To Catch A Fraudster: Building Products for the Dynamic World of Online Fraud
summary
Claims that fraud morphs weekly get backed by examples: Flipkart-era systems see collusion move from repeated addresses to fictional map points to codewords like cricketer names, forcing a shift from regex-based filters to maps + ML for collusion detection. A concrete product move is a rapid UI hack rolled out over a weekend that shows small purchase options (15–20 rupees) to raise tipping and reduce abuse — a weekend "jugaad" UI change that measurably lifts take rates while longer-term models train.
AI-written summary · about
chapters (14)
- 0:00Introduction
- 2:20Indian consumer landscape
- 5:28Customer motivations with Danda
- 7:12Balancing motivations in fraud prevention
- 9:20Five whys for user motivations
- 10:45Customer empathy in India
- 11:07Uber Eats India case study
- 13:13Jugaad in product management
- 14:35Myth of failing fast
- 16:40Adapting to fraudster workarounds
- 18:47Rules for Jugaad
- 19:29Value and adoption in India
- 20:53CCDVTP framework
- 22:16Moving on from failures
related talks
45mAnti-Fraud in Reverse and the Use of ML Methods in It
13mHarnessing Machine Learning to Combat E-commerce Fraud
28mFortifying Digital Trust: Next-Gen Strategies in Fraud Prevention
37mFraud prevention in the age of online privacy
16mLeveraging ML to Create a Reliable Anti-Fraud System
39mFraud Detection: Using ML to Identify and Manage Fraudulent Activities
39mA New Era of Fraud: What Role Can Cyber Play?
40mSherlock and Watson Fight Deep Social Engineering