Ranking Online Listings Using Multi-Parameter Performance Scores
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Online advertising platforms face challenges in maximizing revenue and providing value to advertisers through effective listing rankings, as existing systems fail to accurately predict user click-through rates and prioritize listings based on relevance and geographic importance.
Innovation Solution
An advertising distribution system calculates a rank score for each listing using a performance score, which is determined by weighted parameters such as search relevancy, expected click-through rate, content density boost, and geography type boost, to optimize listing placement in search results and maximize revenue for providers.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Loss of energy
If listings are ranked solely by advertiser payment amount, then provider revenue is maximized, but listing relevance to users and geographic importance deteriorate
Solution Approach 1:
The patent transforms the single-parameter ranking system (payment amount only) into a multi-parameter system by introducing a performance score that combines multiple factors: expected click-through rate, search relevancy score, content density boost, and geography type boost. This parameter change enables the system to simultaneously consider revenue potential and listing quality, resolving the contradiction between maximizing provider revenue and maintaining listing relevance accuracy.
2Ease of operation
If listings are ranked by expected click-through rate alone, then user benefit is maximized, but provider revenue optimization deteriorates
Solution Approach 1:
The patent merges multiple ranking criteria into a unified performance score calculation. By combining expected click-through rate (user benefit metric) with search relevancy score, content density boost, and geography type boost (provider revenue metrics) using weighted multiplication, the system achieves a balanced ranking that simultaneously optimizes user benefit and provider revenue rather than prioritizing one over the other.
3Device complexity
If simple ranking methods are used, then system complexity is reduced, but ranking accuracy and revenue optimization deteriorate
Solution Approach 1:
The patent segments the ranking process into distinct computational components: calculating expected click-through rate, determining search relevancy score, computing content density boost, and applying geography type boost. Each component is calculated separately using specific formulas and then combined through multiplication to produce the final performance score. This segmentation maintains manageable system complexity while achieving high ranking accuracy through comprehensive factor consideration.
Data Source
AI summary
A search query is received. One or more listings is identified responsive to the search query. For each of the one or more of listings, the following are determined: a relevancy score based on one or more parameters in the search query, an expected click through rate, and at least one of a content density boost that is based on one or more fields that are included in or excluded from the listing and a geography type boost that is based on a comparison of one or more geography parameters of the query to one or more geography parameters of the listing. For each of the one or more listings, a performance score is calculated based on the relevancy score, the expected click through rate, and at least one of the content density boost and the geography type boost.