“We were able to leave the fraud to Sift Science, allowing us to focus on growing and scaling our business.”
Julia Kurnia,Founder and Director
Person-to-person microlending nonprofit
Offers lower cost loans due to lean processes and less overhead
Fraud was an existential problem for the business
Manual review process couldn’t scale as the business grew
Detected connections between users and fraudulent accounts using the Network Visualization tool
Utilized the Sift Science console to evaluate all the relevant information about a user
Reduced manual review times by 80%
Improved loan repayment rates by 15%
Zidisha is the first person-to-person microlending nonprofit to connect borrowers with lenders without any intermediaries. As of 2015, lenders in 144 countries are providing loans to borrowers in 9 countries.
Zidisha's lending model is unique in the world of traditional microfinance loans. Microfinance loans generally have large overhead costs due to fees from local banks and administrative infrastructure costs, which ultimately eat a substantial portion of the borrowers' earnings. The high administrative cost is passed on to the borrowers, who pay on average ~37% in fees and interest. Zidisha lowers that cost to 5% by using technology to manage the loan process and eliminating staff and administrative overhead in their borrower countries.
Before discovering Sift Science in early 2014, fraud was an existential problem for Zidisha; anything that affected loan repayment rates threatened the core business and the financial stability for the entire organization. Because Zidisha lacked an extensive network of loan officers, staff, and other personnel, fraud was a very large risk and ever-growing issue.
Julia Kurnia, the founder of Zidisha, found that the organization's largest fraud problem came from fraudsters creating networks of “sock-puppet” accounts in borrower countries. These fraudsters would help people apply for Zidisha loans, take kickbacks from the borrowers, and actively encourage borrowers to not repay the loans.
Julia and her team of volunteers spent countless hours on manual verification by reviewing application details, doing reference checks, making Skype calls, and much more. However, this intensive process couldn't scale, and they had incredible difficulty finding a solution that would allow them to stop fraud accurately and efficiently.
When Julia discovered Sift Science, she was excited to find a scalable solution that would help her team catch fraud without breaking the bank. She fully integrated Sift Science for Zidisha in just a few weeks and started seeing the impact immediately.
Sift Science's ability to analyze hundreds of different attributes to uncover hidden and complex connections between users allowed Julia and her team to quickly see if someone submitting a loan request was linked to suspicious or known bad users. The Network Visualizations tool was incredibly helpful at catching and stopping rings of “sock-puppet” accounts. Additionally, Zidisha's volunteers used the Sift Science console to see a snapshot of all relevant information about a user, allowing them to quickly stop fraud with minimal time investment.
“Sift Science could tell us exactly who were our bad (and good) users were and exactly what they were doing.”
Sift Science's ability to detect fraudsters made it incredibly easy for Julia and her team to prevent fraudsters from targeting their business. Before Sift Science, the time spent on manual reviews and loan verifications was a substantial drain on Zidisha's volunteer resources. Sift Science allowed Zidisha to reduce that time by 80%, since they could now automate the review process and focus on intensive manual reviews for only the riskiest applicants.
By using Sift Science, Zidisha effectively solved what had seemed like an impossible problem: blocking “sock puppet” fraudsters who threatened the financial solvency of the entire organization. With Sift Science's advanced machine learning technology and link analysis of connections between users, Zidisha was able to reduce its fraud problem. The best part? They were able to stop fraudsters while also increasing their loan repayments by over 15%.