This is done by proactively detecting malicious mass registrations, account takeover, and fake listing scams. DataVisor recently worked with a global online marketplace to prevent an aggressive spate of fraudulent attacks that threatened to cause the brand and its customers potentially irreparable damage.
The client operates in more than 40 countries, and helps over 350 million monthly active users connect with one another to buy and sell goods and services. Because trust is central to the success of the platform, the attacks were especially alarming, and the organisation was in need of a powerful and effective fraud prevention solution to restore and build customer confidence before further harm could be done.
The attacks came in the form of mass-registered accounts that would first incubate for weeks, then start launching fake listings to scam good users. The impact on the brand’s reputation was immediate, and customer churn rates began to increase markedly.
Prior to enlisting DataVisor, the client was relying on internal systems for fraud detection, but these systems were not up to the challenge posed by this new breed of attack. The mass registrations were highly coordinated, and the client’s existing tools could only capture a portion of the fraudulent activity. Most concernedly, they were not able to reliably detect large-scale fake listings, due to a fundamental inability to analyse unstructured data and metadata.
This all changed when the client brought DataVisor to the fight.
Combining unrivalled domain expertise with ground-breaking unsupervised machine learning capabilities, DataVisor’s solution introduced a proactive approach that immediately began to accurately surface suspicious accounts and coordinated fraudulent registrations—even those still early in the incubation stage. Additionally, DataVisor’s systems flagged scam content by analysing posts and images and spotting similar attributes and behaviours across accounts.
Operational efficiency increased significantly, made possible by DataVisor’s auto-action and bulk review capabilities. 65% of cases were automatically routed to a friction-heavy queue without manual review. For less suspicious cases, analysts were able to make bulk decisions by reviewing all correlated cases at one time, and accordingly accelerate overall review time.
Below are some of the fraud patterns that DataVisor’s systems detected:
“Hit-and-run” behaviour: 60% of fraudulent accounts made the first attack within 2 hours of registration; 76% made the first attack within 24 hours of registration.
Sleeper cells: some accounts logged in, then remained dormant for weeks prior to a large-scale scam.
Human-operated scam farms: a group of scam armies was highly correlated with behavioural patterns such as event sequences, event time, and intervals. They attacked on weekdays and rested on public holidays and weekends.
Disposable emails for registrations, and bots for scripted logins from cloud hosting IPs.
2-minutes intervals between login and attack, and similar listing descriptions created from templates with shared URLs.
Overall results were dramatic. DataVisor’s fraud management solution caught 88% of fraudulent accounts before the first scam. In the largest fraud ring, the system captured over 68,000 accounts that could have negatively impacted the client and its customers. DataVisor delivered detection accuracy rates that were 20% higher than the client’s existing solutions, and the combination of early, accurate detection and improved efficiency served to prevent any further reputational or financial damage.
Interested in learning more about how DataVisor help clients defeat fraud?
Download DataVisor’s case study booklet: Fighting Fraud with Machine Learning: Stories from the Frontline
About Fang Yu
Fang Yu is the Cofounder/CTO of DataVisor, where her work focuses on big data for security. Fang has developed algorithms for identifying malicious traffic including fake and hijacked accounts, and fraudulent financial transactions. Fang received her PhD from UC Berkeley and holds over 20 patents.
About DataVisor
DataVisor is the leading fraud detection platform powered by transformational AI technology. Using proprietary unsupervised machine learning algorithms, DataVisor restores trust in digital commerce by enabling organizations to proactively detect and act on fast-evolving fraud patterns, and prevent future attacks before they happen. Combining advanced analytics and an intelligence network of more than 4B global user accounts, DataVisor protects against financial and reputational damage across a variety of industries, including financial services, marketplaces, ecommerce, and social platforms.
Every day we send out a free e-mail with the most important headlines of the last 24 hours.
Subscribe now