Work / FinTech & Payments / Institutional Crypto / Custody Wallet

Case Study

Institutional Crypto Custody & Trading Workflows

Designing a secure custody and trading experience that lets institutional clients move, protect, approve, and monitor digital assets with confidence, without hiding the security that makes it safe.

My Role
Lead Product Designer
Responsibilities
Trust & Risk UX, Complex Workflows
Platform
Institutional Web Dashboard
Context
Institutional FinTech · Digital Assets
Year
2023
Illustration prompt: A futuristic 3D isometric illustration of an institutional digital-custody vault, with glowing indigo and silver vault mechanisms, floating blockchain nodes, and signature seals. Rendered as a clean illustration on a fully transparent background (no backdrop, scene, or color fill), so it displays cleanly on both light and dark page themes.

↑ Institutional trust, rendered as a UI: every approval visible, every signer accounted for.


The institutional gap in a retail-built product

As digital assets matured, our platform saw a wave of institutional clients, hedge funds, family offices, corporate treasuries, arrive to manage serious money. They were handed a retail-focused wallet interface built for a single user managing their own funds.

Institutional clients need to move millions in digital assets while reducing operational and counterparty risk. Retail wallets assume one person, one decision. Institutions require governance, and on a blockchain, a mistaken transfer is irreversible, making a "fat-finger" error a catastrophic, not merely embarrassing, risk.


What I owned on this project

I was the Lead Product Designer responsible for translating institutional governance requirements into an interface that felt secure without feeling hostile.

Governance Research

Mapped institutional roles and approval chains directly with traders, operations managers, and treasury leads.

Workflow Design

Designed the Maker/Checker approval engine, from initiation through policy check to dual sign-off.

Trust & Risk UX

Built the address-intelligence layer that translates raw blockchain data into human-readable risk context.

Security Collaboration

Partnered with cryptography specialists to explain MPC key sharding without overwhelming non-technical approvers.


Three governance gaps a retail wallet can't close

01

No segregation of duties

The same person who initiated a trade could also execute it, there was no maker/checker flow at all, a basic governance control institutional clients simply assumed would exist.

02

Approval obscurity

Treasurers couldn't easily verify what they were actually approving before signing a multi-sig request , they were being asked to trust a black box with real money inside it.

03

Address anxiety

Copy-pasting long hexadecimal strings created constant low-grade anxiety, driving frequent small "test" transactions that added friction without actually reducing risk.


The Opportunity

How might we give institutional teams the governance controls a treasury requires, while making every approval legible enough that a signer never has to sign on blind trust?

If we make maker/checker roles explicit, translate blockchain data into human-readable risk context, and give every approver a visible signature timeline, then institutional clients will approve transfers faster and with fewer errors, while gaining full audit readiness by default.


Mapping the governance structure, role by role

Before designing a single screen, I mapped exactly who does what inside an institutional client's treasury operation, because the UI had to mirror a governance structure that already existed on paper, not invent a new one.

Trader

Initiates trades and proposes transfers. Cannot execute without approval.

Operations Manager

Reviews daily limits, manages whitelists, acts as Level 1 approver.

Treasury Lead

Level 2 approver. Holds signing keys via hardware wallet.

Compliance Officer

Read-only. Exports audit logs, monitors AML flags.

Admin

Sets policy rules, e.g. "transfers over $1M need 3/5 signers."


What role-mapping made clear

1

"Approve" and "sign" are two different acts of trust

Approving is a logic check ("does this match policy?"). Signing is a cryptographic commitment. Conflating them in one button was itself a governance risk.

2

Compliance needed visibility, not veto power

Compliance officers wanted to see everything and trust the export, not intervene mid-transaction. Read-only access was a feature, not a limitation, once we framed it that way.

3

Anxiety about addresses was really anxiety about irreversibility

The hex string itself wasn't really the problem. Not knowing whether it was trustworthy was. Naming the destination solved more anxiety than any amount of copy-paste tooling could.


The rules we designed by, and why

Security without confusion

Make risk controls visible but intuitive, security should never become friction for its own sake.

Human-readable blockchain

Translate tx-hashes, gas fees, and hex addresses into plain-English context, always.

Approval clarity

Checkers must see exactly what they're signing, with anomalies clearly highlighted, not buried.

Auditability by default

Every action leaves a clear, exportable trail, compliance shouldn't have to ask for it.


Mapping the Maker/Checker workflow, step by step

Every transfer above policy threshold moves through the same four-stage governance chain, designed so each stage has one clear owner and one clear question to answer.

01 Maker Initiate

A trader drafts a transfer: for example, 100 BTC to a Binance hot wallet. At this stage the system captures intent only; nothing moves and no signature exists yet.

