A method for accurately evaluating advertising effectiveness based on contextualized ad delivery

Abstract

This invention discloses a method for accurately evaluating advertising effectiveness based on contextualized ad placement, belonging to the field of advertising effectiveness evaluation technology. It aims to solve the problems of existing evaluation methods neglecting contextual factors, low accuracy, lack of standardization, and inability to dynamically optimize. The method includes constructing a standardized contextual tagging system covering all user touchpoints, standardizing data collection and preprocessing across the entire advertising placement chain, achieving accurate matching between placement data and contextual tags, layering and decomposing performance indicators and setting contextual baseline values, conducting contextualized attribution analysis to remove non-contextual interference, generating multi-dimensional evaluation results and optimization directions, and continuously improving evaluation accuracy through comparative verification and dynamic iteration mechanisms. This invention solves the problems of chaotic contextual classification and large evaluation biases, improving the accuracy and practicality of advertising effectiveness evaluation, and can be widely applied to various contextualized ad placement scenarios such as internet advertising and offline contextual advertising.

Description

Technical Field

[0001] This invention relates to the field of advertising delivery technology, specifically to a method for accurately evaluating advertising effectiveness based on contextualized delivery. Background Technology

[0002] With the continuous development of advertising technology, contextualized advertising has become a core means to improve advertising effectiveness. Its core is to accurately match advertising content with the characteristics of the user's context, so as to "deliver the right ad to the right user in the right context". However, current advertising effectiveness evaluation methods still have significant shortcomings and are difficult to meet the accurate evaluation needs of contextualized advertising.

[0003] Current advertising performance evaluations primarily focus on basic data such as impressions, clicks, and conversions, neglecting the impact of contextual factors on advertising effectiveness. They fail to differentiate between the effects of the context itself and other factors like ad creatives and delivery strategies, leading to biased evaluation results. Furthermore, existing evaluation methods lack standardized context classification and tagging systems, resulting in inconsistent evaluation metrics across different scenarios and hindering unified evaluation and comparison across multiple scenarios. Data collection and preprocessing are not standardized enough, with invalid and missing data affecting evaluation accuracy. Moreover, the evaluation results lack targeted optimization guidance, failing to provide effective support for subsequent adjustments to contextualized delivery strategies. Summary of the Invention

[0004] To address the aforementioned problems, this invention proposes a method for accurately evaluating advertising effectiveness based on scenario-based ad placement. This method involves constructing a standardized scenario tagging system, standardizing data collection and preprocessing, and achieving precise matching between scenarios and ad placement data.

[0005] To address the aforementioned technical problems, the present invention proposes the following technical solution: a method for accurately evaluating advertising effectiveness based on contextualized ad placement, comprising the following steps: Step 1: Build a scenario-based tag system covering all user touchpoints, collect user data, environmental data, content data and advertising behavior data throughout the entire advertising process, divide the tag categories according to the scenario dimensions triggered by the advertising, set multi-level tag classification rules for each tag category, and form a standardized and reusable scenario-based tag system. Step 2: Synchronously collect relevant data from the entire advertising campaign. For the entire campaign of a single ad, synchronously collect scenario data, exposure data, click data, conversion data, and subsequent user behavior data during ad campaigns. Perform standardized preprocessing on all collected data, remove invalid data, fill in missing data, and remove duplicate data to obtain a unified and standardized dataset to be analyzed. Step 3: Complete the accurate matching of ad delivery data and contextual tags. For each ad delivery record in the dataset to be analyzed, match the tag combination in the corresponding contextual tag system to determine the target delivery scenario and the actual reach scenario of the ad delivery. Calculate the matching degree between the target scenario and the actual reach scenario to form the scenario matching result for each ad. Step 4: Decompose the advertising performance metrics for different scenarios. According to the user conversion path of the advertising campaign, the advertising performance is divided into multiple progressive levels of performance metrics. Combining the attribute characteristics of different scenarios, a benchmark reference value is set for each level of performance metrics under the corresponding scenario, and the normal fluctuation range and abnormal performance range of the metrics under different scenarios are distinguished. Step 5: Conduct scenario-based advertising performance attribution analysis, isolate the interference of non-scenario factors on advertising performance, calculate the contribution weight of different scenario tags and tag combinations to advertising performance, distinguish the performance changes brought about by the scenario itself and the advertising material delivery strategy, and identify the core scenario factors affecting advertising performance. Step 6: Generate multi-dimensional accurate evaluation results of advertising effectiveness. Combine the effect indicator data of the scene matching results and the contribution weight obtained from the attribution analysis to generate a comprehensive score of advertising effectiveness under different scene dimensions. According to the comprehensive score, the advertising effectiveness of different scenes is classified and the corresponding advertising effectiveness evaluation conclusions and optimization directions are output.

