Yellowcard

I owned design end to end across web, iOS, Android, and iPad, from zero to MVP and through ongoing iteration after launch. This case study covers the consumer-facing send and receive flows specifically.

Traditional remittance into and around Africa runs three to five days and costs seven to twelve percent in fees. Yellowcard could settle in minutes using stablecoins, but the people who needed that most had never touched crypto and had no reason to want to.

The challenge wasn’t technical, the rails worked. It was trust and comprehension: how do you get someone to bet their rent money on a system they don’t understand, built on infrastructure they’ve never heard of.

Yellow Card operates in a regulated financial space across multiple African markets, so design couldn’t outrun compliance. Every flow needed clear consent points and audit-ready receipts built in from the start, not added after the fact.

At the same time, the crypto market moves fast enough that waiting for exhaustive research wasn’t realistic. We designed the core happy path first, shipped it, then iterated against real usage rather than guessing every edge case upfront.

Senior product and UI/UX designer, working solo on direction with close input from engineering rather than as part of a larger design team. Consumer flow decisions, visual direction, and the design system all ran through one person.

That meant sitting close to engineering on what stablecoin mechanics actually needed to surface, settlement state, fee, recipient confirmation, and what could disappear entirely, wallets, gas, chain selection.

Nobody sends money home to feel clever about crypto. They want to know three things: how much, to whom, and did it arrive. Everything else, wallets, gas, settlement layers, is infrastructure, and infrastructure should stay out of sight.

We stripped every crypto-native term from the primary flows. A wallet became an account. A gas fee became a transfer cost, shown as one number instead of a line item to decode. A transaction became a transfer, full stop.

That speed gap, minutes instead of days, was the entire product, so the interface had to make it felt, not just stated in a marketing line.

Live transfer status, plain-language settlement confirmations, and timestamped receipts did that work. Not decoration, proof. Each one answers the question a worried sender actually has mid-transfer: is this working?

Financial trust is earned screen by screen, especially in a market that has already been burned by mobile money fraud and fly-by-night agents. A new product gets one shot to look legitimate before someone closes the tab.

Clear branding, explicit fee breakdowns, and receipts that read like something a bank would issue, not something a crypto exchange would, were the small, unglamorous decisions that did the actual trust-building.

Yellow Card processes more than $200M in monthly trading volume. The redesigned onboarding and transaction flows cut failed transactions for new users by 65%, and the 100+ component design system raised design throughput by roughly 40%. Working directly with DevOps brought confirmation times down 30% while holding 99.9% uptime.

Twelve major releases followed launch, each targeting a specific drop-off point seen in the data, lifting engagement 35% and retention 20% over that period.

The best compliment a fintech interface can get is that it’s boring. Boring means nobody had to think twice. Getting there meant saying no to a lot of things that would have made the product look more impressive and feel less usable.

If I were starting over, I’d push the explicit fee breakdown into the very first onboarding screen rather than the first transaction. Trust signals matter most before someone has committed, not after. That’s the instinct I carry into LOOP now.