
Crypto app UX mistakes, and how to fix each one
Imran KhanNovember 19, 202615 min read

table of contents
Most crypto apps don't lose people to a bug. They lose them to a screen that asked for too much, said too little or sounded scared. The fixes are mostly not technical, which is the good news.
This guide lists the mistakes we'd look for first, in the order a new person meets them: the door, the money, the waiting, the warnings, the home screen, the words, and the way the work gets done. For each one you get how to spot it, a real example where we could find one, and what to do instead. At the end there's a 30-minute audit you can run this week.
One thing up front. We're a design studio. We haven't designed a crypto wallet or an exchange. We have audited and redesigned a money app, Dino, designed a Web3 product, Reputable, and designed the website template Finsaro for finance software companies. We use them for what they show and rely on linked sources for the rest. Every outside page here was read on 19 November 2026. Nothing here is legal or financial advice.
The short version
- Don't ask for the hard thing first. A seed phrase, a wallet install or a network choice before the person has seen any value is the most common way to lose them.
- Show the cost before the button. Fees and limits belong on the screen where the decision happens, with a number.
- Every action needs an answer. Sent, pending, failed, done. Silence reads as "broken" or "stolen".
- Save strong warnings for strong risks. Constant alarms train people to click through all of them.
- The home screen answers one question. How much do I have, and what can I do next?
- Label every number. Whose money, which currency, what it's worth in the person's own money.
- Use plain words, with the jargon one tap away.
- Test with someone who isn't a crypto person. Your team can't see these mistakes any more.
How we picked these
We read the pages that rank for this search: Artkai, Purrweb, Masterly's wallet teardown, Brandson and TechCrunch's Coinbase teardown, plus a few more. Four of those five are agencies that sell design, so we treat their claims as opinions. We kept a mistake only when two or more sources agree, or when we could check it ourselves on a real product or a primary page. We left out every percentage that came without a source we could open.
Mistakes at the door
1. A seed phrase before the person has seen anything
Write down twelve words, keep them safe, never lose them. That's often the first thing a new person sees. Brandson calls it "often the worst possible introduction", and Masterly's teardown calls it settled technology debt.
A paper at CHI 2025, Of Secrets and Seedphrases, surveyed 643 cryptocurrency users. Only 43.4% could correctly recognise an image of a seed phrase. Asked what you'd need to move a wallet to a new device, 52% picked "username and password", and 47% picked the recovery seed phrase. Among those who could recognise a seed phrase, around 58% believed they could choose and reset it, the way you would a password. We read those numbers in the paper itself (the PDF on the author's site).
Recognising an image isn't the same as understanding it, and 643 people aren't the whole market. Still, a screen many users can't place is a poor first screen.
How to spot it: your first-run flow shows the seed phrase before the person has done anything they came for.
What to do instead: let the person get to something useful first, and bring the recovery step in when they have something to protect. Explain it in the words people use, not the protocol's. We go through the options, and their trade-offs, in our Web3 onboarding guide.
2. A "Connect Wallet" wall
A person wants to try your product. They tap Connect Wallet and see a grid of wallets they've never heard of. Brandson calls this the wallet connect maze, and Purrweb lists gating the product behind a wallet connection as one of the three highest-impact DeFi mistakes in its FAQ. Purrweb attaches a 30 to 60% figure to it. We can't check that number, so we don't use it.
How to spot it: a logged-out visitor can't see a single real screen or number.
What to do instead: show what the product does first. If a wallet is needed, say why in one plain sentence before the grid, and put the most likely choice first.
3. A network choice nobody can judge
Which exchange? Which blockchain? Which token? Masterly found a wallet whose deposit flow asked exactly those three questions with no explanation of what each choice meant for fees, speed or arrival. Purrweb lists multi-chain fragmentation as a main challenge, and the studio Caliper lists how a wallet shows one token across several networks among the things it examines in its wallet reviews.
Our advice: a network is plumbing. Most people shouldn't have to choose one, and if they must, the screen should say what happens if they pick wrong.
