How a global marketplace unlocked $40M in projected annual fraud savings

The company

A global online marketplace brings millions of customers together with sellers and service providers. Its business depends on making those connections easy, and helping customers transact with confidence.

The challenge: Stopping fraud without slowing down legitimate users

With every transaction, the marketplace must decide: Is this a trusted user or a potential fraudster? As payment fraud grew more sophisticated, costly chargebacks were driving revenue loss and operational overhead.

The team faced several challenges:

• Existing identification methods relied primarily on basic account attributes, making it difficult to connect related fraudulent activity across transactions.
• Any new fraud technology needed to operate with extremely low latency to avoid affecting conversions.
• Limited internal resources made it difficult to deeply analyze fraud data and uncover meaningful patterns.

The solution: Uncovering the fraud signals hiding in their own data

To strengthen its fraud defenses without adding friction, the marketplace turned to Fingerprint.

By analyzing 100+ attributes from a visitor’s device, browser, and network, Fingerprint can recognize returning visitors and surface risk indicators that traditional account identifiers miss.

Fingerprint was implemented by the marketplace’s fraud team to analyze historical data and uncover the device characteristics uniquely predictive of fraud in its environment. The analysis identified signals that consistently appeared in fraudulent transactions but were exceptionally rare among legitimate users, including residential proxies, tampered browsers, and devices associated with unusually high transaction activity across multiple accounts.

These signals helped the team connect seemingly separate transactions and build more precise fraud rules, allowing them to catch more fraud while keeping legitimate transactions moving.

The results: Smarter fraud decisions, backed by clearer signals

With a clearer view of fraud patterns in its own data, the marketplace built more precise rules and improved detection, resulting in:

• $40 million in projected annual fraud savings
• No measurable impact on conversions or application performance
• Faster investigation and rule creation using device-level insights

The team now connects fraudulent activity across accounts and transactions, while preserving the seamless experience legitimate customers expect.