Revolut just made the card number invisible. If your fraud model still leans on PAN-based scoring, you should be sweating.

This isn’t a UI tweak. It’s a structural shift in how payment data is generated and shared. When the card number disappears from the user journey, the traditional risk signals attached to it—BIN, velocity, expiry patterns—lose their predictive edge.

For PSPs and acquirers, the commercial consequence is direct: your fraud models are only as good as the data they consume. If that data is being replaced by tokens, device fingerprints, and behavioural patterns, your legacy risk stack is already outdated.

The real issue is that most fraud teams are still optimising for a signal that’s being deprecated. The shift to behavioural and device-based signals isn’t a choice anymore. It’s a requirement. According to Frederic Yves Michel NOEL, that means new data partnerships, richer session-level intelligence, and AI models that can actually learn from non-card data streams.

For acquirers, the margin pressure is obvious. The cost of fraud isn’t just chargebacks—it’s the false declines you’ll generate by clinging to a signal that’s fading.

For PSPs already moving away from PAN-centric risk, what’s been the hardest part: sourcing the behavioural data or retraining the models?

#Fraud #Tokenization #CardPayments #Fintech #RiskManagement #PSP #PaymentSecurity