Normalized Click-Through Rate Ranking for Search Advertisements
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Search engines face challenges in maximizing revenue from pay-per-click listings as the likelihood of user selection varies among listings, regardless of their position in search results, and existing methods do not effectively account for this variability to prioritize more likely selections.
Innovation Solution
The implementation of a system that calculates and utilizes normalized click-through rates (CTR) to reorder pay-per-click listings based on their likelihood of being selected, ensuring that listings with higher normalized CTRs are placed in more prominent positions, thereby increasing the chances of user selection and generating additional revenue.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If listings are placed in higher positions in search results, then user selection likelihood is improved, but not all listings have equal opportunity to be selected since some listings are inherently more likely to be selected than others
Solution Approach 1:
The patent changes the parameter used for listing placement from simple position-based ranking to normalized click-through rate (CTR) based ranking. By calculating and normalizing CTR for each listing, the system adjusts placement decisions based on actual user behavior patterns, moving from theoretical position-based selection to data-driven probability-based selection.
Solution Approach 2:
The system implements feedback by using actual user click data to calculate and normalize CTR for each listing. This feedback loop allows the system to learn from real user behavior patterns and continuously improve placement decisions, adjusting the ranking based on measured selection probabilities rather than static position assumptions.
2Productivity
If search engines provide listings based solely on bid amount, then implementation is simple, but revenue is not maximized because listings with higher selection likelihood are not necessarily prioritized
Solution Approach 1:
The system uses feedback from actual user click data to calculate normalized CTR for each listing. This feedback mechanism enables the system to learn real user preferences and adjust placement decisions accordingly, maximizing revenue by prioritizing listings that users actually select most frequently rather than relying solely on bid amounts.
Solution Approach 2:
The patent replaces the simple mechanical bid-based ranking system with a more complex but effective statistical model that uses normalized CTR calculations. This substitution allows the system to account for inherent differences in listing selection probability, achieving better revenue outcomes through data-driven decision making rather than straightforward bid bidding.
3Productivity
If all listings are treated equally regardless of their inherent selection characteristics, then fairness is maintained, but the search engine misses opportunities to optimize for listings that are more likely to be selected
Solution Approach 1:
The system changes the placement parameter from uniform treatment to differentiated treatment based on normalized CTR. By calculating and applying individual CTR values to each listing, the system achieves optimized revenue placement while maintaining operational simplicity through automated calculations rather than manual intervention.
Data Source
AI summary
A set of preliminary search results is retrieved based on a search query, the set of search results including a plurality of pay-per-click listings. A normalized click-through rate is calculated for each pay-per-click listing in the set of preliminary search results, the normalized click-through rate being based on an actual click-through rate that is adjusted according to one or more positions in which the pay-per-click listing was previously provided in one or more sets of final search results. The plurality of pay-per-click listings is ordered based on the calculated normalized click-through rates.