Tracking Identification Groups for Cross-Device Data Continuity
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing tracking systems face challenges in maintaining tracking quality across devices and sessions due to issues such as device switching, network changes, ad blockers, limited cookie lifetimes, and compliance with data privacy laws, leading to decreased accuracy in measuring advertisement effectiveness and identifying recurring customers.
Innovation Solution
A system that generates tracking identifications based on entity identifiers, creates identification groups using similarity measures, and assigns tracking data to these groups without requiring explicit consent, utilizing adaptable footprints and encryption to ensure compliance with data privacy rules, enabling cross-device and cross-session tracking.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If cookies are used to track users, then tracking can be achieved, but tracking quality decreases due to limited lifetime and prevention by ad blockers
Solution Approach 1:
The patent introduces an intermediary system that uses device fingerprints and machine learning models to bridge the gap between fragmented tracking data. Instead of relying solely on cookies that expire or are blocked, the system creates a mediator layer that identifies users through device characteristics and behavioral patterns, enabling continuous tracking across sessions and devices without direct cookie dependency.
Solution Approach 2:
The system changes the tracking parameters from cookie-based identification to device fingerprinting and machine learning-based user modeling. By transforming the identification mechanism from transient cookies to persistent device characteristics and learned behavioral patterns, the system maintains tracking quality beyond cookie lifetime while adapting to user actions across different devices and sessions.
2Loss of information
If tracking pixels are used to capture user visits, then tracking data can be collected, but website loading speed decreases due to increased overhead
Solution Approach 1:
The patent extracts the tracking functionality from traditional pixel-based approaches and relocates it to a server-side machine learning system. Instead of loading tracking pixels on every webpage which increases overhead, the system collects minimal device information and behavioral data on the server, where machine learning models process this information to identify users and generate tracking data, significantly reducing client-side overhead and improving loading speed.
3Adaptability or versatility
If device switching or network changes occur, then user interaction continuity is maintained, but tracking sequences are broken and user journey quality decreases
Solution Approach 1:
The system implements dynamic user identification that adapts to device switching and network changes. Instead of static cookie-based tracking that breaks on device changes, the machine learning model continuously learns from new device fingerprints and behavioral patterns, dynamically updating user profiles to maintain accurate tracking across different devices, sessions, and networks while preserving user journey continuity.
4Measurement precision
If comprehensive tracking is implemented to measure advertisement effectiveness, then measurement accuracy improves, but compliance with data privacy laws becomes more difficult
Solution Approach 1:
The patent replaces traditional mechanical cookie-based tracking with a machine learning-based system that processes device fingerprints and behavioral data to create user models. This substitution enables comprehensive tracking for accurate advertisement effectiveness measurement while maintaining privacy compliance through server-side processing, minimal client data collection, and machine learning models that infer user characteristics without storing sensitive personal information.
Data Source
AI summary
A system for providing tracking data includes a processing component configured to: receive an event dataset including a plurality of entity identifiers, generate a tracking identification based on the entity identifiers, generate a plurality of identification groups based on a subset of entity identifiers, determine a similarity measure between the tracking identification and an identification group of the plurality of identification groups, and assign the tracking identification to one of the identification groups based on the similarity measure.