Privacy-Protected Audience Metrics Using Cloud Data Imputation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing audience measurement technologies face challenges in accurately determining unique audience size and demographics due to the elimination of third-party cookies and tags, limiting the ability to track census-wide media impressions and associate demographic information, especially when multiple devices and users are involved.
Innovation Solution
A privacy-protected cloud environment is used to merge high-quality demographic information from audience measurement entity panels with media impression data collected by database proprietors, enabling the combination of data from different parties while maintaining individual privacy, and using machine learning to correct for misattribution and non-coverage errors.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If third-party cookies and tags are used to track census-wide impressions, then user identification and demographic information collection are improved, but user privacy protection deteriorates
Solution Approach 1:
A trusted third-party cloud environment acts as an intermediary between data sources (panel data, census data, advertiser data) and the audience measurement entity. This intermediary processes and merges data without exposing raw personal information to any single party, enabling accurate audience metrics while protecting user privacy through controlled data access and aggregation.
2Loss of information
If multiple data sources are merged to improve audience metrics, then measurement completeness is improved, but data processing complexity increases
Solution Approach 1:
The data processing system is segmented into distinct functional modules within the trusted cloud environment: data ingestion modules for different sources (panel data, census data, advertiser data), data matching modules that correlate records using identifiers, merging modules that combine datasets, and output generation modules. This segmentation manages complexity by organizing processing tasks into independent, manageable components.
Solution Approach 2:
The trusted third-party cloud environment serves as an intermediary processing platform that handles the complexity of merging multiple data sources. It provides standardized interfaces for data input and output, manages data matching and merging operations, and returns processed results to the audience measurement entity without requiring the entity to directly manage the complex integration logic.
3Object-affected harmful factors
If third-party cookies are eliminated, then user privacy protection is improved, but the ability to track census-wide impressions deteriorates
Solution Approach 1:
Instead of relying on third-party cookies that directly track users across websites, the system uses copied and transformed identifiers (such as hashed device identifiers, aggregated demographic profiles, and anonymized user attributes) that preserve the ability to measure impressions without exposing raw personal information. These copied data representations enable tracking while maintaining privacy through transformation.
Solution Approach 2:
The trusted third-party cloud environment acts as an intermediary that replaces the direct tracking function of third-party cookies. It receives anonymized data from various sources, processes it to generate audience metrics, and returns results without requiring persistent cookies on user devices. This intermediary approach maintains measurement capability while eliminating the privacy risks of traditional cookie-based tracking.
Data Source
AI summary
An example system disclosed herein includes programmable circuitry to identify donor adjustment factors and recipient adjustment factors used for correction of media impressions logged by a database proprietor, the donor adjustment factors including first donor adjustment factors associated with a first geographic region and second donor adjustment factors associated with a second geographic region, determine a first reduced donor factor set corresponding to ones of the first donor adjustment factors that satisfy a threshold, determine a second reduced donor factor set corresponding to ones of the second donor adjustment factors that satisfy the threshold, and generate imputation factors based on an aggregation of retained ones of the donor adjustment factors, the retained ones of the donor adjustment factors selected based on the first reduced donor factor set and the second reduced donor factor set, the imputation factors to reduce error in the correction.