Skip to content
Shipped

Commanding attention in a market designed to ignore us

Lira Boost grew from zero to $1.29M AUM in 30 days. Content carried a lot of the work.

Assets under mgmt
$1.29M
Peak in TRY
52.8M
Timeframe
30 days
Project Lira Boost
Company OKX
Role Content Designer
Launch 0 → 1
Market Turkey
01

What we shipped, and what I owned

Problem

Turkish users treated OKX as a daytime trading account. Local inflation pushed banks to offer huge overnight interest rates, so users deposited every morning to trade and pulled funds out every night to chase that interest. The daily ritual was a retention leak.

Solution

We launched Lira Boost. Idle lira sweeps to a partner bank overnight and lands back on OKX by morning. Users earn while they sleep. We keep the liquidity.

What I owned

I worked alongside my design partner on this 0→1 launch. I owned onboarding, entry points, activity management, and external comms like notifications and emails. I also covered core Earn flows when my partner was out.

02

Three hidden rules that could break trust

Earnings didn't start until 4:30 PM. Redemptions took 90 minutes. Early withdrawal forfeited all interest. If users only learned this at the wrong moment, they'd assume the worst.

Earnings start at 4:30 PM, not at deposit.

In the upfront flow

"Your lira is all set to earn tonight. Check back after 4:30 PM."

Funds return within 90 minutes.

In the upfront flow

"Your funds should be available within 90 minutes."

Pull out early and you lose today's interest.

At the decision point

"Redeem now for your funds sooner, or wait 12 hours to keep today's interest."

Pushing back

A 90-minute lock reads as a trap when you only learn about it at withdrawal. I argued for surfacing each rule in the upfront flow instead of hiding it in error states. I also audited the deposit flow and pushed for simpler logic.

The team prioritized it and shipped the changes as a fast follow.

Designing for skepticism

Turkish users carry real economic memory. Three currency crises in a decade will do that. They scan for proof, not promises.

Concept testing showed two distinct readers on the exact same screens. Aspirational users hunted for yield caps. Cautious users hunted for failure states. Every screen had to answer both questions at once.

03

Content systems

Three formats. One rule for each. Never mixed on a single screen.

01 Turkish Lira

Onboarding and education

Full name reads respectful, not jargony

02 TRY

Portfolios and balances

Matches global trading conventions, scans fast

03

Notifications and tight UI

Saves space, reads at a glance

Rule Never mix

On any screen

Mixed formats compound doubt

04

Notifications

Three examples of the same notification, rewritten for the person reading it.

Transcreating content

We weren't translating. We were transcreating.

Word-for-word translation doesn't move skeptical users. Every notification, every label, every error state was rewritten for meaning, tone, and trust in the Turkish context. Same product, rebuilt in the voice users actually hear at 9:41 AM.

$1.29M
AUM in 30 days
52.8M TRY
Peak on platform
Both
Reader types stayed overnight