What was the core problem?

While transitioning from a banking product to a rewarded advertising platform in early 2024, Copper came under attack from fraudsters. The team spent months dealing with sophisticated fake accounts outside target geographies, cross-continental device farms and bots that sought to abuse gaming incentives, surveys, and earning opportunities that Copper offers on behalf of its advertising partners.

‍

What did they try first?

Given their history as a fintech platform, Copper tried to configure Sardine (a Series C fraud prevention platform) to identify fake accounts. The system was rule-based, collecting a myriad of signals and requiring the Copper to configure settings to catch new fraud patterns. The team spent hundreds of hours over several months analyzing patterns in user behavior to identify users most likely to commit fraud. Maintaining this system felt like a game of whackamole - fraudsters quickly discover the reason they got blocked and adapt their behavior. 

‍

Copper needed a solution that was signal-first, not rule-first - an automated platform that could detect "fake users" out of the box without constant manual intervention.

Copper went through a robust bake-off process to find their new fraud provider, comparing Sardine, Socure, Fingerprint, and Verisoul. The products were tested head to head for months on live production traffic - and then chose a combination of Verisoul and Fingerprint.

‍

Why not Socure? While Socure is an industry standard for Identity Verification (KYC), it wasn't agile enough for the rewards use case.

• Fragmented Signals: Socure often charges per API call for distinct signals (e.g., email risk, phone risk), requiring the client to stitch those signals - and manually created rules - together to form a decision. 
• Banking Expenses: Socure’s pricing model is designed for high-value banking users. For a rewards site managing high-volume traffic, the unit economics simply didn't work.

‍

Why Verisoul?

Verisoul provided the winning combination of cost-effectiveness, reliability, and ease of integration/management not seen in the other options. Verisoul’s pre-configured machine learning model allowed Copper to see the benefit in week 1. 

‍

I thought Verisoul offered device fingerprinting. Where does Fingerprint fit in?

Copper initially ran Fingerprint alongside Verisoul, assuming they needed a dedicated device ID provider.

• Redundancy: Testing revealed that Verisoul’s native device fingerprinting was equally robust. 
• Ineffective against sophisticated fraud: Fingerprint provided a signal on device intelligence, but it was not able to catch sophisticated fraud farms that have access to legitimate, unique devices used to get past traditional device fingerprinting point solutions.
• Consolidation: Copper is now deprecating Fingerprint, as it added cost and potential false positives without offering unique coverage.

‍

So, how is Copper using Verisoul?

Verisoul’s solution worked out of the box, and was the obvious choice during the free 1-month testing phase. After months of data collection, Copper has fine-tuned Verisoul’s wide-ranging suite of custom configurations. At first, they used Verisoul’s signals only at registration, but have since expanded to checking users pre-offer and at withdrawal.

‍

The best part? Setting Verisoul up took hours, not months. 

‍

Copper has seen benefits in several core areas since switching to Verisoul, including decreasing fraud rates, cost, and time, as well as experiencing growth in their partnership pipeline.

‍

1. Decreasing reversal rates by 37% - with Verisoul, Copper has been to more effectively catch fraudsters vs Sardine
2. Reduction in cost for their fraud solution: Copper now pays 76% less for Verisoul than they did for Sardine. In addition, Verisoul has proven to be an all-in-one platform, allowing them to move off of Fingerprint as well. 
3. Time savings: Copper integrated Verisoul in just days compared to months of dedicated effort with Sardine. In addition, the overhead of maintaining the model falls on Verisoul’s machine learning models and “research lab” approach to detecting new fraud patterns, not homegrown, in-house approximations inside of the Sardine platform.
4. Direct integration with gaming offerwalls. Using a banking-grade fraud solution has been key in helping Copper integrate directly with gaming offerwalls. In underscoring their prioritization of high quality traffic, end advertisers feel more confident working with Copper’s partners. A win-win-win for everyone!