Fraudzuki was developed in collaboration with Luke Todd & Co, originally designed to help Gumtree.com (UK’s number #1 classifieds site, part of the eBay Classifieds group) prevent policy breaches on its site. The provided set of tools also help detect fraudulent activity on the site, ensuring a better overall user experience.
With this project, we used previously acquired experience in Trust & Safety, in order to design & develop the product, We were also working really close with the customer support team who manage Trust & Safety on the site.
The developed modules helped automate repetitive tasks and searches which the team used to perform manually on a daily basis, helping the team focus more on analysing trends, which can then be used to automatically detect malicious behaviour using FraudZuki. The outcome of automating this process resulted in the same processes being performed automatically and more frequent, providing a better service with no additional cost.
The system contains a Rules Engine, which works using site searches and an anomaly detection module that automatically tracks patterns in volumes based on a customizable set of rules. Fraudzuki can also read image text, a feature which is currently proving very valuable to detect contact information (such as email addresses or phone numbers) in images posted by malicious users.