Multi-Touch Attribution Using Hashed Identifier Matching
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The growing concern for user privacy in mobile and web applications has hindered the effective matching of datasets for advertising analysis, as users are reluctant to share identifiable information, complicating the computation of metrics like attribution, lift, and ranking in online advertising.
Innovation Solution
An advanced matching system separates impression and conversion datasets into separate data buckets, implements minimum bucket sizes, constructs user identifiers by concatenating data fields, hashes them, and computes the intersection using multiple identifiers to maintain privacy while maximizing matching likelihood.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If user identifiable information is collected for advertising analysis, then matching accuracy for attribution metrics is improved, but user privacy is compromised
Solution Approach 1:
The patent creates hashed copies of user identifiers that preserve matching capability while eliminating direct identification. Instead of using original user IDs, the system generates hash values that can be compared across datasets without exposing the underlying personal information, thus maintaining measurement precision while protecting privacy
Solution Approach 2:
The patent introduces hashed identifiers as an intermediary between raw user data and analysis systems. This intermediary layer enables dataset matching and attribution calculation while preventing direct access to identifiable information, effectively mediating between the need for accurate measurement and user privacy protection
2Reliability
If multiple identifiers are used for matching, then matching likelihood is maximized, but data complexity increases
Solution Approach 1:
The patent segments user identification into multiple independent identifier types (device IDs, advertising IDs, contextual identifiers). By dividing the identification process into separate segments rather than relying on a single complex identifier, the system increases matching likelihood while managing complexity through modular processing of each identifier type
Data Source
AI summary
Systems and methods herein describe privacy preserving multi-touch attribution. The described systems access a plurality of impression events and a plurality of conversion events, and for each impression event and each conversion event, wherein each impression event and each conversion event are associated with user identifiers, the described systems generates a hashed user identifier based on the associated user identifier, initiates a key agreement protocol comprising a key, generates an encrypted identifier by encrypting the hashed user identifier with the key, and stores the encrypted identifier.