Leverage Federated Data to Fight Financial Crime with Feedzai Risk Ledger

Fraud is big business, with increasingly sophisticated criminal rings launching complex cross-channel attacks against a multitude of institutions, industries and verticals. Fraudsters know that more and more businesses have fraud prevention solutions, so they make sure to pick several targets to increase their chances of success, moving from one business to another and leaving few traces of their activity.

For fraud analysts, knowing customers’ behavior, typical fraud schemes and attack vectors is necessary but no longer sufficient. If they assess only a handful of data points, even the most advanced fraud solutions will fail due to tunnel vision. To prevent this, datasets need to include federated data from different sources in order to reconstruct the missing information and reveal the fraudsters’ actions.

Feedzai supports this through Feedzai Risk Ledger, a consortium of pseudonymized data aggregated from our diverse customer base and from third-party vendors. It provides insights that enable comprehensive, timely, and accurate financial crime detection and prevention at scale. To see why consortium data is critical to fighting fraud today, consider this example Feedzai observed at a client, as seen through Feedzai Genome, our dynamic visualization engine.

Image 1: Without Risk Ledger

Image 2: With Risk Ledger

On the left is the visual pattern of what appears to be a regular, loyal customer using the same card, address, email and device over several months.

The picture on the right shatters this idyllic illusion. Thanks to consortium insights, we see that this person is actually a criminal who tested several cards elsewhere and then used one of these cards to commit fraud for months—without even changing the contact details.

After several months, our client received fraud chargebacks for each transaction made by the “loyal” customer on the left. The fraudster was blocked, but it was a high price to pay for not having consortium insights in the first place—which would have helped our client avoid the fraudulent transactions altogether.

The idea of using federated data for deeper insights, while intuitive, is not without its challenges. Traditional data consortia are not always equipped to address rising privacy concerns or to capitalize on modern technologies such as machine learning to boost detection effectiveness. Their data is often narrow (limited to a single industry or use case) and thus might miss emerging fraud attacks. It also often lacks context, hindering the ability to assess the relevance of the data to the problem at hand.

Feedzai Risk Ledger changes the fraud landscape by overcoming these challenges in three key ways: through data diversity, by protecting privacy and by harnessing the power of machine learning.

Increasing data diversity

To understand why data diversity matters, let’s take an example from everyday life. Choosing a restaurant has become incredibly easy thanks to an abundance of restaurant rating websites. But, there are a few things that can seriously limit the value of such a platform. The platform might have few data points overall (be it restaurants or reviews). Or, it might not have any data about restaurants in your city, despite having plenty of data about restaurants elsewhere, limiting the breadth of the ratings. The platform might also lack multiple points of view, which would limit the depth and accuracy of the ratings, as different people assign different importance to food, service and ambiance; they also have different budget and dietary restrictions, therefore being less likely to visit and rate every restaurant. To avoid these pitfalls, it’s critical that the platform have lots of diverse data with enough data points across each area covered.

The same applies to fraud prevention. Feedzai data originates in more than 200 countries and territories, as well as across verticals (banks, credit unions, acquirers, processors and merchants), and across the entire customer lifecycle (account opening, account monitoring, transactions and compliance). The scale and breadth of this data ensure wide coverage, while the diversity of participants is key to catching fraudsters red-handed regardless of how and where they choose to execute their attacks. Indeed, the card testing attempts discussed above could have been easily detected by combining data from merchants, banks and third party vendors:

Protecting Privacy

For each of its many data sources, Risk Ledger incorporates data security and privacy by design. All personal data ingested in Risk Ledger is pseudonymized and pre-aggregated into profiles, so no sensitive information is shared among participants—only insights are.

Harnessing the power of machine learning

Because consortium insights lack the nuance that makes your business and your fraud trends unique, it’s not a replacement for your data. Where it excels is providing a different perspective that augments your knowledge where it’s limited.

Common “blackbox” solutions, such as consortium scores, don’t contain enough context to ensure this new perspective is actually relevant to your use case. For instance, going back to the restaurant review site example, the score is usually a mere starting point. Before committing to a new culinary experience, users usually take a look at the menu and sometimes read the reviews. These are both examples of what we call “whitebox” data: data plus context, which provides a better basis for decisions than standalone, “one-size-fits-all” scores or models.

When it comes to making decisions based on rich information, modern machine learning technology is the one to beat. Built on top of a dense knowledge graph and trillions of Segment of One profiles, Risk Ledger’s whitebox data takes full advantage of our machine learning engine. This achieves the optimal blend of the existing data and the relevant consortium insights that best complement it.


How does diverse federated data, coupled with advanced machine learning, benefit risk and compliance teams in practice? In short, Risk Ledger provides three key advantages:

1. A more complete fraud and AML solution

Risk Ledger insights augment your field of view and cover the blind spots of your fraud and AML prevention system, allowing you to detect fraud or suspicious transactions that wouldn’t be captured otherwise. Clients can prevent millions of dollars worth of fraud and stop money laundering with Risk Ledger insights alone, and even more by combining Risk Ledger with their own data.

2. Detecting fraud and suspicious transactions earlier

Greed is revealing. Most fraudsters who continuously abuse your business will eventually be caught by your current fraud and AML prevention system. But this comes at a cost. Many fraudsters operate for weeks or even months, however, before being noticed. Risk Ledger insights provide a consortium “background check” to stop fraudsters early in their tracks.

3. Reducing customer friction

In commerce, a frictionless customer experience is a big competitive advantage. Every fraud prevention system misfires once in a while. This is a consequence of making decisions based on little data. Just like juries are composed of more than one person, your data strengthened with consortium data will provide more robust decisions and reduce the false positives that can introduce friction for good customers.


Feedzai is deeply committed to fighting financial crime through robust solutions to manage risk and keep customers safe. Risk Ledger is the result of Feedzai’s continuous work to provide risk departments with the best access and visual insights about financial crime patterns. It follows the recently-launched Feedzai Genome, a tool that uses graphic link analysis on top of Feedzai’s advanced AI platform to find networks and patterns of financial crime more effectively and faster than ever before. By plugging directly into Genome, Risk Ledger supercharges risk and compliance teams.

Subscribe to stay infomed