Continue Reading
Not long ago, opening a business bank account or onboarding as a merchant to a payments platform could take three to five business days. Manual document review. A compliance analyst somewhere in a queue. The customer waiting. That friction had a real cost: drop-off rates during financial product onboarding ran between 40% and 70% depending on the product, with a significant portion of that drop-off happening during identity verification steps.
The companies that cracked real-time KYC didn’t just improve user experience. They changed the economics of the entire distribution model.
The visible cost of slow KYC was drop-off. But the hidden costs were larger. Every application that sat in review for 48 hours was a customer support ticket waiting to happen. It was also an opportunity for a competitor to complete onboarding first. In categories where conversion is won or lost on friction, a two-day delay in identity verification is a product-level problem, not an operations problem.
On the compliance side, manual review created its own issues. Human reviewers are inconsistent. Review queues created backlogs that led to approval pressure. And manual processes don’t scale well — adding compliance headcount has a linear relationship to cost that destroys margins as volume grows.
What automated identity verification actually looks like today is a layered stack. No single technology solves all of it. The combination that enables real-time or near-real-time decisions:
| Layer | Function | Key Signals |
|---|---|---|
| Document verification | Authenticate government IDs | Document validity, hologram, MRZ data |
| Biometric liveness | Confirm live person, match to document photo | Face match score, liveness detection |
| Database checks | Cross-reference identity data | SSN/DOB verification, credit header data |
| Sanctions/PEP screening | OFAC, UN, EU watchlists | Name matching, date of birth |
| Device/behavioral signals | Fraud pattern detection | Device fingerprint, IP, velocity checks |
When this stack operates in concert, the majority of clean applications can be approved in under 60 seconds. The edge cases — flagged names, document anomalies, high-risk jurisdictions — route to human review, but those represent a small fraction of total volume at a well-run compliance operation.
Consumer identity verification got to real-time relatively quickly. Business verification — KYB, or Know Your Business — is harder and the automation is less mature. The challenge is data fragmentation. Business ownership records, UBO information, and entity verification data are spread across state-level registries, third-party databases, and self-reported information that needs to be cross-referenced.
For simple LLCs and corporations in states with good registry data, automated KYB can get to a decision in minutes. For complex ownership structures, foreign entities, trusts, or businesses in states with poor public records, manual review is still necessary. The companies building KYB infrastructure are essentially aggregating and normalizing messy public data sources, and that’s a genuinely hard technical problem with a big market behind it.
Every basis point of onboarding conversion improvement in a high-volume financial product is worth meaningful revenue. We’ve seen portfolio companies recapture 15-20% of previously lost applicants by moving from manual to automated identity verification. At scale, that’s tens of millions of dollars in LTV.
Faster onboarding creates a tension with fraud prevention. Historically, the delay in manual review provided a natural fraud speed bump — not a reliable one, but some benefit. When you move to real-time approvals, the fraud models need to be significantly better to compensate.
The companies doing this well have invested in fraud signal networks. A device that was used in a fraudulent application at one company should trigger additional scrutiny at another. The economics of sharing that intelligence are complicated — companies are reluctant to share data that exposes their own fraud patterns — but consortium models and third-party fraud intelligence networks are making it more viable.
Automated KYC does not reduce regulatory responsibility. It changes how that responsibility is discharged. Financial institutions and their fintech partners remain responsible for the quality of their identity verification, regardless of whether a human or an algorithm made the decision. That means the automated systems need to be auditable, explainable, and demonstrably consistent with regulatory guidance.
The companies that are building compliance-grade automation — audit logs, decision explanations, model documentation — are the ones that will scale without running into regulatory friction. The ones treating identity verification as a pure UX optimization without the compliance infrastructure are building on sand.
We’re actively looking at infrastructure plays in this space. The identity verification and KYB market is still fragmented, and the integration complexity between verification vendors, core banking systems, and compliance workflows creates real opportunities for companies that can simplify the stack.
Building compliance infrastructure? We invest in this category.