[0006] Preferably, when constructing a scenario-based tagging system, the tag categories include time scenario, space scenario, user behavior scenario, content context scenario, and device environment scenario. Each tag category has a three-level tag classification rule: the first-level tag is the scenario category, the second-level tag is the scenario sub-type, and the third-level tag is the scenario specific feature, so as to achieve full-dimensional coverage of the advertising scenario.

[0007] Preferably, when performing standardized preprocessing on the collected data, all data are first formatted according to a unified field specification, and then abnormal, invalid, and duplicate data are identified through data validation rules. The identified abnormal and invalid data are removed, and missing core field data is filled in by the mean of the same scenario and dimension to ensure the completeness and accuracy of the dataset to be analyzed.

[0008] Preferably, when completing the accurate matching of advertising data and contextual tags, the contextual features of a single advertising record are first extracted, and then the extracted contextual features are matched one by one with the tags in the contextual tag system to determine the full tag combination corresponding to the advertisement. At the same time, the target contextual tags preset before the advertisement are placed are compared to calculate the overlap between the actual reached contextual ...

[0009] Preferably, when performing scenario-adaptive advertising performance metrics, the advertising performance is broken down into exposure, click, interaction, conversion, and retention metrics according to the user conversion path. Each level of metrics is assigned a benchmark reference value for the corresponding scenario. The benchmark reference value is calculated using historical campaign data from the same industry, product category, and scenario.

[0010] Preferably, when conducting scenario-based advertising performance attribution analysis, first fix the advertising material placement budget, placement duration, and non-scenario variables of the target audience, then compare the differences in advertising performance data under different scenarios, calculate the marginal contribution of different scenario tags to advertising performance, and at the same time eliminate the effect interference caused by the adjustment of placement strategy to accurately locate the actual impact of scenario factors on advertising performance.

[0011] Preferably, when generating accurate multi-dimensional advertising performance evaluation results, the advertising data is first grouped according to the scenario dimension, and then the performance indicators under each scenario are weighted to obtain a comprehensive score. According to the comprehensive score, the scenario performance is divided into four levels: excellent, up to standard, needing optimization, and ineffective. At the same time, corresponding evaluation conclusions and actionable optimization directions are output for different levels of scenarios.

[0012] Preferably, after outputting the advertising effectiveness evaluation conclusion, a comparative verification will be conducted for high-quality scenarios and scenarios that need optimization. By controlling a single variable, the impact of scenario factors on advertising effectiveness will be verified. At the same time, the benchmark values ​​and attribution weights of the scenario-based tag system indicators will be corrected to improve the accuracy of subsequent evaluations.

[0013] Preferably, new advertising placement data, scenario data, and performance data are continuously collected, the scenario-based tagging system is updated and iterated regularly, the benchmark reference values ​​of performance indicators are dynamically adjusted, and the contribution weights of attribution analysis are optimized and corrected to ensure that the evaluation method is adapted to the ever-changing placement environment and user behavior characteristics, and to maintain the accuracy of the evaluation.

