Man taking webinar using laptop at coffee shop

Learn Best Practices for Graph-Based Identity Databases



Determining the legitimacy or risk of the identities behind a transaction during manual review can require multiple internal and external tools. The process is investigative by nature from basic checks like validating a person’s address to more advanced research to find a missing link between a name and address. Manual review agents need access to tools that enable them to accurately and efficiently mitigate fraud without turning away good customers and increasing false positives.

Oftentimes, associations between people can play an important part in understanding the details of a transaction. For example, a person could ship a birthday gift using their Seattle billing information to their mother in Texas who has a different last name as well as an IP and shipping address over 3k miles away. In this example, something that may look risky at first-sight could easily be approved by leveraging shared addresses, alternative names such as maiden names, and relatives to make a connection between the billing and shipping name.

Connections between people are often complex and can be made in multiple ways through digital and traditional identity data attributes such as names, addresses, phone numbers, and email addresses. Traditionally, the databases that powered manual review tools were structured statically and organized by row (think: Excel spreadsheet). However, in today’s world of sophisticated fraud, successful manual review requires dynamic graph-structured databases that are able to make, break, and maintain a web of connections in real-time (think: hub and spoke model or spider web).

Why is a graph-structured database necessary for manual review?

  1. Generally, people’s lives don’t fit neatly into a row— one person’s identity can be tied to multiple names, aliases, family members, and friends as well as multiple addresses, phone numbers, and emails that are all legitimate. A dynamic database is required to understand the complete picture of an identity behind a transaction.
  2. Networks between people evolve constantly making it necessary to add, remove, and reconnect links in real-time. Additionally, one update to a person’s network can impact multiple connections. A graph-structured database allows updates to be made in real-time across an entire database vs. a static structure that requires updates row by row.
  3. It’s imperative for manual review agents to approve good orders as quickly and efficiently as possible. Traditional database structures only have a few key fields to linearly search for information, which slows down the approval process. However, a graph-structured database makes it easier to navigate because each identity data attribute (name, email, phone, etc.) serves as a unique entry point to a person’s network— again, think about a hub and spoke model rather than a spreadsheet. As agents dive deeper to investigate an identity, any of these data points will provide access to the full picture.

Without the proper tools to review potentially risky transactions, manual review agents may delay or reject legitimate orders, thus increasing customer friction and customer insult rates. Much like social media platforms that use graph-structured databases to connect friends, Ekata uses a graph-structured databases to connect people and their associated identity data. Our global manual review solution, Pro Insight, provides a clear and focused user experience that leverages 20+ years of data sourcing, allows for easy navigation of the Ekata Identity Graph of over 8 billion links, and puts the power of real-time machine learning and sophisticated data science directly in the hands of the user.

To learn more about Pro Insight, contact us.

Mastercard Identity Avatar

About the Author

Related content