On average 50,000 people a day were trying to open a Mara account, with a $1M World Cup push about to double that. A 27 step onboarding gauntlet wasted almost 90% of them, half of which came with a bill. Paid to get them, and paid to lose them.





email



phone
details



KYC · biggest drop




PIN
biometrics
homeHead of Design and Product Owner, running three cross-functional teams. The whole wallet, mobile, for Africa. I led design, took on engineering and support alongside it, and presented roadmap and growth to investors, led by Coinbase. 2022-24.
funnel analysis
Onboarding asked for everything up front: email and SMS both, Country three different ways, personal details, a PIN and biometrics before there was anything to protect. Each step, another reason to quit. And KYC was the toughest to pass, losing more users than every other step combined. Without a strong reason to do it, most users found this step simply too much hassle for a product they hadn't even tried yet.
business case
With every check, every permission, every data point, front-loaded. The flow was cumbersome and heavy for users to go through. And when they got to KYC and started a new, nested flow, that weight compounded and naturally many gave up. But the service charged for the API calls regardless of outcome. So the bill came in, even when users dropped out.














solution
Once onboarded, all users had been through KYC. But only 15% ever transacted, which made KYC useless for the other 85%, that's $340,000 wasted. AML rules require verification before money moves, not at signup. By moving KYC to the transaction flow, it only triggers on someone with clear intent: a strong context for the prompt, and a real reward to complete it. That made useful checks climb to 98%, and reduced the cost dramatically.







iteration
With ample room still to improve, I turned to the remaining leaks. Email became both the unique identifier and when needed, the communication channel, saving another $100k monthly in SMS verification fees. Personal details became a single ask, with Country folded in. And PIN provided security, pushing the biometrics prompt to a first successful login.
With just three unique actions, onboarding became effortless and user acquisition rose to 2.1M new users a month.








audit
The naira was devaluing abruptly, losing 70% of its value across this period, and Nigerians were moving close to $60B of it into crypto, as a store of value.
Our onboarding was clearing 88% by then, adding 2M users a month, and only 15% of them ever traded. In a market that hungry, the clear problem was our app. It was a mess. Without a single guiding thread.
Home had a secondary navigation splitting the screen. Send, a flagship feature, actually covered the main action behind the keyboard.
messy home
Hard send
design system
It was obvious fixing just transactions wouldn’t solve our low rates. The money had to be able to come in in order to transact, so the leak would likely move to Deposits. Once that was in order, users would likely feel trapped if Withdrawal didn’t work. And Send was really the one that needed more help. So it was clear, a full redesign was the only fix.
I reviewed every flow and mapped all actions to a single repeatable set of steps: action, preview, commit, outcome. This became the foundation of the app, and gave it a rhythm that made all flows instantly familiar.
impact
The new onboarding looked obvious in hindsight, stripped down and purpose-built. But the bloated version made sense too, once, to whoever added each piece. In isolation, even “add KYC to onboarding too” was a reasonable call. That's the trap: every step defensible on its own, but broken as a whole. As was every other flow.
What Mara lacked was someone who defined and held the structure, the system, not the individual flows. Someone who could make the whole product work as one. That's what I did.
Perfection is achieved, not when there is nothing more to add,
but when there is nothing left to take away.Antoine de Saint-Exupéry · Wind, Sand and Stars · 1939