[0014] The advantages of this invention compared to the prior art are: A standardized and reusable scenario-based tagging system has been constructed, covering all user touchpoint scenarios. Through a three-level tag grading rule, the system achieves accurate segmentation of scenarios across all dimensions, solving the problems of chaotic scenario classification and lack of unified standards in existing evaluation methods, and providing a solid foundation for scenario-based evaluation.

[0015] The system standardizes the data collection and preprocessing process across the entire advertising delivery chain. By unifying the format, removing invalid data, and completing missing data, it ensures the integrity and accuracy of the dataset to be analyzed. At the same time, it achieves precise matching between advertising delivery data and scene tags. Combined with layered performance indicators and scene-based attribution analysis, it accurately locates the impact of scene factors on advertising performance, significantly improving the accuracy of advertising performance evaluation.

[0016] A comparative verification and dynamic iteration mechanism was established. The evaluation parameters were corrected through comparative verification, and the label system, benchmark value and attribution weight were dynamically adjusted through continuous data collection. This ensures that the evaluation method can adapt to the ever-changing delivery environment and user behavior characteristics, and maintain the accuracy and practicality of the evaluation in the long term. Detailed Implementation

[0017] The present invention will now be described in further detail.

[0018] Example 1 This embodiment uses contextualized advertising on short videos on the internet as an example to illustrate the method of the present invention in detail. The specific steps are as follows: Step 1: Construct a scenario-based tagging system covering all user touchpoints. Collect various data throughout the entire advertising campaign process. This includes user data such as age, gender, interests, and consumption habits; environmental data such as device model, network type, and geographic location; content data such as ad creative type, content theme, and presentation format; and campaign behavior data such as campaign time, campaign channel, and campaign budget. Based on the scenario triggering the campaign, the tag categories are divided into five main categories: time scenario, space scenario, user behavior scenario, content context scenario, and device environment scenario. Each tag category has a three-level hierarchical rule: the first-level tag is the broad scenario category, the second-level tag is the sub-scenario type, and the third-level tag is the specific scenario characteristic. This forms a standardized and reusable scenario-based tagging system that can be flexibly called upon and expanded according to different campaign needs.

[0019] Step Two: Synchronously collect and preprocess relevant data from the entire advertising campaign. For the entire campaign of a single beauty-related short video ad, synchronously collect scenario data, such as the campaign time being a weekday evening rush hour and the geographical location being a home environment in a first-tier city; exposure data, such as the number of exposures and the number of people exposed; click data, such as the number of clicks and the click-through rate; conversion data, such as the number of orders and the conversion rate; and subsequent user behavior data, such as the number of repeat purchases and the retention days. Standardize and preprocess the collected data. First, unify all data formats according to a unified field specification, for example, unify time data to year-month-day-hour format and rate values ​​to percentage format. Then, identify abnormal data through data validation rules, such as abnormal records with a click-through rate exceeding 100%; invalid data, such as blank exposure data; and duplicate data, such as repeated clicks by the same user on the same ad. Remove abnormal and invalid data. For missing core field data, such as missing geographical location data, supplement it using the average of the same scenario and dimension, such as the average geographical location of other campaign records from the same weekday evening rush hour, to finally obtain a unified and standardized dataset for analysis.

[0020] Step 3: Complete the precise matching of ad delivery data with contextual tags. For each ad delivery record in the dataset to be analyzed, extract its contextual features, such as the delivery time being a weekend lunch break, the geographical location being a second-tier city office setting, and the user interest being beauty. Match the extracted contextual features one by one with the tags in the contextual tag system to determine the full tag combination corresponding to the ad: time context being a weekend lunch break, space context being a second-tier city office setting, and user behavior context being beauty interest. At the same time, compare the target contextual tags preset before the ad delivery, such as the target context being a weekend lunch break, a first-tier city home setting, and beauty interest, to calculate the overlap between the actual reach context and the target context. In this embodiment, the overlap is 60%, forming the contextual matching result corresponding to the ad and identifying the links where the contextual adaptation is insufficient.

