The Problem Space
Age-restricted brands have long struggled with online verification. Signing up to purchase age-restricted products online is long and tedious, thanks to the law. ID uploads are slow and easy to spoof, and failing at any point in this process takes you back to the very beginning. It's not uncommon for a user to get confused, drop off, or get denied, even when they’re doing everything right.
The space was full of compliance-heavy tools with poor UX — and high abandonment to match.
Video 1: JUUL's 15-step verification process (yes, 15)
A Smarter Way to Verify Age
The market was ripe for disruption, and our team set out to build something different: a lightweight biometric system that used facial recognition and liveness detection to confirm a user’s age — no IDs, no friction. The underlying tech was solid, but the challenge was how to turn it into a product people would actually trust and use.
I was brought on shortly after the inception of Arcoscan to lead the UX and UI from the ground up. This meant turning a complex, backend-heavy system into an experience that felt simple, clear, and human.
Video 2: Testing Arcoscan's tech
Image 1: Arcoscan's instruction/error list
Exploring the UIÂ Possibilities
Before designing anything, I spent time looking at how other biometric tools handled similar flows — tools like CLEAR at airports, ID.me, and even onboarding flows in fintech apps with selfie verification.
Using this as inspiration along with others, I came up with UI components for each of the instructions we had to resolve during an age verification check.
Image 1: Apple's Face ID setup
Image 2: CLEAR's airport stalls
Image 3: Basic idea for the Arcoscan UI
Image 4: Telpo POS facial scan
Build for Trust, Not Just Speed
After researching the market, it was clear that the first challenge was clarity with instructions. Users needed to grant camera access, center their face, perform a subtle gesture, and wait for a backend result — all in a few seconds.
I leaned into neutral, minimal UI — not only to avoid overwhelming the user, but to keep focus on the task. I prototyped motion cues for gestures, rewrote onboarding copy for calmness and clarity, and built error states that guided users back into the flow instead of locking them out.
We built an early MVP through Webflow and embedded an iframe into the demo page to test it. Even during early internal testing, we saw how easy it was for people to fail — and how much design could help them succeed.
Image 5: Translating the tech into UI components
Image 6: Camera mockup of Arcoscan UI
Image 7: Building a responsive screen to test our tech (in Webflow)
The JUUL Pilot
Months later, we landed a 3-month pilot with JUUL — a validation moment that also became a stress test. The timing was perfect as we were looking for strategic partners to use our technology, and JUUL was looking to replace their existing age verification process. After proving our tech could achieve a >95% confidence score, we just had to clean up the UI and provide a new UI layout that could match the JUUL brand. I created the new layout for the JUUL workflow.
Working closely with JUUL’s product and engineering team, I redesigned key parts of the experience:
- Introduced a “Hold Still” visual cue that guided users into proper face alignment
- Replaced static instructions with animated gesture guidance
- Rewrote error handling and added real-time feedback to recover from failed scans
Image 8:Â Designing a JUUL-specific age verification flow
Image 9: Initial results post-implementation
Early Signals
Our first batch of users onboarding to JUUL with the Arcoscan technology were abandoning their experience at record highs, which was not a good sign for us.
Looking at the data more closely, we were able to figure out poor scan alignment, tight timeouts, and vague instructions were making it hard to complete a scan successfully.
We also worked through bugs in the backend — including “image too blurry” and unexpected timeout failures — translating technical issues into better design, better pacing, and clearer expectations.
The Experience Got Better — Fast
Over the course of the pilot, user success improved dramatically. The abandonment rate dropped steadily as we refined the flow — especially on mobile — and JUUL’s own data confirmed the experience was smoother and more reliable than where we started.
By the end of the 3-month pilot, we had reduced total abandonment (including failed verifications) from 51% down to 20%. While that was a huge improvement, it wasn’t quite enough to secure a long-term partnership.
Ultimately, JUUL chose to move forward with a more established vendor that had already reached industry-leading benchmarks. As a small startup, we didn’t have the scale or maturity yet — but the pilot validated that our design held up under pressure and that we were solving real problems for real users.
Image 10: Virtual usability test with a user
Reflection
This project reminded me that great UX isn’t just about polish — it’s about clarity in chaos. Designing for Arcoscan meant working through technical unknowns, aligning product with legal nuance, and earning trust one interaction at a time.
I didn’t just design screens — I helped shape a product from scratch, guided it through a live pilot, and iterated fast based on real human behavior. Even without a long-term launch, I’m proud of what we built.
As we fine-tuned the experience, we even removed the head silhouette — making the flow cleaner and closer in feel to Apple’s Face ID. Our goal is still to partner with a major alcohol or tobacco brand, and proving Arcoscan’s potential in the wild.
As we improved performance, we started to wonder if we were sacrificing clarity for speed — and whether users had enough time to understand what was happening. The search continues.
Video 3: Testing the new "Face ID" UX
more Projects
Promontory (Coming Soon)
LinkedIn for investors.
Square
Email design system.