Unlocking Customer Value Through Graph Reasoning
Customer improved their targeting and cut their compute costs 10x.
With tens of millions of customers and no physical storefronts, our customer needed a better way to understand how clients interact and influence each other. By using graph reasoning with RelationalAI, they turned peer-to-peer payment data into actionable insights—while saving time, money, and engineering effort.
The Challenge: Understanding a Social Network of Money
Our customer’s product isn’t a thing—it’s a network. Users send money, trade stocks, and share bitcoin. Every transaction adds relationships, creating a massive web of connections.
They needed to answer key questions:
Who are the most important customers to retain?
Which customer groups are drifting apart?
What does each customer’s relationship network reveal?
But traditional analytics tools couldn’t keep up. Running queries on tens of millions of users and their 10+ monthly transactions meant huge self-joins and long runtimes. Graph algorithms ran overnight, if they finished at all. It wasn’t scalable or cost-effective.
“The value of the company increases the more people use it. That’s why we need to understand customer behavior.”
—Head of Network Science & Behavioral Modeling
The Solution: Graph Reasoning in Snowflake with RelationalAI
Customer brought graph reasoning into their Snowflake environment using RelationalAI’s native app—no ETL needed.
Customers became nodes, and transactions became edges in a knowledge graph.
Metrics like transfer counts and averages defined relationship strength.
Graph algorithms like Eigenvector Centrality helped identify key influencers.
Community detection using InfoMap revealed groups that behave as tightly-knit financial ecosystems.
Instead of building external systems, customer ran these analyses directly in Snowflake, preserving security, governance, and simplicity.
“With RelationalAI, we can run algorithms on the relationships—not just the raw data. All from within Snowflake.”
The Results: 10x Faster and Cheaper—and More Focused
A Strategic Advantage: From Observation to Action
The customer didn’t just analyze the network—they began shaping it.
Product managers now build for influence, not just demographics.
Marketing teams know who and where the key communities are.
The innovation cycle has accelerated, with faster experimentation and feedback.
“So the more healthy this network becomes, the more value we create—and capture.”
10x reduction in compute time and cost.
Increased speed to insight, allowing teams to experiment faster.
Improved marketing ROI by targeting key influencers and clusters.
“We now design products for our central nodes. We can focus resources where they matter.”
Customer also began defining KPIs around network centrality—prioritizing users who influence others. This helped drive adoption of new features like debit cards and optimize incentive strategies.
Project Information
Category
Finance, Tech
Applications
- Customer influence and network segmentation
- Product spread and incentive targeting
- Centrality-based product design