How to spot it: a first deposit takes three or more choices the person can't judge.
What to do instead: pick a default for them, show the one detail that matters (will this arrive?) and tuck the rest behind an "advanced" control.
Mistakes about money
4. The fee that shows up at the last second
The ethereum.org gas page explains that the total fee is a base fee plus a priority fee, the tip, and that the fee "is paid regardless of whether a transaction succeeds or fails". A person can pay and get nothing. If your screen shows the fee late, in an unfamiliar unit, you're asking them to accept that without seeing it.
Brandson calls gas fees a "surprise tax". Artkai lists unclear financial statements and fee information as a main problem too. Markswebb's Telegram wallet comparison found the send form gave no fee or processing time before the amount was entered, so people tried to send their whole balance and the transfer failed. Its fix: show the fee in the send form before the amount.
How to spot it: the fee first appears on the confirm screen, in gwei or in a token the person didn't choose.
What to do instead: show an estimate early, in the person's own currency, in words ("about this much, can change"). Say plainly that a failed transaction can still cost something. Where your design lets you sponsor or hide gas, say who pays.
5. Buttons that don't say what they do, and limits with no number
TechCrunch's Coinbase teardown dates from 2020, but two of its points still hold. First, a prominent "Get Started" button that, in the writers' words, "just puts bitcoins in your basket". Their fix: don't lie in your call-to-action. Second, card payments labelled "Invest small amounts" with no figure. "How much is a small amount?" they ask. Their fix: if you have limits, state them.
Both are small, and a nervous person can read either as "something's being hidden".
How to spot it: read every button aloud. If a button says "Get started" and does something else, or a limit has an adjective instead of a number, it's this mistake.
What to do instead: name the action ("Buy bitcoin"), name the number ("up to this much per day"), and keep the promise.
6. Numbers nobody can place
Markswebb, a UX studio, compared two ways of cashing out crypto, one through Telegram's wallet. Two findings stand out. The history didn't show what a received amount was worth in fiat, or the rate at that moment. And the currency list wasn't sorted by balance, so a coin with zero balance could sit above the one holding all the money. A person could think the wallet was empty.
We saw the same family of mistake at close range on Dino. Its home screen opened on a group's balance, with "your balance" under it, a personal balance below that, and a list of groups each carrying two more figures. Nothing said whose money was whose. The audit's rule became: every number says whose money it is, and every screen has one obvious thing to do next. Dino is a payments app, not a crypto app, but the mistake is the same.
How to spot it: point at any number on any screen and ask "whose, in what currency, worth what to me?" If you can't answer in a second, a new user can't either.
What to do instead: label every balance (total, available, staked, pending are four different numbers), show a fiat value beside each token amount, and sort lists by what the person actually holds.
Mistakes about waiting and failing
7. Silence after the tap
TechCrunch's second fail is about feedback. After adding a card, the person taps "Add Card" and lands back on the homepage with no notice of success or failure. "All feedback needs to be thought of and built," the writers say.
Onchain it's worse, because waiting is normal. On the ethereum.org transactions page a sent transaction is broadcast and added to a pool of pending transactions before anyone includes it. Artkai quotes an app review from a person whose transaction showed as pending after the chain had confirmed it many times over, and who concluded the app had "hijacked" their wallet. Artkai's advice is to mark an action on screen as soon as it's likely to succeed, give the status a link, and wind it back with an error if it fails.
How to spot it: tap every button in your app and count the ones that give no visible answer within a second.
What to do instead: design each state: submitted, waiting, done, failed. Say what the person can do while they wait, and give the failed state a next step, not just a red box.
8. The error that blames the user, or the wrong thing
In April 2025 Decrypt reported that Coinbase had reportedly fixed an interface error in which a failed password login was labelled "2-step verification failed" in the account activity log. The report, which credited BleepingComputer's Lawrence Abrams, says people took that to mean an attacker had the right password and was stopped only at the second step. Some spent hours checking for malware and changing passwords. The log reportedly now says "Password attempt failed". Coinbase didn't respond to Decrypt's request for comment.