02 Risk Engine Policy Check

The system automatically flags the transfer against policy, over $1M triggers a 2-of-3 signature requirement. This step removes any ambiguity about how many approvals a given transfer needs before a human ever looks at it.

03 Checker 1 Approve (L1)

The Operations Manager verifies the destination address against the whitelist and reviews the logic of the request. This is a policy-and-context check, not yet a cryptographic commitment.

04 Checker 2 Sign (L2)

The Treasury Lead signs the payload using a hardware wallet (Ledger/Trezor). This is the only stage that produces an irreversible cryptographic commitment, deliberately kept separate from every earlier "approve" action.


The choices that shaped the outcome

Decision 01 · Separate "approve" and "sign" vs. one combined action

Tension: One combined button is faster to use, but blurs a policy check with an irreversible cryptographic commitment.

Choice & trade-off: We kept them strictly separate. The extra click was a deliberate cost, institutional clients told us directly they wanted friction at exactly this point, not less of it.

Decision 02 · Read-only compliance access vs. override rights

Tension: Some clients initially wanted compliance officers able to directly block a pending transaction.

Choice & trade-off: We kept compliance strictly read-only with full export rights, preserving a clean separation-of-duties model rather than adding a fourth actor into the live approval chain.

Decision 03 · Translated address labels vs. hex-only display

Tension: Security-minded stakeholders worried a friendly label might create false confidence versus raw, unambiguous hex.

Choice & trade-off: We showed the translated entity name as primary, with the raw address always one click away, trust without hiding the ground truth.


Institutional confidence at scale

Pending Approval: 100 BTC

To: Binance Hot Wallet (Whitelisted)

2/3 Signatures
Trader initiated
Ops L1 approved
Treasury L2 signature pending

↑ The Multi-Step Approval Center, every pending signature visible, nothing hidden behind a single "Send" button.

Destination Security

bc1qxy2kgdygjrsqtzq2n0yrf2493p83kkfjhx0wlh

Known Binance Wallet (whitelisted 30 days ago)

Risk Warning

Gas fees are currently unusually high (120 gwei). Consider delaying the transfer.

↑ Human-Readable Trust & Risk, the address-intelligence layer that turns a hex string into an entity you can recognize.


Design doesn't happen in isolation

Cryptography specialists: explaining MPC without overwhelming

I worked directly with the security team to design a "vault" metaphor for multi-party computation key sharding, so approvers could trust the mechanism without needing a cryptography degree to understand it.

Compliance: validating the read-only model

Compliance officers reviewed every export format before launch to confirm the audit trail actually satisfied the reporting formats their regulators expected, not just what was easy for engineering to build.

Institutional relationship managers: surfacing real client language

Relationship managers who spoke with treasury leads daily fed back exactly which phrases ("segregation of duties," "whitelist ageing") clients actually used, language that shaped the UI copy directly.


How we arrived at the solution

Methodology

Tree Testing

Tested the proposed information architecture for the multi-signature approval flow directly with institutional crypto traders.

↳ Outcome: Simplified navigation let users find pending approvals 3x faster.
Workshop

Security Protocol Co-Design

Worked directly with cryptography experts to design a UI that explains key sharding (MPC) without overwhelming the user.

↳ Outcome: A visual "vault" metaphor that built trust among institutional clients.

Illustration Placeholder

Prompt: A stylized illustration of a security-protocol co-design workshop, cryptography experts and a designer reviewing key-sharding diagrams on a glass wall, in indigo and silver accent tones. Rendered as a clean illustration on a fully transparent background (no backdrop, scene, or color fill), so it displays cleanly on both light and dark page themes.


Institution-grade UX drove adoption

-85%
Transfer errors
Via clear address resolution
2.4x
Faster approvals
Clear maker/checker timeline
94%
Confidence score
Measured in post-launch CSAT
100%
Audit readiness
Compliance exports passed review

Honest reflections from the process

What Worked

1

Mirroring an existing governance structure instead of inventing one

Institutional clients already had a maker/checker model on paper. The fastest path to trust was making the UI match what they already practiced, not proposing something novel.

2

Treating "approve" and "sign" as separate acts of trust

Once we split the logic check from the cryptographic commitment, the whole flow became explainable in a single sentence to any new approver, a strong sign the model was right.

What I'd Do Differently

1

Bring compliance into the workflow design earlier

Compliance validated export formats late in the process. Involving them while the maker/checker flow was still being sketched would have caught reporting-format requirements sooner.

2

Test the workflow under adversarial conditions, not just happy-path approvals

Most validation sessions walked through a clean approval. I'd want dedicated sessions rehearsing disputed or rejected transfers, since that's exactly where governance UX gets tested hardest in production.