Ad Quality Estimation Using Session Behavior Segmentation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing on-line advertising systems rely on imperfect click-through rate (CTR) metrics to assess advertisement quality, as they focus solely on the advertisement creative rather than the landing document, leading to inaccurate evaluations.
Innovation Solution
A method that utilizes observed user behavior beyond CTR, such as session features like ad selection duration, number of previous ad selections, and search queries, to construct a statistical model that estimates advertisement quality, allowing for more accurate assessments by correlating known qualities with measurable user actions.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of operation
If click-through rate (CTR) is used to measure advertisement quality, then the measurement process is simple and easy to implement, but the measurement precision is insufficient because it does not account for user satisfaction with the landing document
Solution Approach 1:
The patent segments the user interaction process into multiple distinct phases: ad impression, ad click, landing page view, and user action completion. By dividing the measurement into these segments, the system can evaluate different aspects of user behavior (click rate vs. completion rate) to obtain a more comprehensive and precise quality assessment that goes beyond simple CTR.
Solution Approach 2:
The patent introduces a new dimension to ad quality measurement by moving from binary click/no-click data to multi-dimensional user behavior data including session duration, pages viewed, and action completion. This dimensional expansion allows the system to capture user satisfaction nuances that CTR alone cannot detect, thereby improving measurement precision while maintaining operational feasibility through automated tracking.
2Measurement precision
If multiple user behavior parameters are collected and analyzed to improve ad quality measurement, then the measurement precision improves, but the device complexity increases due to the need for statistical models and data processing systems
Solution Approach 1:
The patent implements feedback mechanisms by continuously monitoring user actions on landing pages and using this data to update ad quality estimates. The system feeds back user behavior patterns (session duration, pages viewed, conversions) into the quality assessment model, allowing dynamic refinement of ad rankings. This feedback loop improves precision by adapting to real user patterns while managing complexity through automated iterative processing.
Solution Approach 2:
The patent replaces manual quality assessment with automated statistical modeling and machine learning algorithms. Instead of requiring complex manual analysis of user behavior, the system uses computational models to process and interpret user action data, automatically generating quality scores. This substitution reduces operational complexity while enhancing measurement precision through sophisticated data-driven approaches.
Data Source
AI summary
A system obtains ratings associated with a first set of advertisements hosted by one or more servers, where the ratings indicate a quality of the first set of advertisements. The system observes multiple different first user actions associated with user selection of advertisements of the first set of advertisements and derives a statistical model using the observed first user actions and the obtained ratings. The system further observes second user actions associated with user selection of a second advertisement hosted by the one or more servers and uses the statistical model and the second user actions to estimate a quality of the second advertisement.