and mode reconstruction, the traditional advertisement putting is widely dependent on a rough strategy driven by experience, and the traditional advertisement putting mode is widely covered as a core target and lacks of accurate positioning of target audiences, so that advertisement resource waste is serious. More importantly, the cross-platform data island phenomenon is prominent, user behavior data are scattered in different ecologies such as social media, search engines and electronic commerce platforms, the data interface opening rate is less than 30%, and the cross-platform user portrait matching accuracy is less than 60%, so that advertisers cannot comprehensively capture the full-link value from cognition to conversion of users. Meanwhile, the existing evaluation system excessively depends on short-term indexes such as click rate and conversion rate, and neglects the association of long-term value precipitation of brands and life cycle value of users, so that budget allocation is unbalanced, a large amount of investment is consumed on low-efficiency flow, and the superposition of false flow accounts for up to 18% of industry mess, so that ROI calculation distortion and advertiser loss are further aggravated. The method comprises the steps of collecting advertisement display quantity, click quantity, conversion quantity, consumption cost and other throwing data in a minute-level period through a multi-platform data interface, predicting the ROI of each advertisement platform in a future period by utilizing a machine learning model based on history and real-time data, constructing an optimization problem aiming at maximizing the whole ROI under the constraint of total budget, solving an optimal budget allocation strategy of each platform in each time period through an optimization algorithm, synchronizing the budget allocation strategy to the advertisement platform through an automatic control module, executing the budget allocation strategy, and recording the ROI change before and after optimization through a monitoring module to realize model iterative updating. On the basis of the above-mentioned patent framework, in order to realize truly landable, high-response dynamic optimization system, the prior art realizes the high frequency data synchronization of minute level, and through multidimensional feature extraction and modeling of platform and period, the ROI prediction has stronger time sequence adaptability and platform specificity, thus provide reliable basis for the accurate decision of budget allocation, this scheme also adapts the core demand of financial advertisement compliance, high risk management and control, and through strengthening the accurate modeling of the exclusive dimension of advertisement putting in the financial industry and platform, period and financial product type, the ROI prediction has stronger time sequence adaptability, platform specificity and financial product adaptability, its concrete implementation mode is as follows: In the prior art, in order to achieve the improvement of the flexibility of dynamic continuous optimization and supervision of a financial advertisement ROI, firstly, the data interface of the advertisement platform is connected through the management platform of the financial advertisement, key delivery data are collected in real time at a minute-level period (preferably once every 5 minutes) to ensure the timeliness of the data, the key delivery data comprise advertisement display quantity, click quantity and the like, financial product type labels (such as credit products, fund financing and the like), the fit scores of platform user images and financial product risk levels, intention user asset level estim