Okto · Redesign
Concept study
a product design case study

From the multi‑chain maze to one‑tap DeFi.

Rethinking Okto — a self‑custody Web3 trading wallet — so that first‑time users could swap, bridge and earn across chains without ever meeting a seed phrase, a gas token, or the word “chain”.

RoleLead Product Designer Scope0→1 redesign · IA · flows · UI · system Team1 design lead, 2 PMs, 6 eng SurfaceiOS & Android Year14 wks · concept study
ETH SOL APT ? ? BEFORE redesign CHAIN ABSTRACTION $2,418.50 All chains · one balance USDC → HYPE Swap · 1 tap AFTER

Fig. 00 · The thesis

The whole project in one frame: collapse three chains, a seed phrase and a gas token into a single, legible action. Every artifact below is hand-drawn and downloadable as vector (SVG) or high-res PNG.

01 — At a glance

A powerful wallet that beginners couldn’t get past the front door.

Okto was already a genuinely capable self-custody wallet: cross-chain swaps, bridging, staking, perps. But its power was its problem. Every concept that makes Web3 work — keys, chains, gas, bridges — was also a wall a newcomer had to climb before doing anything useful.

My remit was not to add features. It was to make the existing power survivable for a first-timer without diluting it for the pros — and to do it on Okto’s own terms, the line written on its homepage: “Secure, yet simple.”

What I owned

  • Problem framing & the redesign strategy
  • Generative & evaluative research
  • Information architecture & core flows
  • End-to-end UI + the lo-fi system

How I partnered

  • Two PMs on scope, sequencing & metrics
  • Six engineers on feasibility of chain abstraction
  • Security on the MPC / keyless model
  • Support, mined for the real failure points
where it actually broke

02 — The problem

Seven steps stood between “I’m in” and a first trade. Most people quit at step three.

I mapped the real path a new user walked, end to end, against how they felt at each step. The line didn’t dip gently — it fell off a cliff the moment we asked someone to be their own bank.

Three moments did the damage: writing down a 12-word seed phrase, understanding which chain they were on, and discovering they needed a separate gas token to move their own money. None of these are features users wanted. They’re plumbing we were making them learn.

FRUSTRATION LINE Install Create wallet Back up 12 words Fund wallet Pick a chain Bridge for gas First swap ! ! ! “Write 12 words by hand? Really?” “Why do I need ETH to move my USDC?” “It just… failed.”

Fig. 01 · Emotional journey, before

Reconstructed from onboarding analytics + 9 first-run interviews. Red nodes sit below the frustration line — the points where intent to continue collapsed. The cliff is the seed-phrase step.

03 — The old screens

Three walls — one per screen.

Each of the three drop-off moments had a screen behind it. Drawn lo-fi on purpose: at this stage the argument is about structure and burden, not pixels.

Back up your wallet Save these 12 words in order ⚠ Never share these words 123456789101112 I’ve stored them safely Continue Hand-copying12 words isthe #1 drop-off Self-custody= 100% on you.Heavy for afirst-timer.

Before · Onboarding

A correct security pattern, fatal for activation. We front-loaded the scariest concept in self-custody.

Wallet Ethereum ▾ $612.40 (Ethereum only) SendReceiveSwap Assets ETH0.21 ETH$540 USDC72 USDC$72 Switch chain to see Solana → Per-chainbalance —assets hidein a dropdown “Where didmy Solanago?” 😕

Before · Home

The wallet exposed our data model. Users had to know what a chain was just to find their own money.

Swap From USDCon Solana100 To HYPEon Hyperliquid≈ ? ✕ Not enough ETH for gas Step 1 of 4: Bridge to Ethereum Network fee ~$3.80 in ETH Slippage tolerance 0.5% Swap Needs aseparate gastoken to moveyour own money Bridge → wait→ approve →swap = 4 steps

Before · Swap

The killer: you needed the right gas token, on the right chain, before you could touch your own funds. Many gave up here.

listen before you draw

04 — Research & insight

I stopped asking how to teach Web3, and started asking why we were teaching it at all.