One wrong label caused real fear and real work.
How to spot it: list every message your app can show and ask, for each one, "is this exactly true?"
What to do instead: say what happened in the plainest true words, and say what to do next.
Mistakes about warnings
9. Alarms everywhere
Crypto is dangerous, and apps know it, so they warn about everything. Brandson's name for the result is warning blindness: when every screen has red text, people click through all of them, including the one that matters.
Artkai makes the opposite complaint: vague warnings, with words like "funds" that mean different things to different people. Both can be true in the same app.
How to spot it: screenshot every warning in your app. If more than a handful look equally loud, none of them is.
What to do instead: rank your warnings. Keep the loudest for the moments you can't undo (sending to a new address, approving broad access). Make the routine ones quiet. Help people check ("this address is new to you") instead of only scaring them. Our Web3 UX guide covers approvals and signing in detail.
Mistakes on the home screen
10. A home screen that serves the team
Masterly's teardown says the Coinbase app's home screen competes with itself, with a balance, buy and sell buttons, discovery content, rotating promotional banners and price tickers, and says it is designed for engagement more than clarity. That's one studio's view, and it's a studio that sells design. But try the test yourself on your own app: what's the first thing you see, and is it the thing a new person came for?
Brandson's version is trading terminals for people who don't trade: dense data and glowing charts for the many people who came to do one simple thing. Coinbase's chief executive Brian Armstrong himself said in August 2023, as CryptoSlate reported, that the user experience in its main app was "broken" for dapps and layer-2 activity, something he said the company's Onchain Summer campaign was exposing.
How to spot it: ask five colleagues to name the one thing the home screen is for. If you get five answers, it's for none of them.
What to do instead: let the home screen answer two questions: how much do I have, and what can I do next? Put everything else one step away. Dino's audit gave us the same lesson in a different product. The redesign opens on the person's own balance with three actions under it, and group money appears as a card you pay from, with your share printed on it. Two kinds of balance are never in the same place without a label.
11. A zero-balance dead end
A new wallet starts at zero. Masterly points out that this is the first screen a person sees after setup and that one wallet it tested shows a bare zero balance with no hint of what to do. Another showed "Add crypto to get started", which it called functional.
How to spot it: make a brand-new account and screenshot the first screen. Is there one clear next step?
What to do instead: treat the empty state as the last onboarding screen. Give it one next action, in plain words.
Mistakes in the words
12. Jargon as the interface
Slippage tolerance, approve spending cap, bridge, wrap. Artkai's list of problems starts with arcane terminology, and Brandson gives a swap: "set how much price change you'll accept" instead of "slippage tolerance". Purrweb's top DeFi mistake is showing raw contract language instead of a plain outcome like "You will receive about this much".
How to spot it: copy all the labels on one screen into a document and hand it to someone outside crypto. Mark what they can't explain.
What to do instead: say what will happen, in the person's terms, and explain a technical word inline the first time. Keep the exact term one tap away for people who want it.
13. Colour as the only signal
This one none of the ranking pages mention. Price changes in crypto apps are almost always green and red. The W3C's guidance on use of colour (WCAG success criterion 1.4.1) says colour shouldn't be the only visual way of conveying information. A person who can't tell red from green gets nothing from a chart that relies on them.
How to spot it: put your screen in greyscale. Can you still tell what went up and what went down?
What to do instead: add a sign, an arrow or a word beside the colour. It costs almost nothing. Our crypto app design guide covers numbers and contrast in more detail.
Mistakes in how the work gets done
14. Third-party data nobody checked
TechCrunch's teardown found a bank account shown as the wrong bank, because the data came from a third party. The writers' point: when it's wrong, your users blame you, not the provider. Crypto apps pull in third-party prices, token names and logos all day.
How to spot it: list every piece of outside data on your screens and open each one on a real device.