[0021] Step four involves breaking down advertising performance metrics into layers tailored to different scenarios. Following the user conversion path, advertising performance is divided into five progressively higher levels of metrics: exposure, click, interaction, conversion, and retention. Exposure metrics include impressions and number of impressions; click metrics include clicks and click-through rate (CTR); interaction metrics include comments, likes, and shares; conversion metrics include orders, conversion rate, and average order value; and retention metrics include 7-day retention rate and 30-day retention rate. Combining the characteristics of different scenarios, and using historical campaign data from the same industry, product category, and scenario, a baseline reference value for each metric level is calculated for the corresponding scenario. For example, the baseline reference value for CTR in a weekday morning rush hour office scenario is 3.5%, with a normal fluctuation range of 3.0% to 4.0%. Values ​​exceeding 4.0% or falling below 3.0% are considered abnormal, providing a reference standard for subsequent performance evaluation.

[0022] Step 5: Conduct scenario-based attribution analysis of advertising performance. Fix non-scenario variables such as ad creatives, budget, campaign duration, and target audience. For example, fix the ad creative as a beauty short video, the budget as 50,000 yuan, the campaign duration as seven days, and the target audience as women aged 20-35. Compare the differences in advertising performance data under different scenarios. For example, compare the click-through rate and conversion rate of a weekday morning rush hour office scenario versus a weekend lunchtime home scenario. Calculate the marginal contribution of different scenario tags to advertising performance. For example, the marginal contribution of the home scenario to the conversion rate is 2.8%. Simultaneously, exclude the interference caused by adjustments to the campaign strategy, such as changing the campaign channel midway, to accurately pinpoint the actual impact of scenario factors on advertising performance. For example, if the conversion rate is found to be significantly higher in the home scenario than in the office scenario, the core reason is that users have more time to understand the ad content and complete the conversion in the home scenario.

[0023] Step Six: Generate a multi-dimensional, precise evaluation result of the advertising performance. This involves combining scene matching results (60% overlap), layered performance metrics such as click-through rate (CTR) of 3.8% and conversion rate of 5.2%, and contribution weights from attribution analysis (35% contribution from the home scene). The overall score is calculated by weighting the ad's performance across different scenes; in this example, the overall score is 82. Based on the overall score, the scene placement performance is categorized into four levels: excellent, satisfactory, needs optimization, and ineffective. The ad is rated as satisfactory overall. The evaluation conclusion is also output: scene matching is average; performance in the home scene is excellent; performance in the office scene does not meet the benchmark. Optimization directions are suggested: increase the placement ratio in the weekend lunchtime home scene and optimize the presentation format of the ad creative in the office scene.

[0024] After outputting the evaluation conclusions, comparative verification was conducted for the high-performing scenario (home scenario) and the scenario requiring optimization (office scenario). Keeping other variables such as ad creatives and budget constant, only the scenario was adjusted. Equal-scale campaigns were conducted in both the home and office scenarios to verify the impact of scenario factors on ad performance. The results showed that the conversion rate in the home scenario was 3.1% higher than in the office scenario, consistent with the attribution analysis. Based on these verification results, the scenario-based tagging system was revised, adding sub-tags for the home scenario, such as "home leisure" and "home skincare." The benchmark click-through rate for the office scenario was adjusted down to 3.2%. The attribution weights were optimized, with the contribution weight of the home scenario adjusted to 38%, improving the accuracy of subsequent evaluations.

[0025] We continuously collect subsequent ad placement data, new scenario data (such as newly added nighttime home scenarios), and performance data. We update and iterate the scenario-based tagging system monthly, dynamically adjust the benchmark reference values ​​of performance indicators quarterly, and optimize and correct the contribution weight of attribution analysis every six months to ensure that the evaluation method can adapt to changes in the internet short video advertising environment and changes in user behavior characteristics, and continuously maintain the accuracy of the evaluation.