The early instinct on the team was “better onboarding” — smarter tooltips, a friendlier seed-phrase tutorial. Research killed that idea. The problem wasn’t that people couldn’t learn the concepts. It’s that they shouldn’t have to.

Method9 first-run interviews MethodOnboarding funnel analytics Method320 support tickets, coded MethodTeardown · 5 wallets Method5 usability tests (proto)

Four insights that set the whole direction

01

Nobody asked for self-custody

People wanted the assets. Custody was a tax they tolerated, not a feature they valued. So the security had to become invisible, not celebrated.

02

“Chain” is an implementation detail

Users reasoned in coins and rupees, never in networks. Every time the UI said “select chain,” we were exposing our own database schema to a beginner.

03

Gas is the cruelest gotcha

Being locked out of your own funds by a token you didn’t know you needed doesn’t read as “secure.” It reads as broken. This was the single biggest trust-killer.

04

Warnings don’t build trust

Big red “never share this!” banners raised anxiety without raising safety. Trust came from clarity and reversibility — knowing what would happen and being able to undo.

One sentence from a 27-year-old in our study became the brief: “I just want to buy what my friends are buying — why is this harder than UPI?” If a payments app could be that simple, a wallet had no excuse.

Aditya · 27 First-time investor · UPI-native · Tier-2 India “Why is this harder than UPI?” Goals Buy & swap coins in a few tapsOne app for everythingNever lose funds to a mistake Fears Forgetting a secret phraseSending to the wrong placeHidden fees & scams Comfort with crypto plumbing — low

Fig. 02 · Primary persona

Not a demographic prop — a decision filter. Every design choice below was checked against Aditya: would this make sense to someone who has never heard the word “blockchain”?

in their own words

What changed for users

Not excitement. Just the absence of confusion.

"Wait… that's it?"

"I don't need to install MetaMask first?"

Curious Investor · First onboarding session

"I'm already inside."

"I was in the app before I realised my wallet was ready."

Curious Investor · Wallet creation flow

"No fear this time."

"I wasn't scared I'd lose some secret phrase and lock myself out."

Curious Investor · Post-session reflection

the turn

05 — The reframe

If the plumbing is the problem, hide the plumbing.

This is where the project changed shape. We weren’t going to teach people chains and gas more gently — we were going to take them off the screen entirely, using the same chain-abstraction tech Okto had under the hood but had been exposing as controls.

North star

Hide the chain. Kill the seed phrase.
Never block a user from their own money.

Three principles fell out of that, and I used them to settle every argument for the rest of the project:

1

Custody, made invisible

An MPC, keyless wallet created from a social login. Recovery feels like signing back in — not guarding a secret you can permanently lose.

2

One balance, every chain

Assets unified into a single number and list. The network becomes a quiet badge we resolve, never a question we ask the user.

3

Intent over mechanics

People say what they want — swap, send, earn. We handle routing, bridging and gas underneath, then tell them plainly what happened.

BEFORE — DEEP, BRANCHING, GATED AFTER — FLAT, INTENT-BASED Open app Choose chain Per-chain home Find token Enough gas? Bridge Swap no yes loop back ↻ Open app Unified home Swap Send Earn Done

Fig. 03 · Information architecture

Same capabilities, restructured. The left tree forces a chain decision and a gas detour before any value. The right tree is two taps deep and routes the mechanics out of sight.

06 — Exploration

Eight minutes, six homescreens, one survivor.

Before committing, I sketched the home concept fast and cheap to pressure-test the “one balance” idea against alternatives. Keeping it ugly kept the team arguing about the model, not the styling.

balance + actions tabbed by action single feed card stack search-first unified + 3 actions ★ this one

Fig. 04 · Crazy-eights, home

The winner put one balance up top and three verbs — Swap, Send, Earn — within thumb reach. Search and chain filters survived as secondary, never as gates.

same power, no homework

07 — The redesign

The wallet now starts working before you’ve learned a single new word.

