Why fraud investigations are better with data and analytics

Virtually all industries and organizations can fall victim to fraud, be it financial fraud, insurance fraud, identity theft, or other types of criminality. Anti-fraud detection systems can flag suspicious activity - but that’s only the beginning.

When a detection system flags a customer, transaction, or behavior as suspicious, an analyst must determine if it’s an actual fraud incident. They also need to understand if the case is isolated or part of a larger fraud scheme. A single fraudulent transaction may be evidence of a criminal ring, or the suspicious behavior indicative of a wider scam. 

In this article, we’ll look at the typical steps of a fraud investigation. Getting to the bottom of the case requires exploring what the customer or behavior is connected to, so we’ll also look at the role data plays in a fraud investigation.

Fraud investigations can of course vary according to the industry and the fraud typology. But many investigation processes will follow the same general flow. Let’s look at the main steps in a typical investigation of financial fraud.

The first step is to gather all the information you have on the case, including transaction records, statements, and any other relevant data. This information can be obtained from various sources, including the victim, the bank's internal systems, and external sources like law enforcement.

Once you have collected all the relevant information, you need to determine if a fraud has actually occurred. This involves identifying any suspicious activities, unusual patterns, or discrepancies in the data.

Once you have identified the fraud, you need to secure all the evidence related to the fraud. This includes preserving records of transactions, emails, and other relevant documents that could serve as evidence in the investigation.

Next, you need to interview the parties involved, including the victim, the suspect, and any other witnesses. This will help you to gain a better understanding of the circumstances leading to the fraud and to gather additional evidence.

Analyzing the data involves using analytical tools to identify any trends or patterns in the data. This will help you to uncover any hidden connections or relationships that may exist between different transactions or parties involved.

Once you have completed your investigation, you need to report your findings to the appropriate authorities, such as law enforcement, the victim, or senior management at the financial institution. A fraud report should include a summary of the investigation, the evidence gathered, and any recommendations for preventing future fraud.

Finally, the financial institution needs to take appropriate action based on the findings of the investigation. This could include freezing the account, terminating the suspect's access to the account, or filing a police report. The financial institution should also take steps to prevent future fraud by implementing stronger security measures and monitoring systems.

Gaining a full understanding of any given fraud case requires finding evidence within your data: personal data, transaction data, public data, data about businesses, and more.

Data structure has a major impact on the speed and efficiency of a fraud investigation. Many organizations use manual fraud investigation techniques, which are time-consuming and can miss key information. Finding the information to understand if wrongdoing has occurred or not is a long process when the data is scattered across different tabs. Verifying the identity of individuals connected via financial transactions, for example, requires opening a first tab with the person’s transactions and recipients. Then you need to repeat the process for each recipient.

Graph analytics facilitates the dynamic exploration of relationships within a large dataset. An investigator can explore and visualize who and what a client is connected to through an address, a phone number, an email, transactions, etc. Detecting accomplices becomes much faster.

Graph algorithms quickly answer many of the questions that are relevant to fraud investigations. Is one of your clients indirectly connected to a politically exposed person? A pathfinding algorithm will reveal if they are. Is a given fraudster actually part of a bigger network? A community detection algorithm can help you understand if they are, and who their accomplices might be.

Let’s look at an example of how a solution like Linkurious, based on graph analytics, can help you gain investigative efficiency. Synthetic identity fraud is when a criminal uses a mix of real and fake or stolen information to open an account or take out a loan. Here’s what that can look like: Maxine Williams and John Leahy share a phone number. John and someone called Thomas Zhao share an address. And Thomas and Maxine share an email address. It’s likely that they are part of a fraud ring using synthetic identities. Tracking down these connections manually would be very time-consuming.

Doing a thorough fraud investigation that gives you a complete picture of what happened helps inform risk. And better understanding fraud risk means better fraud prevention. 

Fraud investigation is part of a virtuous circle. The intelligence gathered from an investigation helps identify new patterns and criminal methodologies. And that information can in turn be used to create countermeasures to prevent the fraud from occurring again. It can also help scale fraud detection, particularly when using a graph analytics solution, in which it’s easy to create new alerts to flag new types of suspicious activity.