Live SSAI Cohort Caching for Scalable Targeted Ad Insertion
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing digital advertising systems face challenges in managing high concurrency of live streaming events, particularly in over-the-top (OTT) media services, where instantaneous advertisement insertion is required, leading to bottlenecks that cannot be resolved by increasing infrastructure, and integration with content customization exacerbates these challenges.
Innovation Solution
Implementing a scalable Dynamic Ad Insertion (DAI) system that groups viewers into cohorts based on shared user characteristics, allowing for cached content delivery and reduced demand on the technical infrastructure, while maintaining targeted advertising capabilities.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If traditional digital advertising systems deliver individualized ads to each viewer in live streaming events, then ad personalization is achieved, but system bottleneck and infrastructure complexity increase significantly
Solution Approach 1:
The patent segments the audience into cohorts based on shared characteristics (demographics, interests, behavior patterns) rather than treating each viewer individually. This segmentation allows the system to deliver personalized advertising experiences to groups simultaneously, reducing the complexity burden while maintaining adaptability. The cohort-based approach divides the vast audience into manageable segments that can be processed efficiently during live events.
Solution Approach 2:
The patent merges multiple individual viewer profiles into cohort-level aggregations that preserve key personalization attributes. By combining data from multiple users into cohort representations, the system maintains the ability to deliver targeted ads while significantly reducing the computational burden. This merging strategy allows infrastructure to handle fewer, consolidated requests rather than millions of individual personalized ad deliveries.
2Speed
If traditional systems process individual ad requests for each viewer simultaneously, then real-time ad insertion is achieved, but processing capacity requirements become unsustainable
Solution Approach 1:
The patent performs preliminary actions by pre-segmenting audiences into cohorts before live events begin and pre-determining cohort-based ad selections. This advance preparation allows the system to handle real-time ad insertion demands during live events without processing individual viewer requests on the fly. The cohort assignments and ad selections are established beforehand, enabling rapid deployment during the actual event while maintaining real-time performance.
3Adaptability or versatility
If infrastructure capacity is increased to handle individualized ad delivery at scale, then ad personalization capability is maintained, but cost and complexity increase beyond sustainable levels
Solution Approach 1:
The patent creates cohort-level copies or representations that capture the essential personalization characteristics of multiple individual users. Instead of processing each user's unique profile, the system uses cohort copies that represent groups of users with similar attributes. This copying approach maintains personalization capability while dramatically reducing the infrastructure capacity needed, as one cohort representation serves multiple users simultaneously.
Data Source
AI summary
Server-side ad insertion (SSAI) systems for use with live broadband video streams. Users of a content delivery network (CDN) are grouped into a plurality of cohorts, which are groups of users having a matching set of user characteristics. When a video ecosystem receives a request to stream a live broadband video stream to a user, the cohort for that user is identified. Then, an advertisement to be shown to all members of that cohort viewing the live broadband video stream is identified and incorporated into the live broadband video stream at the video ecosystem to create a cohort-targeted live video stream for the user. The video ecosystem may then transmit the cohort-targeted live broadband video stream to a user device associated with the user. Content related to advertisement identification and insertion may be cached to expedite target advertisement processing for subsequent viewers in the same cohort.