The three walls, gone. Onboarding is a login you already have. Home is one balance. A cross-chain swap is one tap, with gas quietly paid in whatever token you’re holding. The capability never shrank — it just stopped asking permission.

Create your wallet Takes about 15 seconds Continue with Google Continue with Apple Secured by MPC No seed phrase to write down A login youalready know— ~15s Keys split viaMPC. Nothingto lose or leak.

After · Onboarding

The seed phrase is gone. MPC keyshares mean recovery is a re-login, so the scariest step simply doesn’t exist anymore.

Wallet All chains $2,418.50 across all your chains SwapSendEarn Assets ETHEthereum$1,640 HYPEHyperliquid$520 USDCSolana$258 One balance,every chain Chain is a tinybadge here —never a gate

After · Home

One number, one list. Chains live as small badges on each asset, so they inform without ever blocking. Earn is promoted to a primary verb.

Swap From USDCSolana100 To HYPEHyperliquid≈ 4.12 Best route · Solana → Hyperliquid ✓ Network fee paid in USDC No bridging. No native gas token. Swap 1 tap · any chain Gas auto-paidin the tokenyou already hold 1 tap. No bridge,no gas token,no chain pick.

After · Swap

Gas Exchange pays the network fee from the token being swapped. Routing and bridging happen underneath a single button. The 4-step gauntlet became one tap.

08 — Before / after

The same task, re-counted.

Not a restyle — a removal. The clearest way to see the redesign is to count what a first-time user no longer has to do.

? BEFORE $2,418Swap AFTER WHAT THE USER NO LONGER DOES 12 words0 5 chains exposed1 balance 4-step swap1 tap gas token requiredgasless

Fig. 05 · The delta

Every row is a concept subtracted from the new-user’s path. The power to do all of it is still there for those who want it — it’s just no longer mandatory reading.

Log in · Google Wallet ready Pick token Confirm · gas auto Done no cliffs this time ↗

Fig. 06 · Emotional journey, after

The same axis as Fig. 01, redrawn. Five steps, all above the line, trending up. Nothing here requires a tutorial.

honest about the numbers

09 — Impact

What we’d expect this to move.

A note I’d make to any reviewer: this is a concept study, not a shipped product. So these are directional targets — modeled from the real onboarding funnel and wallet benchmarks, and framed as hypotheses the beta would test, not results I’m claiming.

2.4×
modeled lift in activation — wallet created → first on-chain action (≈28% → ≈68%)
<2min
time to first trade, down from ~11 minutes of setup
+34pts
onboarding completion rate, end to end
−60%
support tickets tagged “gas” or “seed phrase”
FIRST-ACTION ACTIVATION · MODELED Before ~28% After ~68% 2.4× TIME TO FIRST TRADE ~11 min < 2 min

Fig. 07 · Modeled outcome

The headline hypothesis: removing the seed phrase, chain selection and gas wall roughly doubles the share of new users who reach a first action, and collapses setup time from minutes to seconds.

what stuck with me

10 — Reflection

What I’d carry into the next one.

  • Abstraction is a design decision, not just an engineering one. The hardest, highest-leverage calls in this project were about what to hide — and earning the trust to hide it.
  • Defaults are the product. The overwhelming majority of users never open settings. The “all chains, one balance” default isn’t a convenience — for most people it is the wallet.
  • Don’t soften a bad step — delete it. We didn’t write a friendlier seed-phrase tutorial. We removed the need for one. Subtraction beat polish every time.
  • Honesty under the hood earns more trust than warnings on top. Plainly telling someone “fee paid in USDC” did more for confidence than any red “never share this” banner ever did.

If this were going to beta next

I’d pressure-test the unhappy paths the happy path hides — failed routes, partial fills, slippage on thin pairs — and design the honest, recoverable version of each. I’d add an opt-in power mode for users who genuinely want manual chain and gas control. And I’d treat every number on this page as a hypothesis to confirm with real instrumentation, not a trophy.