Why Onboarding Friction Is a Design Choice, Not a Failure
Every extra step at sign-up loses a customer. Every skipped step risks the wrong one staying.
Tamás Simon, CCAS
Compliance Officer
3 min read
In February 2025, the DOJ announced a $504 million settlement with OKX and its Aux Cayes Fintech entity: no FinCEN registration, no effective AML program, and, per the DOJ's own release, employees who on certain occasions advised customers how to provide false information to get past the company's KYC process (a separate, customer-produced instructional video, not an OKX employee, walked users through using a VPN to conceal their US presence, something OKX allowed to circulate rather than authoring itself). Ten months later, OFAC settled with Exodus, a wallet provider, for $3.1M over 254+ sanctions violations, the same underlying pattern in a different corner of the industry: staff recommending customers use a VPN to disguise their real location. Two different business models, two different regulators, the same eighteen months, the same root cause.
The point of friction is where it should sit.
Standard identification at onboarding, enhanced due diligence for higher-risk customers, and a full look-through for corporate accounts are three levels of friction applied to three levels of assumed risk. Should friction scale with what a customer has actually declared, or with what the system infers before they've transacted at all? Most mature programs resolve this with three broad risk tiers, low, medium, high, assigned at onboarding from inferred factors, then revised as real transaction history accumulates.
Verification is a moment, monitoring is a process.
A customer who clears onboarding today doesn't stay cleared forever. The standard answer layers two triggers: event-driven rescreening every time a relevant list updates, plus a separate periodic risk-rating refresh for higher-tier customers specifically.
What this means.
OKX and Exodus are eighteen months apart and in different segments of the industry, but they're the same story: friction treated as a UX problem to route around, by the people closest to it, rather than a control to enforce. A well-placed step tied to a documented risk tier is worth defending internally, not just designing well.




