File 01
2022
Making regulated onboarding feel like 4 minutes, not 4 days
Sector
Series-C fintech
Scale
~40M users · 3 EU markets
Role
Lead Product Designer
Team
2 PM · 6 Eng · 1 Research
Duration
7 months
Client
30-second read
For hiring teams
Moved identity verification from the first screen to the first deposit, without breaking KYC rules.
Cut 11 required fields to 4 by reusing data we already had a legal basis to hold.
Activation up 18–22%, KYC drop-off down about a third.
The flows shown are abstracted. Real screens, the research synthesis, and three rejected directions are in the full file.
Fig. 1 — Abstracted interface
Context
New customers had to upload a passport, a selfie, and proof of address before seeing a single screen of the product. Most quit at the passport step. Leadership wanted fewer steps. Compliance wanted zero risk. Both were right. That was the brief.
Constraints
—
Three markets, three regulators, and three definitions of “verified.”
—
Every screen needed legal sign-off, a two-week cycle.
—
The verification vendor contract was fixed for 18 months.
Decisions
Decision
When do we verify identity?
Upfront, before anything (status quo) · At a balance threshold
Rejected
At first deposit
Chosen
It was the only option all three regulators accepted in writing. People could explore the product with no money at risk, and we verified at the exact moment verification made sense to them.
Decision
What do we actually need to ask?
The full 11-field form · Progressive disclosure
Rejected
Prefill + 4 confirmations
Chosen
We already held most of this data under a legal basis. Asking again didn’t reduce risk. It only signalled distrust.
Decision
How do we explain the wait?
Spinner · “We’ll email you”
Rejected
Status timeline with honest times
Chosen
Most support tickets were a version of “is anything happening?” Showing the queue honestly beat hiding it.
Outcome
+18–22%
Activation across markets, over two quarters
~⅓
Less drop-off at the KYC step
~½
Fewer “where is my verification” tickets
Reflection
I spent too long trying to make one flow satisfy three regulators. Three thin variants on a shared spine would have saved six weeks.
Full file
File 01 — Locked
Regulator comparison matrix
Two rounds of usability findings
The rejected “verify later” direction
Real screens, before and after