Ad Optimization System Parallel Bidding Latency Reduction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Publishers face challenges in optimizing advertisement yield and reducing advertisement latency in real-time bidding environments due to infrastructure limitations, technical inefficiencies, and increasing complexity from changes in major platforms like Google and Meta.
Innovation Solution
A system that integrates client-side and server-side optimizations, allowing parallel bid requests to header bidding partners and mediation platforms, caching the highest bid and ad object, and submitting them in a single hop to reduce latency and maximize ad yield.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If parallel bid requests are sent to multiple header bidding partners and mediation platforms, then ad yield is maximized through increased competition, but system complexity and processing overhead increase
Solution Approach 1:
The system segments the ad bidding process into distinct modules: header bidding partners, mediation platforms, and a unified arbitration layer. Each component handles specific demand sources independently, allowing parallel processing while maintaining manageable complexity through clear separation of concerns.
Solution Approach 2:
The patent introduces a mediation platform as an intermediary that receives bids from multiple header bidding partners and ad networks, then arbitrates to select the winning bid. This mediator layer simplifies the overall system architecture by centralizing the decision-making logic and reducing the complexity that would otherwise exist in coordinating multiple parallel bidding systems.
2Device complexity
If sequential waterfall architecture is used to query demand sources, then system complexity is reduced, but ad delivery latency increases significantly
Solution Approach 1:
The system performs preliminary actions by sending bid requests to multiple header bidding partners and mediation platforms simultaneously before the ad delivery deadline. This parallel preliminary bidding allows the system to gather multiple bids in advance, then quickly arbitrate to select the winner, thereby reducing overall delivery latency while maintaining manageable complexity.
3Productivity
If multiple network hops are performed between mediation platforms, header bidding partners, and ad networks, then comprehensive bid collection is achieved, but total processing time increases
Solution Approach 1:
The patent merges multiple bid collection pathways into a unified arbitration process. Instead of treating header bidding partners and mediation platforms as separate sequential steps, the system combines their bid responses and arbitrates among them simultaneously, reducing the number of effective network hops and processing time while maintaining comprehensive bid collection.
4Device complexity
If publishers rely on third-party mediation platforms to manage demand sources, then integration complexity is reduced, but flexibility and control over ad inventory are limited
Solution Approach 1:
The mediation platform is designed with multi-functionality to handle both header bidding partner integrations and direct ad network connections within a single unified system. This universal approach allows publishers to manage diverse demand sources through one platform, reducing integration complexity while maintaining flexibility through configurable demand source priorities and custom arbitration rules.
Data Source
AI summary
A system and a method for optimizing advertisement yield and reducing latency. The system receives an ad request and initiates parallel bid requests to a mediation platform, header bidding partners, and ad networks. The system then determines the highest bid from header bidding partners in real-time while simultaneously determining a prioritized list of demand sources from the mediation platform. The highest bid and its associated ad object are cached and submitted to the mediation platform in a single hop. The mediation platform mediates this bid against other sources. The system selects the ad object, for display on a client device, either from the header bidding partner's ad object if the mediation platform doesn't confirm a higher bid, or the mediation platform's ad object if it does. The system utilizes caching, parallel processing, and backfill strategies to reduce latency and optimize ad selection. It can also dynamically adjust backfill ad network positioning based on real-time data.