​​​Real-time link analysis: fighting fraud with Persona’s Graph​​

Graph uncovers these connections, allowing you to stop fraudsters from onboarding to your product, taking over accounts, and scaling their attacks. Even if they switch devices or hide their IP address, Persona’s device identifiers can help reveal the connection and suspect ​​​behaviour​.

Graph-based link analysis is an effective barrier against fraud rings, deepfakes, identity mules, and other sophisticated attacks. It can also offer important insights that you can use to protect your customers who are in the midst of a social engineering trance.  

Graph can connect with other parts of Persona’s platform in real time. What processes in a fintech’s or bank’s customer life cycle does the solution address? 

The platform approach enables you to use the insights you collect from link analysis throughout user life cycles:  

• Onboarding: Use the results to increase assurance when you’re running initial checks during onboarding. You can also automate decisions, such as blocking users who share the same device with 10 other users.  

• Transactions: Rerun Graph to check for suspicious signals and connections before approving specific transactions.  

• Account reviews: Schedule recurring Graph queries, such as ​‘​​​find clusters of 15+ accounts sharing device fingerprint created in the last seven days​’​,​​ and receive Slack or email alerts with the results.  

• Account management: Tag clusters of bad actors, compromised devices, or identification elements. Add these tagged groups to blocklists and create rules to auto-deny future attempts from connected bad actors. 

• Investigations: Immediately see Graph connections in Cases, our investigations hub, to determine if it's an isolated incident or part of a coordinated ring. You can also start an investigation in Graph Explorer, the visuali​s​​​ation tool.  

What can link analysis see that identity verification fraud checks cannot? 

Identity verification checks whether the person is who they claim to be at a singular moment in time. Link analysis checks whether the user is connected to known fraud in the past or future. It’s an important distinction.  

For example, a fraud analyst at Persona noticed that users were submitting IDs with different portraits and matching birthdays. A single birthday won’t flag risk signals, and you wouldn’t create a verification check against a specific birthdate because you’d erroneously block a lot of good users.  

But the analyst was suspicious, and he used Graph to quickly uncover a fraud ring using an ID template. He blocked dozens of bad accounts and figured out how to block similar attempts in the future based on the accounts’ shared attributes.  

How is Persona’s Graph different from other fraud solutions on the market? 

Persona’s Graph integrates automatically with Persona’s platform to sync data from our verification flows, like device identifiers, IP addresses, and user-submitted images. Alternative solutions can be restricted to batch processing, but if you want to catch attacks as they scale and make decisions during the user session, you need real-time link analysis. 

Graph is also really flexible and allows you to sync custom data, like your promotion codes or hashed payment information. Graph query results can also be returned to you via API or webhooks if you want to trigger, monitor, or analy​s​​​e them elsewhere. 

There are many ways to approach link analysis, and comparing the options is definitely important. Some organi​s​​​ations opt to build and maintain these tools on their own, while others buy solutions. We actually recently demoed Graph to the fraud team at a company with an in-house link analysis tool. Now, they’re using Persona’s Graph for fraud detection and large-scale investigations, and their internal tool for risk scoring.  

There’s also a benefit for fraud teams who use Graph to add more context to their audits and board reports. Instead of saying, ​‘​​​we blocked 500 accounts this quarter​’​,​​ they can add the context that ​‘​​​we identified and blocked 12 fraud rings connected on xyz properties ​​​totalling​ 500 accounts​’​.​​  

Can you share some real examples where Graph-enabled fintechs or FIs detect and stop fraud? What are some learnings that these shared? 

Of course! Without naming customers directly, I can say we work with a lot of fintechs and financial institutions, and I can share some important trends we’re seeing.  

We’ve been seeing synthetic identity fraud and identity mule attacks, sometimes with deepfake selfies. We’ve also been seeing more cross-platform fraud, where bad actors test ​​​defences​ across multiple fintech products to find the weakest link, then scale attacks within that target organi​s​​​ation. 

Some recent examples: 

• A large fintech platform used Graph to find over 1,700 accounts linked to flagged devices and determined ​that ​over 99% of the accounts were fraudulent. There were five browser fingerprints linked to over 300 accounts.  

• A fintech lender found that 97% of fraudulent accounts were linked to five or more accounts by IP address, browser fingerprint, Social Security number, phone number, or email address. We estimated they could save at least ​USD ​​​8.9​ ​​​​mln​ by automating decisioning and blocking fraud with Graph.  

• A mobile payment platform with tens of millions of users manages chargebacks by automatically checking whether the account filing a dispute is linked to other suspicious accounts.  

You can read more customer stories on the Persona website. We also recently published an ebook about how link analysis works and what an effective link analysis strategy looks like in 2026. Or, if you’d prefer to watch or listen, there was a recent webinar with Handshake AI focused on pattern and link analysis.  

​​​About the author​​ ​​

Anita Ilango is a product manager at Persona, where she works on Graph, a product that helps teams uncover hard-to-detect fraud using link analysis. Previously, Anita co-founded Goldilocks Labs, exploring new approaches to search and knowledge discovery. Before that, she was an early employee at Clay, serving as the company's first product manager. Earlier in her career, Anita was a Lead Venture Partner at Contrary, working with early-stage founders. She holds a B.S. in Computer Science from UCLA.​​ ​