Cross-Platform Audience Measurement With AI Deduplication
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for tracking media viewership, particularly across multiple platforms, face challenges such as over-counting and under-counting due to server log manipulation and the limitations of third-party cookies, leading to inaccurate audience metrics and duplicate impressions.
Innovation Solution
Implementing a system that utilizes machine learning models, specifically random forest regression, to deduplicate audience data across platforms by leveraging first-party cookies and integrating cross-platform correction circuitry to adjust audience sizes based on multiple levels of aggregation, using AI to process input data and generate accurate total audience metrics.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If server log tracking is used to monitor media viewership, then viewership information can be obtained, but server log manipulation causes over-counting and under-counting leading to inaccurate audience metrics
Solution Approach 1:
The patent introduces a measurement service provider as an intermediary between content providers and audience tracking. This intermediary uses panel data from selected users and machine learning models to generate accurate audience metrics without relying on manipulatable server logs, thus resolving the contradiction between measurement precision and server log reliability
Solution Approach 2:
The patent replaces the mechanical server log tracking system with an AI-based measurement system that uses machine learning models (such as random forest regression) to estimate audience sizes. This substitution eliminates the vulnerability to log manipulation while maintaining the ability to track viewership across multiple platforms
2Adaptability or versatility
If third-party cookies are used for cross-platform tracking, then audience data can be collected across platforms, but cookie limitations and manipulation lead to duplicate impressions and inaccurate metrics
Solution Approach 1:
The measurement service provider acts as an intermediary that receives panel data from multiple platforms and uses machine learning to deduplicate audience members. This approach maintains cross-platform tracking capability while eliminating duplicate impressions through AI-based identification of unique audience members across TV, digital, and other platforms
Solution Approach 2:
The patent changes the fundamental parameter of audience identification from cookie-based tracking to panel member-based tracking with machine learning deduplication. By changing how audience members are identified and counted across platforms, the system maintains versatility while improving measurement precision
3Quantity of substance
If audience data is aggregated across multiple platforms without correction, then total reach can be estimated, but over-counting occurs due to duplicate audience members viewed on multiple platforms
Solution Approach 1:
The patent replaces simple aggregation of audience data across platforms with machine learning-based estimation using random forest regression models. The system takes panel data from multiple platforms as input and outputs corrected total audience sizes that account for duplicate viewing, thus resolving the contradiction between measuring total reach and counting unique audiences
Solution Approach 2:
The system uses feedback from panel member identification and cross-platform correlation analysis to continuously improve audience size estimates. By comparing panel data across platforms and using machine learning models to identify patterns, the system provides feedback loops that refine measurements and eliminate over-counting
Data Source
AI summary
Methods, apparatus, systems, and articles of manufacture are disclosed to determine total audience ratings. An example apparatus includes metrics generator circuitry to generate a first audience size for media accessed by first devices of a first media platform at a first level of aggregation, the first level of aggregation corresponding to the media accessed on a first television network and on a first website, generate a second audience size for the media accessed by the first devices of the first media platform at a second level of aggregation, the second level of aggregation corresponding to the media accessed on the first television network and accessed on the first website and a second website, comparator circuitry to compare the first audience size to the second audience size, adjustor circuitry to reduce the first audience size based on the second audience size, and audience determination circuitry to determine a total audience size.