Conversion Rate Prediction Model for Ad Placement
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current methods for predicting conversion rates of content publisher-third-party content provider pairs are inefficient, as they rely on manual processes and lack data-driven models to optimize ad placement and revenue generation for content publishers.
Innovation Solution
A computer-implemented method and system that analyze log data to identify content publisher-provider pairs, determine conversion rates, and transform feature data into a predictive model to forecast conversion rates, enabling data-driven ad placement and revenue optimization.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If manual processes are used to predict conversion rates, then the system complexity is low, but the prediction accuracy and efficiency are insufficient
Solution Approach 1:
The patent replaces manual processes with an automated machine learning-based system. The processor automatically identifies content publisher-provider pairs from log data, determines conversion rates, extracts feature data, and trains predictive models without human intervention, thereby improving accuracy while managing complexity through automation
Solution Approach 2:
The system transforms raw log data into feature data through parameter extraction and transformation. It converts impression and conversion counts into conversion rate ratios, and further transforms these into model inputs through feature engineering, enabling accurate predictions by changing the representation of data parameters
2Productivity
If data-driven models are implemented to optimize ad placement, then the productivity and revenue generation improve, but the device complexity and computational requirements increase
Solution Approach 1:
The patent segments the prediction process into distinct stages: identifying content publisher-provider pairs, determining conversion rates, extracting feature data, and training predictive models. This segmentation allows the system to handle complexity systematically by breaking down the overall task into manageable sub-tasks that can be processed independently
Solution Approach 2:
The system performs preliminary actions by pre-processing log data to identify and store content publisher-provider pairs, their conversion rates, and feature data before the actual prediction task. This pre-computation and organization of data enables more efficient model training and faster predictions, improving productivity while managing computational complexity
Data Source
AI summary
Systems and methods for predicting a conversion rate of a content publisher-third-party content provider pair are disclosed. A processor identifies, from log data, a plurality of publisher-provider pairs. Each publisher-provider pair corresponds to a content provider associated with at least one creative receiving at least one conversion when served on information resources of the publisher. The processor determines, for each publisher-provider pair, a conversion rate. The processor determines, for each publisher-provider pair, feature data of the publisher and feature data of the content provider. The processor can transform the determined feature data of the publishers and content providers and the conversion rates of the plurality of publisher-provider pairs into a conversion rate prediction model useful for predicting a conversion rate of a given publisher-provider pair different from the identified plurality of publisher-provider pairs.