How ATM Online Boosted Risk Models and Cut Fraud in Vietnam with JuicyScore

A Strong Partnership for Smarter Lending

ATM Online and JuicyScore began working together in late 2020 with a pilot project to enhance the credit application flow in Vietnam. By integrating JuicyScore’s stable device ID technology and rich data vector, ATM Online was able to build a stronger risk tech stack, tackle fraud challenges, and improve the performance of its scoring models.

The pilot’s key outcome was the integration of JuicyScore’s score and vector parameters into ATM Online’s antifraud and credit risk assessment systems. In February 2021, the company launched JuicyScore’s solution in full commercial use across its Vietnamese platforms.

JuicyScore’s technology enabled ATM Online to filter out high-risk and fraudulent applications at scale — protecting the company’s portfolio and customer trust.

Denis Aleksandrov, Risk Director at ATM Online Group in Asia, notes:

“JuicyScore’s unique data vector — combining online behavior markers, connection quality signals, and device parameters — added 5 Gini points to our models’ predictive power. Combined with new antifraud rules based on JuicyScore’s device ID, we lowered credit risk and achieved rapid ROI in just a few months. We expect a 5x ROI as we fine-tune the strategy over the next six months.”

JuicyScore’s solution also helped cut commercial fraud significantly.

About ATM Online

ATM Online is an international digital lending company offering fast, flexible consumer loans through a single-platform solution. Established in 2015, it uses holistic scoring and highly predictive risk models. With more than 500,000 registered users and a credit portfolio exceeding $7 million, ATM Online maintains an impressive 90% customer return rate. The company integrates all major local payment channels, including Payoo, Viettel, and direct bank transfers.

Key Results

+5 Gini pointsAdded to credit scoring models

5x ROIWithin the first 6 months

90% retention rate9 out of 10 customers returning

“JuicyScore’s unique data vector — combining online behavior markers, connection quality signals, and device parameters — added 5 Gini points to our models’ predictive power. Combined with new antifraud rules based on JuicyScore’s device ID, we lowered credit risk and achieved rapid ROI in just a few months. We expect a 5x ROI as we fine-tune the strategy over the next six months.”

Denis Aleksandrov, Risk Director at ATM Online Group in Asia