How OKX's designers study apps that aren't available in their market

Scarlet Huang designs the part of OKX where users manage their money: accounts, sub-accounts, the security flows of the user center, across both app and web. At a crypto exchange, that surface carries unusual weight.

“Our users have serious expectations. They care more about account security, because it’s about money.”

Like every designer at a global exchange, she studies the competition. Unlike most, she often can’t even open it. Crypto and fintech apps are licensed market by market, so the products she most needs to benchmark are frequently unavailable where she works.

“I cannot get on [a lot of these apps] easily. I cannot even go on the app and use it as a normal user. So I got all my screens from Mobbin to reference.”

When she opens Mobbin

Mobbin isn’t a default at OKX the way Figma is. It lives with a specific cohort of designers doing reference-heavy work, and Scarlet is one of them. She didn’t bring it in; it was already in her team’s stack when she arrived. Before OKX she’d been in IT consultancy building internal platforms, where there was nothing to reference and no need for a library like this. The job changed when the work became public-facing and competitive.

She opens it maybe once or twice a month, for what she calls serious research, and it’s usually one of two situations. The planned kind: a deep-research project with time to study the market, like a sub-account architecture or a security flow, which sends her to direct competitors, or for cross-industry problems, anywhere relevant. And the unplanned kind, which surprises no one.

“We try to keep daily work well planned, but there’s always an urgent task, or we suddenly need to change something. If it’s urgent, it’s easy to just go to Mobbin and see similar patterns instead of doing user testing from scratch.”

In that mode it’s triage more than research: a way to a sensible design fast when nobody has time to test from zero.

Looking past the same few apps

The freshest thing on her plate when we spoke was an AI avatar feature, still at proposal stage, which lets users customize avatars and enhance photos. References weren’t scarce. Quite the opposite: every obvious app had already converged on the same way of doing it.

“AI is still relatively new, so the common patterns on TikTok and the big social apps look the same. I went to Mobbin to find hidden gems, apps I hadn’t used before, with newer flows for changing your photos or profile. That gave me more ideas for the whole interaction.”

That’s the trouble with a popular pattern. The more apps adopt it, the less it has left to teach. The good ideas were in the apps she’d never opened, the ones you’d never find on your own.

What Mobbin couldn’t find

We have a habit of asking customers what didn’t work, which is either diligence or mild masochism. Either way, we asked Scarlet for a bad example too.

“I was revamping a feature where users create API keys in batch, more than one at a time. I tried to search for batch-creation patterns, but the keyword doesn’t work that well in Mobbin. It’s a design pattern, but not a common one, especially on app side. It’s a very enterprise-platform thing.”

She also pointed out that some of her closest competitors aren’t in our catalog yet. We furiously took notes. It’s exactly the feedback shaping where Mobbin’s headed: deeper finance-industry coverage, and an MCP that surfaces patterns instead of just matching keywords.

There was an MCP mix-up, too. Scarlet had wanted a Mobbin MCP for a while, didn’t realize we had one, and had tried a competitor’s instead.

“I tried a competitor of yours because they had an MCP. But it didn’t work well. I could only use it in Claude Code, not Cursor, and it just gave me a link to their site instead of showing me the results. It’s like extra steps.”

Ours launched a few weeks before we spoke. She’ll try it and report back.

What if Mobbin went away tomorrow?

We asked.

“It would be very painful. Going through all those apps would cost extra time. And with crypto and fintech regulated differently in every market, a lot of them I can’t even access.”

That’s the reality of designing in a regulated industry: the products you most need to learn from are often the ones you can’t get to.

Scarlet still can’t open many of the apps she benchmarks. With Mobbin, she doesn’t have to.

“If it’s urgent, it’s easy to just go to Mobbin and see similar patterns instead of doing user testing from scratch.”