10.5 Case Studies¶
Case 1: Large Indian Private Bank — Digital Onboarding Transformation¶
| Aspect | Before | After |
|---|---|---|
| KYC method | Branch-based with Aadhaar biometric | Digital eKYC with face liveness |
| Onboarding time | 3-5 days | 8 minutes average |
| Monthly new accounts | 45,000 | 180,000 |
| Cost per account | ₹350 ($4.20) | ₹25 ($0.30) |
| Fraud rate | 0.4% | 0.03% |
| Liveness approach | N/A | Hybrid (passive + active) |
| Key challenge | High device diversity (budget Android phones) | Tuned for low-end devices, progressive quality guidance |
Case 2: European Neobank — Deepfake Attack Incident¶
| Timeline | Event |
|---|---|
| Month 1 | Neobank launches with passive-only liveness (vendor A) |
| Month 4 | Fraud team detects 23 accounts opened with AI-generated faces (StyleGAN) |
| Month 5 | Investigation reveals additional 150+ synthetic identity accounts |
| Month 6 | Switched to vendor B with deepfake detection; added active challenges for high-risk |
| Month 8 | Synthetic identity fraud dropped to zero; overall fraud rate down 94% |
| Lesson | Passive-only liveness without deepfake detection is insufficient for digital banking |
Case 3: Southeast Asian Digital Bank — Accessibility-Driven Design¶
| Challenge | Solution | Result |
|---|---|---|
| Diverse user base including elderly with limited tech literacy | Passive-first approach with simple guidance overlay | 89% first-attempt completion rate |
| Multiple languages (Thai, Malay, Bahasa, English) | Localized challenge prompts and guidance | 12% improvement in completion for non-English speakers |
| Hot climate causing face moisture/reflection issues | Adaptive specular highlight handling | 40% reduction in outdoor false rejections |
| Low-end devices (< $100 phones) | Lightweight on-device model + server-side verification | Works on devices with 2GB RAM |
Next: Future Trends →