[0026] Example 2 This embodiment uses offline physical store scenario-based advertising as an example to illustrate the method of the present invention in detail. The specific steps are as follows: Step 1: Construct a scenario-based tagging system covering all user touchpoints. Collect various data throughout the entire advertising campaign process. User data includes age, gender, spending power, and shopping habits of in-store users; environmental data includes the type of surrounding business district, peak hours, and weather conditions; content data includes ad poster format, promotional content, and display location; and campaign behavior data includes ad placement time, placement area, and material type. Based on the scenario triggering the campaign, tag categories are divided into five major categories: time scenario, space scenario, user behavior scenario, content context scenario, and device environment scenario. Each tag category has a three-level hierarchical rule: the first-level tag is the major scenario category, the second-level tag is the sub-scenario type, and the third-level tag is the specific scenario characteristic, forming a standardized and reusable scenario-based tagging system that can be flexibly called upon and expanded according to the different campaign needs of different stores.

[0027] Step two involves synchronously collecting and preprocessing relevant data across the entire advertising campaign. For a single offline apparel store advertisement, the following scenario data is collected synchronously: campaign time (weekend afternoon), surrounding commercial area (community commercial area), medium to high foot traffic, and sunny weather; exposure data (number of people reached by the poster, number of people visiting the store); click data (number of people inquiring about the poster); interaction data (number of people trying on clothes); conversion data (number of purchases, transaction amount); and subsequent user behavior data (repurchase rate, number of new members). The collected data undergoes standardized preprocessing. First, all data formats are unified according to standardized field specifications, such as unifying time data to year-month-day-hour format and monetary data to yuan format. Then, abnormal data is identified through data validation rules, such as records with negative transaction amounts; invalid data, such as blank reach data; and duplicate data, such as records of the same user visiting the store repeatedly during the same time period. Abnormal and invalid data are removed. For missing core field data, such as missing weather data, the missing data is filled in by the average value of the same scenario and dimension, such as the average weather value of other advertising records with high traffic in the community business district on a weekend afternoon. Finally, a unified and standardized dataset to be analyzed is obtained.

[0028] Step 3: Complete the precise matching of ad placement data with contextual tags. For each ad placement record in the dataset to be analyzed, extract its contextual features, such as the placement time being a weekday morning, the surrounding business district being an office building business district, the pedestrian traffic being moderate, and the weather being cloudy. Match the extracted contextual features one by one with the tags in the contextual tag system to determine the full tag combination corresponding to the ad. The time context is a weekday morning, the space context is an office building business district with moderate traffic, and the user behavior context is working women with moderate spending power. At the same time, compare it with the target context tags preset before the ad placement, such as the target context being a weekend afternoon with high traffic in a community business district and high spending power of working women. Calculate the overlap between the actual reach context and the target context. In this embodiment, the overlap is 50%, forming the contextual matching result corresponding to the ad and identifying the links where the contextual adaptation is insufficient.

[0029] Step four involves breaking down advertising performance metrics into layers tailored to different scenarios. Following the user conversion path, advertising performance is divided into five progressively higher levels of metrics: exposure, clicks, interaction, conversion, and retention. Exposure metrics include the number of people reached by the poster and the number of people visiting the store; click metrics include the number of people inquiring about the poster; interaction metrics include the number of people trying on clothes; conversion metrics include the number of transactions, transaction amount, and average order value; and retention metrics include the seven-day repurchase rate and member retention rate. Combining the characteristics of different scenarios, and using historical campaign data from the same industry, product category, and scenario, a baseline reference value for each level of metric is calculated for the corresponding scenario. For example, the baseline reference value for transaction amount in a high-traffic community business district on a weekend afternoon is 80,000 yuan, with a normal fluctuation range of 70,000 to 90,000 yuan. Values ​​exceeding 90,000 yuan or falling below 70,000 yuan are considered abnormal performance ranges, providing a reference standard for subsequent performance evaluation.

