Coin Cloud Wallet
Crypto's conversion problem isn't a UI problem. It's a trust problem. We fixed the trust first.
Project Details
Coin Cloud operated one of the largest Bitcoin ATM networks in the US, alongside a companion mobile wallet. Two surfaces. Two form factors. Millions of users making real financial transactions.
The product needed a foundation before it could scale. It needed flows that handled money movement with clarity and compliance. And it needed an onboarding experience that could turn a first-time crypto user into a confident one.
This work covers all of it.
Role: Staff Product Designer - End-to-End Product
Client: Coin Cloud
Industry: Fintech / Crypto
Scope: Design System · Mobile UX · Transaction Flows · Compliance Patterns · Research & Analytics · Mixpanel
The Problem
Crypto has a confidence gap.
Not a technology gap. Not a product gap. A gap between what a first-time user believes about moving money digitally and what the product is actually asking them to do.
Coin Cloud operated one of the largest Bitcoin ATM networks in the US, alongside a companion mobile wallet. Two surfaces. Two completely different form factors. Millions of real users making real financial decisions — most of them for the first time.
The product had traction. What it didn't have was a foundation. Design decisions had accumulated without a shared system behind them. Flows that should have felt effortless created friction exactly where confidence mattered most. And the data confirmed it — users were dropping out of the transaction flow at a rate that represented real, measurable revenue leaving the table.
The brief was to fix the product. The real job was to fix the trust.
The Foundation
Most teams reach for the design system last, after the screens are built and the inconsistencies are already in production. We built it first.
That wasn't a philosophical choice. It was a practical one. Two surfaces, mobile wallet and physical ATM kiosk, had to feel like one product to a user who was moving between them with real money. If the design language drifted between surfaces, so did the user's confidence. And in a financial product, lost confidence doesn't recover.
The system established a semantic token architecture across color, typography, spacing, and components. Every token named for its function, not its appearance, not because that's best practice, but because names that describe meaning survive redesigns. Names that describe appearance don't.
Color — Primary, Secondary, Surface, Blacks and Grays, Error, and Success states. The naming convention means a developer and a designer can talk about the same decision without translation.
Typography — Inter, scaled across nine levels from xs to Display. Every level carries a specific weight relationship designed to move a user through a high-stakes financial decision without creating friction at the moments that matter most.
Components — Buttons, toggles, inputs, icons, markers, and crypto asset symbols. Full state coverage. Documented for handoff. Nothing aspirational — everything in the build.
Onboarding
The first transaction a user completes isn't the buy. It's the account.
Onboarding in a crypto product is where trust is either established or forfeited. Most products treat it as a checklist. We treated it as the product's opening argument.
Two complete flows, designed from launch screen to account approval.
Default Flow — Launch, Welcome, Phone/Email Login, Verification, Account Creation, Email Verification, PIN, Face ID, User Verification, Account Approved.
Every step designed. Every error state accounted for. Every moment where a user might hesitate given a reason to continue.
Product Not Available Flow — When Coin Cloud's services weren't available in a user's region, the experience needed to say so clearly without making the user feel like they'd wasted their time. Designing this edge case wasn't optional. Every user who hits a wall and gets a graceful response is a user who comes back when the product is available. Every user who hits a wall and gets a dead end doesn't.
Locations
The physical network was only valuable if users could find it.
Coin Cloud's ATM network was a competitive moat. Thousands of locations. Real cash in, real crypto out. A physical presence no pure-digital competitor could replicate.
None of that mattered if the app couldn't get a user to the nearest machine without friction.
The Locations feature was designed end to end: map view, list view, favorites with empty state, a filter system with applied and unapplied states, a location detail sheet with directions and support access, and search with suggested results.
Every edge case has a screen. The empty favorites state. The zero search results state. The location detail when a machine is temporarily out of service. These aren't polish — they're the moments where a user decides whether the product is reliable or not.
Transactions
Abandonment at 25% isn't a design problem. It's a diagnosis.
When we pulled the Mixpanel funnel data, the drop-off wasn't random. It was concentrated, specific steps, specific moments where users hit something they didn't understand or didn't trust and left.
That's actually useful information. Random abandonment means the problem is everywhere. Concentrated abandonment means the problem has an address.
We ran 50+ usability sessions alongside the funnel analysis, qualitative insight to explain what the data was showing, quantitative signals to measure whether the fixes worked. The combination is what separated guessing from knowing.
The redesigned transaction flow reduced abandonment by 25% and increased conversion by 25%. Those numbers came from real users making real decisions with real money, not a prototype test.
Buy Flow — Asset selection, payment method, personal use acknowledgment, amount entry with quick-select options, minimum purchase state, currency selection, order preview, confirmation. Every step pressure-tested against the sessions. Every friction point addressed before it hit production.
Sell Flow — The mirror of the buy flow, designed with the same rigor. Asset selection, amount entry with available balance displayed, sell preview, currency selection, complete.
The compliance gate deserves its own paragraph. The personal use acknowledgment modal is a legally required disclosure. It cannot be removed. It cannot be buried. The design challenge is making it honest without making it a roadblock, clear enough to satisfy a regulator, frictionless enough that a user who has nothing to hide doesn't feel accused of something.
Getting that balance right is what regulated product design actually looks like. It's not about making compliance invisible. It's about making it proportionate.
Settings
Nobody talks about Settings until it breaks something.
In a financial product, Settings is where users manage their money, their security, and their ongoing relationship with the product. It's not a secondary surface. It's where trust gets maintained after the transaction is done.
Designed across eight areas: payment methods, default currency, notification preferences, security settings, statements and history, agreements and policies, app theme, and support.
Every section consistently structured. Every option clearly labeled. No buried toggles. No ambiguous language around security settings. If a user is trying to remove a payment method or export a statement, the product should make that straightforward, because a product that makes it hard to leave, or hard to understand what's happening with your money, is a product users stop trusting.
Dashboard
The empty state is the product's first impression for every new user.
The dashboard brought it all together: wallet balances across multiple assets, top movers, and a direct ATM finder CTA connecting the mobile experience to the physical network.
The populated state is the obvious design problem. The empty state is the one most teams skip.
We didn't. Because the first time a new user completes onboarding and lands on the dashboard, the product has one screen to show them what their life looks like when this works. An empty state that looks like an error is an opportunity wasted. An empty state that shows the shape of what's coming is an onboarding moment in disguise.
In Closing
Two surfaces. One foundation. A 25% reduction in abandonment and a 25% lift in conversion, not from a redesign of the visual layer, but from understanding exactly where confidence broke down and building the system that restored it.
Design systems, compliance gates, and trust signals aren't constraints. They're the craft.
The trust was always the product. The UI was just how we made it visible.