Leading Audience Data and Identity Resolution Provider Case Study

How Metal Toad helped a leading audience data provider use graph machine learning to connect fragmented consumer data, improve identity matching workflows, and support more accurate consumer identity resolution.

BUSINESS CONTEXT

THE CHALLENGE

The client, a premier provider of audience data, faced a critical industry challenge: fragmented consumer identity. As consumers move across devices and platforms, their digital footprints become increasingly scattered. The client’s traditional identity resolution methods relied on rigid "deterministic" and "probabilistic" rules, which struggled to link disparate data points to a single, unified consumer profile.

For their end-users—marketers and data analysts—this data fragmentation resulted in incomplete profiles, wasted ad spend, inaccurate attribution, and generic customer experiences. The client needed a breakthrough method to uncover hidden connections between isolated data points without compromising massive scale or strict global data privacy standards.

Data Inteligence (1)-1
THE SOLUTION

THE TECHNOLOGY: Graph-First Identity Resolution Powered by AI

Metal Toad engineered an AI-powered, "graph-first" identity resolution engine to elevate the client's legacy matching systems. Leveraging AWS Cloud Native Services, our team architected a robust pipeline centered around Amazon Neptune for highly flexible relationship modeling. To move beyond basic exact-data matching, we integrated Amazon SageMaker and GraphStorm.

By deploying advanced Graph Neural Networks (GNNs), the solution performs predictive "link matching" to identify missing edges between isolated data points. Instead of just matching labels, the machine learning model analyzes the structural "shape" of the data, predicting hidden consumer relationships and assigning a confidence score to each predicted link. The entire infrastructure was secured using enterprise-grade encryption via AWS native services, including Amazon S3 and Neptune, ensuring full compliance with global data privacy standards.

 
 
 
 

 Struggling with fragmented customer data?

Metal Toad can help you design graph-based AI solutions that connect identity signals across systems.

THE IMPACT

THE MEASURABLE OUTCOMES

  • Produced actionable machine learning insights within the first 3 weeks of a rapid 6-to-8-week Proof of Concept (PoC) engagement.
  • Validated measurable improvements in data matching accuracy and relationship linking compared to the baseline of their legacy implementation.
  • Delivered a fully modular, production-ready pipeline including SageMaker model training notebooks and standard output formats immediately compatible with major CDPs, DSPs, and analytics platforms.
 
 
 
 

Ready to connect identity signals across systems?

Metal Toad can help you design graph-based AI solutions that improve identity matching and customer data workflows.

Schedule a  Consultation

See how graph-based AI can help connect identity signals across systems.

Talk to an expert

Featured Case Studies

Case Studies : RPGMatch Bio-creation

Discover how Metal Toad transformed user engagement with RPGMatch by implementing an automated bio generation feature, enhancing player connections...

Case Studies : Wheel of Fortune

For over 40 years, contestants on Wheel of Fortune have spun the wheel to win big on TV. And today, viewers at home can join in the excitement — by...

Learn more about GenAI on our blog

Artificial Intelligence

How do I pick the right GenAI Model?

Learn how to choose the right GenAI model for your business with insights on AWS Bedrock and key industry players. Discover our expert tips and...

Joaquin Lippincott, CEO Mar 10, 2025

Learn more about our GenAi Service offering

Services: Generative AI Services

Discover cost-effective Generative AI solutions with Metal Toad, leveraging AWS expertise to enhance workflows, drive innovation, and deliver...