Onboard V3: From crypto wallet to global money app

Process

How we worked

1. Sprint. A one-week design sprint to rethink the core experience from first principles: what Onboard is for, and how every product should fit together.

2. Prototypes. I built interactive Figma prototypes for every core journey: onboarding, the dashboard (two competing versions), exploring accounts, and six ways to move money, from USD-to-bank to bank-to-crypto-wallet.

3. Testing while engineering built. The backend took months, so we didn't wait. Over one week, Henry (PM) and I ran moderated remote tests with 9 high-value transacting users, measuring task success, time and satisfaction for each journey.

4. Handoff. Annotated flows covered every state, including empty, pending, failed and refunded.

Each decision below shows what testing changed.

Decision 1

Organise around money, not technology

V2 was built crypto-first: every action on home (Add, Transfer, Swap) and every prompt beneath them led to crypto, with the card stacked on top as a secondary account. To become a money app, cash had to be the centre, not another card in the stack. So home had to follow how people actually think: my money, and what I want to do with it.

The ambition went beyond a redesign. With Project Unity, Onboard was becoming a regulated money app for a broader market: one place to hold, spend and move money, with crypto there for those who want it. If people were going to trust us with all of that, home couldn't feel like a toolbox. That set up the question:

How might we make Onboard the last finance app you'd ever download, without front-loading every money tool at once?

Our answer was to rebuild the app around three groups:

• Cash. Everyday balances like USD and EUR
• Cards. Virtual cards for spending and subscriptions
• Crypto. Trading and the self-custody DeFi wallet, still there for those who want it

Home shows one total balance, with each account a swipe away and two actions up front: Add money and Send money. Send sits at the centre of the bottom nav, always one tap away.

V3 home at launch: every account in one carousel, five swipes wide

What testing changed

V3.1: from complete to focused

V3's home was comprehensive: every account a swipe away, each with its own set of tools. That turned out to be the problem. Swiping through five accounts gave people too many choices, and we weren't getting them to the tools they cared about fast enough. Most of the time, people aren't thinking about where they put their money. They just know they have it.

We had designed for completeness when people needed focus. V3.1 gave home a sharp one: cash. Home now shows what you have and what you can spend, with Card and Crypto in their own tabs, so people can get straight on with the task in hand. Cash is also our preferred path to activation, because it's where Onboard monetises, so focusing home on it pointed new users at the actions that matter most.

How home evolved: V2 crypto-first, V3 organised by money, V3.1 cash first

Decision 2: the core flow

One send flow for every kind of money

In V2, every product had its own send flow. In V3 there's one. You pick where the money is going, and the app chooses the rails behind the scenes: bank transfer, mobile money or stablecoin.

That first choice came straight from testing. Our early prototype grouped send options by asset type, and "Send from any cash account" was misunderstood by almost every participant. So we grouped by destination instead.

The flow at launch (sending dollars to a bank account in Kenya)

1. Send. One question: To bank or mobile money, or To a crypto wallet?
2. Send from. Defaults to your USD balance. Change lets you pick another.
3. Amount. As you type, you see the fee (often free), the total we'll convert, the rate, what the recipient gets in KES, and "Arrives in 10 mins".
4. Recipient. Pick a saved recipient or add a new one.
5. Review. Bank, account number and an expandable cost breakdown, before anything moves.
6. Sent. "We'll notify you when this transaction is completed."

Then we reordered it. In testing, every participant expected to choose who they're paying before how much, the way a bank works. V3 launched with the amount first. In a later release we moved the recipient to the start: recipient → amount → review → confirm.

We also renamed vague buttons. Participants read "Continue" as "Send now" and hesitated, so each step's button says what happens next: Next → Review → Confirm.

The principle: no surprises. The cost, the rate and the arrival time all show up before you commit, and so do your transfer limits, because unclear limits made cross-border users nervous in testing.

One send flow for every balance, from choosing where the money goes to a confident close

After you press send

Cross-border money takes minutes, not milliseconds, and the gap between "sent" and "arrived" is where trust breaks. So every transfer gets a live tracker:

• Home shows an Ongoing transaction card. Tap it to track.
• The receipt is a three-step timeline: Cash sent → Conversion ongoing → Funds received.
• If something goes wrong, the fix is on the same screen. A failed payment offers Retry. One we can't process offers Refund, and the status updates once it's refunded.

Every state has a next step: pending, success, retry, refund, refunded

Decision 3: activation

From sign-up to active user

Activation was our weakest number: only about 1 in 8 new users ever moved money. Signing up was easy. The steps after it (verifying, funding, spending) were where people stalled. So I designed activation into the product itself.

Finish setup in time, get $20

A time-boxed checklist that turns three scattered first steps into one goal: verify your identity, add $100, and make three card purchases of $20, before the deadline, and get $20.

• It lives on home. A setup card with a live countdown shows the next step and a Start button, so the first thing a new user sees is what to do next.
• It gets out of the way. In testing, people disliked cards they couldn't clear, so the checklist can be dismissed.
• Every step has one action. Each item carries its own button (Verify me, Add money, Open cards), so no one has to hunt for where to go.
• Waiting is a state, not a dead end. While KYC is processing, the step says View details instead of Start, and the screen sets expectations: usually under 15 minutes.
• Failure is recoverable. If verification fails, people see why, how many retries are left, and tips to pass next time.
• Progress is visible. Steps tick off on a timeline, and the reward card says exactly when the $20 is paid.
• Only real spending counts. Purchases must be $20 or more, and cash-like transactions (gambling, gift cards, transfers) don't count, so the reward drives genuine use.

Step 1, verify identity: the setup card, the checklist, and every KYC state

Step 2, add money: the next step surfaced on home and in the checklist

Step 3, card purchases: straight to cards, and all three steps done in time

Refer and earn carries the same idea into growth: $10 for each friend, paid once they spend up to $100 on their USD card, so referrals reward active users, not empty sign-ups.

What changed: in the months after the checklist launched, the share of new users starting verification rose from 37% (June) to 54% (September), and reached 74% by January 2026. Activation rose from 11% to 13.6% in Q4.

Activation built into the core: the setup reward, refer-and-earn, and instant account details

What didn't work (at first)

Our power users pushed back

The single send flow broke the habits of our crypto-native users. They relied on legacy rails like Direct accounts, which we'd deliberately moved deeper into the app to keep the default clean. To them, it felt like we'd traded power for ease.

What we did: instead of rolling back the unified flow, I brought their favourite workflows and tools back inside the Crypto tab. Power users got their shortcuts back where they already expected crypto to live. Everyone else kept the simple, unified experience, untouched.

What I learned: simplifying for the majority is half the job. The other half is giving experts a short, findable way back to the control they rely on.