What to do instead: check how it displays, and put someone in charge of looking again. TechCrunch's guest expert says his team solved it by checking each entry and hard-coding the response.
15. Only the happy path gets designed
Rejected signatures, dropped connections, wrong networks, a transaction that never lands.
How to spot it: ask to see the screen for "the person rejected the request in their wallet". If nobody can show you one, it doesn't exist.
What to do instead: our advice is to draw the failure screens at the same time as the success screen, each with a next step. A team that skips them finds them in production.
16. Testing only with people who already know
Your team can't see jargon, because to them it's just words. Brandson's advice, and ours: watch someone who isn't crypto-native try to finish your main action without help.
How to spot it: everyone who has tested the product works in crypto.
What to do instead: put a phone in the hands of a person from outside crypto, say nothing, and write down every pause and question. Each one is a mistake you've found.
A 30-minute audit you can run this week
This is how we start on an existing product, cut down. On Dino we mapped 16 screens of the existing app, audited the five that carried the most confusion, and turned every note into a design decision. That ended in 21 redesigned screens. The same shape works for you in half an hour.
The form below is from Timerack, a time and attendance product we designed. It isn't a money app, but it shows the plain, labelled kind of form a confirmation step needs.
- Pick the one thing a new person comes to do. Buy, send, swap, save, look around.
- Make a fresh account and do it. Write down every screen, every choice and every word you had to guess.
- Mark each screen against the 16 mistakes above. A tick for each one you find.
- Hand a phone to someone outside crypto and ask them to do the same task. Don't help. Note where they stop.
- Fix in this order: whatever stops people (a wall, a seed phrase, a missing state), then whatever scares them (warnings, wrong labels, hidden costs), then whatever slows them (jargon, clutter).
By the end you'll have a list, which is more useful than another round of redesign.
What we've done and what we haven't
We haven't designed a crypto wallet or an exchange. If you need someone who has shipped a licensed exchange, ask for that specifically from any studio you talk to, ours included. What we do is audit and redesign money screens, and design products for new markets: Dino for a group payments app, Reputable for Web3 in dark and light, and Finsaro for a finance software website template (its dashboard screens are stand-ins, made for a product that doesn't exist yet). Our mobile app design page says we audit what's live, find where people drop out and rework those paths first. For how that's priced, see our pricing page.
FAQ
The ones the ranking guides and our own checks agree on: a seed phrase or wallet wall before any value, hidden fees, silent or unclear transaction states, constant warnings, jargon, and dense screens built for traders. We cover sixteen above.
Actions can't be undone, the words are new, and the numbers change by the second. A mistake costs real money, so people are more nervous than in most apps. That makes plain words and honest status more important, not less.
Some is. Brandson's answer is that most of what people see isn't, and we agree with that as an opinion. A network choice or a gas unit is often a design choice, not a law of nature.
Not by itself. Our advice is to slow people down only on the step that can't be undone, and to keep the rest easy. A scary screen on every step protects nobody.
Whatever stops people before they see value: the wallet wall, the seed phrase, the missing next step. Then what scares them. Then what slows them.
Pick one main action and count how many people who start it finish it, step by step. Watch a few people try without help. We don't quote benchmark percentages, because the ones we found had no source we could open.
Ours depends on the number of screens. For a rough guide, our mobile app design page says most app projects take four to eight weeks from start to finish. An audit alone is shorter. Book a call and we'll tell you what yours needs.
If you have a crypto, fintech or money app and want its screens audited, book a call. Our mobile app design team starts by mapping what's live and finding where people stop.
About the author
Imran Khan, Founder & Creative Head
15 years of industry experienceFormer Head of Design at Day TranslationsFormer Creative Lead at SupersideWork at Superside for brands including Meta, Red Bull and Ray-Ban

related articles
Book a call
Pick a time that suits you
Get in touch
Let’s build something amazing together
Let’s build somethingamazing together
For product, interface and web-system work that needs clear direction and careful execution.