[0030] Step 5: Conduct scenario-based attribution analysis of advertising effectiveness. Fix non-scenario variables such as ad creatives, budget, duration, and material type. For example, fix the ad creative as a clothing promotional poster, the budget as 20,000 yuan, the duration as seven days, and the material type as a storefront banner. Compare the differences in advertising effectiveness data under different scenarios. For example, compare the transaction amount and number of people trying on clothes between high-traffic community shopping districts on weekend afternoons and mid-traffic office building shopping districts on weekday mornings. Calculate the marginal contribution of different scenario tags to advertising effectiveness. For example, the marginal contribution of high-traffic community shopping districts to transaction amount is 3.5%. Simultaneously, exclude the interference caused by adjustments to the campaign strategy, such as changing poster content midway, to accurately pinpoint the actual impact of scenario factors on advertising effectiveness. For example, it was found that the transaction amount under high-traffic community shopping districts on weekend afternoons was significantly higher than that under mid-traffic office building shopping districts on weekday mornings. The core reason is that the target user group is more precise under high-traffic community shopping districts, and users have more time to understand the promotional content and complete the conversion.

[0031] Step Six: Generate a multi-dimensional, precise evaluation result of the advertising effect. This involves combining scene matching results (50% overlap), layered performance metrics such as transaction amount of 75,000 yuan and number of trial users (120), and contribution weights obtained from attribution analysis (40% contribution from high-traffic areas in community business districts). The overall performance of the advertisement in different scenarios is weighted and calculated to obtain a comprehensive score, which in this example is 78 points. Based on the comprehensive score, the scene placement effect is divided into four levels: excellent, satisfactory, needs optimization, and ineffective. The advertisement is rated as satisfactory overall. The evaluation conclusion is also output: scene matching is average; the effect is excellent in high-traffic areas of community business districts on weekend afternoons; the effect in mid-traffic areas of office buildings on weekday mornings does not meet the benchmark. The optimization direction is suggested: increase the proportion of advertising in high-traffic areas of community business districts on weekend afternoons and optimize advertising on weekday mornings.

[0032] The above description is only a preferred embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any equivalent substitutions or modifications made by those skilled in the art within the scope of the technology disclosed in the present invention, based on the technical solution and inventive concept of the present invention, should be covered within the scope of protection of the present invention.

Claims

1. A method for accurately evaluating advertising effectiveness based on contextualized ad placement, characterized in that, Includes the following steps: Step 1: Build a scenario-based tag system covering all user touchpoints, collect user data, environmental data, content data and advertising behavior data throughout the entire advertising process, divide the tag categories according to the scenario dimensions triggered by the advertising, set multi-level tag classification rules for each tag category, and form a standardized and reusable scenario-based tag system. Step 2: Synchronously collect relevant data from the entire advertising campaign. For the entire campaign of a single ad, synchronously collect scenario data, exposure data, click data, conversion data, and subsequent user behavior data during ad campaigns. Perform standardized preprocessing on all collected data, remove invalid data, fill in missing data, and remove duplicate data to obtain a unified and standardized dataset to be analyzed. Step 3: Complete the accurate matching of ad delivery data and contextual tags. For each ad delivery record in the dataset to be analyzed, match the tag combination in the corresponding contextual tag system to determine the target delivery scenario and the actual reach scenario of the ad delivery. Calculate the matching degree between the target scenario and the actual reach scenario to form the scenario matching result for each ad. Step 4: Decompose the advertising performance metrics for different scenarios. According to the user conversion path of the advertising campaign, the advertising performance is divided into multiple progressive levels of performance metrics. Combining the attribute characteristics of different scenarios, a benchmark reference value is set for each level of performance metrics under the corresponding scenario, and the normal fluctuation range and abnormal performance range of the metrics under different scenarios are distinguished. Step 5: Conduct scenario-based advertising performance attribution analysis, isolate the interference of non-scenario factors on advertising performance, calculate the contribution weight of different scenario tags and tag combinations to advertising performance, distinguish the performance changes brought about by the scenario itself and the advertising material delivery strategy, and identify the core scenario factors affecting advertising performance. Step 6: Generate multi-dimensional accurate evaluation results of advertising effectiveness. Combine the effect indicator data of the scene matching results and the contribution weight obtained from the attribution analysis to generate a comprehensive score of advertising effectiveness under different scene dimensions. According to the comprehensive score, the advertising effectiveness of different scenes is classified and the corresponding advertising effectiveness evaluation conclusions and optimization directions are output.

