A DEX rebuilt from chaos to a released product
Three designers had passed through before us. What was left was 100+ screens that disagreed with each other, and a release date that had already been promised. We aligned the visual language, standardised the system, and shipped the features that made trading worth repeating.
Audit, redesign, new flows and features, product growth.
Competitor analysis began the day after the first call.
Audited, standardised and brought under one system.
After a new order type reached the trading flow.
The crypto market does not forgive visual noise.
The product had to be ready for release fast. Inside it were three previous designers' conflicting approaches, fragmented components, and an interface that read as a collection of local fixes rather than one trading system.
- Mixed styles and patterns from screen to screen — three design languages competing inside one product.
- 100+ screens of inconsistent quality and logic, with no shared source of truth.
- Friction sitting in exactly the flows that earn: trading, settings, wallet interactions.
- Nothing that gave a trader a reason to come back and trade more often.
- A refreshed visual layer — without an expensive foundational refactor.
- A system flexible enough that development would not have to wait on design.
- Parity with the market on core functionality, and a real edge in a few places.
- More executed trading actions, because platform revenue is fees.
A redesign that never made development wait.
The brief was not "redraw everything from scratch". It was: bring the product into a coherent state, fast. We aligned the visual language, standardised the patterns that repeat, and let the team keep shipping while we did it.
An overloaded screen with fragmented blocks and weak visual hierarchy. The key numbers get lost among secondary data and are not easy to scan at a glance — the interface reads as heavy, and assembled from local fixes.
A cleaner, more structured screen built around the numbers a trader actually acts on. The hierarchy is explicit, the analytics that were buried are now visible, and the daily workflow is readable in one pass.
Chart, holder analytics, a live trades feed and the order panel at once, none of them shouting. Hierarchy is carried by weight and spacing rather than by colour — in a market this red and green, colour was already spoken for. Everything a trader checks before pressing Buy sits within one glance of the button.
Total value, the default asset and the four wallet actions come before the list, so funding and moving assets remain one-tap tasks.
Preset amounts, slippage and fees in one strip, buy above and sell below. The panel floats over any screen, so a decision never costs a navigation.
Avatar, name and referral-ready username are resolved in one calm step. The flow gives a new trader a recognisable profile without turning setup into a separate journey.
Protocols and quote assets stay visual and scannable, while deeper metrics remain grouped below. A dense market search becomes a sequence of quick, reversible choices.
The feature that moved revenue.
A DEX earns on trading fees. So the product question was not "what else can we add" — it was: how do we increase the number of executed actions without making the interface heavier? We introduced a new order type: conditional execution.
One extra line in a flow traders already knew. Conditions sit beside take-profit and stop-loss, so the new capability is discovered in place instead of taught.
Conditions a trader would otherwise sit and watch for. Fee thresholds, developer sell-offs, market-cap triggers — the waiting is handed to the platform.
Execution brings the trader back. The notification is the return trigger: the order fired, and there is a reason to open the product again.
growth in trading volume after the new order types shipped — and with it, fee revenue. At launch, most competitors still did not offer conditional orders. Tested on a focus group of 100 traders, February 2026
The screens that decide whether people trust you with money.
In a self-custody product the reputational risk is not on the dashboard. It sits in the flows where someone can permanently lose access to their funds — and those are the screens teams leave until last. We started there.
SELF-CUSTODY Your funds. Your keys. Your responsibility.
Twenty more decisions like these.
Three of them are above. The release pace meant reading the market quickly and shipping only what moves activation, retention and everyday value — this is what the other twenty covered.
Chart, order book, holder analytics and the order panel on one screen.
Market, limit, DCA and the conditional execution that changed the numbers.
Total value, realised and unrealised PNL, positions and trade history.
New tokens across every launch platform, on a single page.
What the market is talking about, ranked and time-stamped.
Tagged wallets, live activity and alerts without leaving the terminal.
Opinion leaders scored on reach, call quality and risk.
PNL cards a trader builds, styles and posts from inside the product.
Fee-back mechanics wired into the flows that already earn.
What it gave the product.
Instead of a set of disconnected screens, the platform gained a direction: clearer for new users, more convincing for active traders, and easier for the team that has to keep releasing.
More reasons to trade, a stronger position against competitors, and a traceable contribution to revenue through the trading mechanics themselves.
A scalable design system and standardised flows: features ship faster, and changing one thing stops meaning changing it in nine places.
Faster trading and faster scanning, new capability through conditional orders, and a product that finally holds up on a phone.
What we actually did.
UX audit & friction map
Competitor analysis from day one, then a pass over 100+ screens to find where the product contradicted itself and where the earning flows leaked.
Product redesign
One visual language across the platform, rebuilt around scanning speed — with the release schedule untouched.
Design system
1000+ elements covering every state and scenario, so the team could ship features without re-deciding the basics each time.
Feature & flow design
20+ features worked through end to end, including the conditional order type that changed how much the platform earns.