Risk Management Show Why MORE Data Doesn't Mean BETTER Risk Models
13 snips
Feb 23, 2026 Artem Lalaiants, CEO of RiskSeal and expert in fraud prevention and real-time credit signals. He explains why simply adding more data fails to improve models. He highlights orthogonal digital signals like subscription behavior, real-time scoring for instant underwriting, and strategies to score the underbanked across markets. Practical advice on combining bureau and alternative signals for better coverage.
AI Snips
Chapters
Transcript
Episode notes
Why Credit Bureaus Fail For International Fintechs
- Traditional credit bureaus are regionally limited and often fail for fintechs expanding across markets.
- Artem Lalaiants explains fintechs need real-time, portable signals because bureau coverage and predictiveness vary widely by geography.
Add Orthogonal Signals Not More Bureau Data
- Seek orthogonal data rather than just more of the same to improve model uplift.
- Artem Lalaiants recommends adding signals like digital behavior and paid subscriptions that are predictive yet independent of bureau data.
Real Time Signals Enable Instant Digital Decisions
- Speed and low-friction UX demand real-time signals; traditional bureaus can be too slow and request heavy documentation.
- RiskSeal delivers sub-second collections of digital signals to enable instant decisions and reduced friction.