2. The method for accurately evaluating advertising effectiveness based on contextualized ad placement according to claim 1, characterized in that: When constructing a scenario-based tagging system, the tag categories include time scenario, space scenario, user behavior scenario, content context scenario, and device environment scenario. Each tag category has a three-level tag classification rule: the first-level tag is the scenario category, the second-level tag is the scenario sub-type, and the third-level tag is the scenario specific feature, so as to achieve full-dimensional coverage of the advertising scenarios.

3. The method for accurately evaluating advertising effectiveness based on contextualized ad placement according to claim 1, characterized in that: When performing standardized preprocessing on the collected data, all data are first formatted according to a unified field specification. Then, abnormal, invalid, and duplicate data are identified through data validation rules. The identified abnormal and invalid data are removed, and missing core field data is filled in by the mean of the same scenario and dimension to ensure the completeness and accuracy of the dataset to be analyzed.

4. The method for accurately evaluating advertising effectiveness based on contextualized ad placement according to claim 1, characterized in that: When accurately matching ad delivery data with contextual tags, the scenario features of a single ad delivery record are first extracted. Then, the extracted scenario features are matched one by one with the tags in the contextual tag system to determine the full tag combination corresponding to the ad. At the same time, the target scenario tags preset before ad delivery are compared to calculate the overlap between the actual reached scenario and the target scenario, thus quantifying the degree of scenario matching.

5. The method for accurately evaluating advertising effectiveness based on contextualized ad placement according to claim 1, characterized in that: When performing scenario-adaptive ad performance metrics, ad performance is broken down into exposure, click, interaction, conversion, and retention metrics according to the user conversion path. Each level of metrics is assigned a benchmark reference value for the corresponding scenario. The benchmark reference value is calculated using historical campaign data from the same industry, product category, and scenario.

6. The method for accurately evaluating advertising effectiveness based on contextualized ad placement according to claim 1, characterized in that: When conducting scenario-based advertising performance attribution analysis, first fix the advertising creative budget, campaign duration, and non-scenario variables of the target audience. Then compare the differences in advertising performance data under different scenarios, calculate the marginal contribution of different scenario tags to advertising performance, and at the same time eliminate the effect interference caused by the adjustment of the campaign strategy to accurately locate the actual impact of scenario factors on advertising performance.

7. The method for accurately evaluating advertising effectiveness based on contextualized ad placement according to claim 1, characterized in that: When generating accurate multi-dimensional advertising performance evaluation results, the advertising data is first grouped according to the scenario dimension. Then, the performance indicators under each scenario are weighted and calculated to obtain a comprehensive score. Based on the comprehensive score, the scenario performance is divided into four levels: excellent, up to standard, needing optimization, and ineffective. At the same time, corresponding evaluation conclusions and actionable optimization directions are output for different levels of scenarios.

8. The method for accurately evaluating advertising effectiveness based on contextualized ad placement according to claim 1, characterized in that: After outputting the advertising performance evaluation results, we will also conduct comparative testing on high-quality scenarios and scenarios that need optimization. By controlling for single variables, we will verify the degree of influence of scenario factors on advertising performance. At the same time, we will revise the benchmark values ​​and attribution weights of the scenario-based tag system indicators to improve the accuracy of subsequent evaluations.

9. The method for accurately evaluating advertising effectiveness based on contextualized ad placement according to claim 1, characterized in that: We continuously collect new advertising placement data, scenario data, and performance data; regularly update and iterate the scenario-based tagging system; dynamically adjust the benchmark reference values ​​of performance indicators; and optimize and correct the contribution weights of attribution analysis to ensure that the evaluation method adapts to the ever-changing placement environment and user behavior characteristics, thereby maintaining the accuracy of the evaluation.