I redesigned investor onboarding from a six-document, 45-minute paper trail into one 12-minute adaptive flow — collapsed jurisdiction branching, replaced wizards with a state machine, surfaced compliance status in real time. Shipped to Canaccord, Raymond James, and Sprott across $2B+ in regulated transactions.
Founded the design practice and ran it solo for three years. As the company grew to 30+, design's role got built inside it — every investment defended in numbers the business already tracked: drop-off rates, audit risk, dealer churn. Owned the design system, the jurisdiction-rules engine, and the compliance dashboard. Kept the in-house team lean — 2 full-time designers (Prague, Serbia) + external designers brought in as projects required. By the time Markette acquired the company, design was reviewing every major release. Time to move on.
Cross-functional collaboration with engineering (backend implementation), compliance officers (regulatory interpretation), and the institutional dealers (user research observation on their floors) informed the work — but the design ownership was mine.
60% of investors abandoned at document upload.
Six to eight documents in one session, no save-and-resume. The same flow for a California accredited investor and a UK retail investor. Compliance officers had no dashboard view of who was stuck or why. The platform was processing private placements for top institutional dealers, but losing investors before they finished signing up.
Investors didn't abandon because the process was hard. They abandoned because it gave them no information, no progress, and no way to pause.
I designed the research. The dealers ran it. Canaccord, Raymond James, and Sprott had daily access to real investors, so I set the protocol and their teams observed sessions on their floors. Their notes came back weekly. Platform drop-off analytics ran in parallel and confirmed the same patterns across every jurisdiction.
Cut investor onboarding from 45 minutes to 12 — without losing compliance across IIROC, Reg CF, MiFID, ISA, and FINTRAC.
Regulators think in states. Users think in steps. The product's job is to translate between the two.
Residence, tax residence, and citizenship — three things investors kept getting wrong on a single form. Split them into three screens, one question each. Small enough to leave and come back to.
Document upload as a single guided capture step instead of an 8-document dump. Viewfinder framing, clear progress, one document at a time. Investors stopped uploading the wrong file.
Source-of-funds was free-text before — investors wrote "savings", "work", "inheritance", and compliance had to interpret each one. Replaced it with a category pick. Evidence is only asked for when the category needs it.
Ran for 11 years across 20+ regulatory jurisdictions.
NIGO (not-in-good-order) submissions dropped enough that ECM teams stopped chasing corrections and started closing deals. The compliance team grew slowly while investor volume kept climbing — automation handled the high-confidence cases, humans owned the exceptions.
The same engine eventually shipped to two sister products: Lending (Singapore, Mexico, UK, US deployments) and Donations & Rewards (InvestYYC + Alberta BoostR with Calgary Arts Development and ATB Financial).
Progressive disclosure reduced cognitive load in a domain where cognitive load directly costs money. The jurisdiction-aware rules engine is the feature compliance officers cite most often. Save-and-resume was the single biggest behavioral change — it turned a high-stakes single session into a returnable process. Trulioo and Onfido integration scaled identity verification across 20+ jurisdictions without a 20× headcount increase on the compliance team.
Mobile-first from day one. The platform was retrofitted for mobile years after desktop launch, creating layout debt that took multiple sprints to address. Earlier investor-side usability testing too — research relied heavily on compliance officer observation, and spending equal time with the end investors (not just the admins watching them) would have surfaced friction sooner.
Three takeaways. Compliance isn't a UX blocker. It's the brief. Wizards break when paths split. State machines don't. The people closest to users — dealers, compliance officers — know more than any lab. I use all three now.
