Home / Work / Dandelion

Cross-border payments

Dandelion

Redesigning international transfers around trust, not rails.

Customers were choosing Wise and Remitly over their primary bank for international transfers, not because CommBank could not move money, but because competitors made the experience feel safer. I led product design across a new cross-border experience that made speed, status, and cost feel legible from the first screen through post-transfer tracking, aligning product, engineering, legal, and compliance around a journey customers could actually trust.

Role
Lead product designer
Team
1 designer, 1 PM, 8 engineers, accessibility, legal and compliance
100%
Task completion in usability testing
22
Additional corridors with route-specific requirements
3
Delivery methods: bank deposit, digital wallet, and cash pick-up
Dandelion transfer flow showing country, amount, corridor, and live FX conversion

The redesigned flow made cost, timing, and corridor logic visible before commitment instead of hiding uncertainty until the end.

Challenge

The problem was not the rail. It was the feeling of sending money into a void.

CommBank's legacy international transfer experience asked customers to tolerate too many unknowns at once. They were redirected out of the app, fees were disclosed late, corridor requirements appeared mid-flow, and once the money left the account there was no meaningful in-app visibility into what happened next.

That combination produced something rare: a banking feature people actively avoided using with their primary bank, even though the bank already owned the relationship and the funds.

What the old flow felt like

  • 3-5 business days with weak expectation setting
  • Unclear FX margins and late fee disclosure
  • Manual tracing instead of live status visibility
  • Redirects to NetBank for selected currencies

What competitors were proving

  • Near real-time settlement where possible
  • Clear pricing before commitment
  • Native mobile flows
  • Tracking that reduced anxiety after send
"
It just feels really clunky. I'm always convinced I'm going to stuff it up and my money's going to go off to some random stranger.
Participant feedback from transfer research

Insights

Customers cared about three questions: Will it arrive? When? What will it really cost?

Research reframed the brief. Customers did not need a lesson in payment rails. They needed the experience to make the infrastructure feel legible. The anxiety in the journey came less from speed and more from incomplete information, broken momentum, and weak end-state reassurance.

Insight 01

The absence of status information was not neutral. It actively created anxiety.

Insight 02

Customers would tolerate waiting if timing and outcomes were explained clearly.

Insight 03

Surprise requirements mid-flow were one of the most avoidable causes of abandonment.

Insight 04

The experience needed one mental model even when the underlying rails differed.

Design

The redesign shifted the journey from rail-led to customer-led.

The most important moves were structural. I surfaced live FX and corridor expectations earlier, added a "what you'll need" preparation step before data entry, kept eligible journeys fully native inside the app, and gave end states the same design attention as the entry flow.

Dandelion transfer review screen showing rate, fees, and delivery timing
Transparent review state with exchange rate, fees, and arrival expectations visible before send.
What you'll need screen with annotated design rationale
Preparation screen that reduced the surprise factor for country-specific transfer requirements.
Transfer confirmation and in-app tracking screens
End-state and tracking design turned the quietest part of the journey into a confidence-building moment.
Transfer timeline with processing milestones
Tracking replaced the old manual trace process with visible progress and a stronger recovery model.

The core design principle was consistency of mental model: customers should not have to know whether they were on SWIFT or Dandelion. They should only feel that the experience was predictably CommBank.

Decisions

Three decisions shaped the final experience more than any visual polish.

Confidence over speed messaging

We resisted turning the experience into a pure speed story. Reassurance and clarity had a bigger effect on trust than aggressive claims about instant delivery.

Explicitly name uncertainty

Unknown overseas fees were surfaced clearly and earlier. Transparency built more trust than trying to hide imperfect information.

One unified customer model

Different rails remained an internal concern. From the customer point of view, the experience was designed to feel seamless and familiar.

Outcomes

The experience became faster, clearer, and easier to trust.

The strongest validation came from usability performance and the structure of the shipped phase 1 experience. The redesign translated new infrastructure into something customers could understand and act on with confidence.

100% Task completion in usability testing, giving the team confidence that the redesigned journey held together under scrutiny.
22 Additional corridors designed into phase 1, each with its own regulatory and information requirements, handled through one clearer journey structure.
3 Delivery methods supported in the experience: bank deposit, digital wallet, and cash pick-up, with groundwork in place for stablecoins.

Reflection

The real design job was making infrastructure feel human.

This project reinforced a lesson I keep returning to in fintech: infrastructure improvements do not automatically create experience improvements. Faster rails and better economics only matter when the design makes them understandable, trustworthy, and emotionally manageable.

What I would carry forward from this work is the importance of structural decisions. The best choices here were about timing, sequencing, wording, and end-state behavior. Those are the things that changed the emotional shape of the journey.

Next case study

CommBroker

How I turned a broker portal built on memory and workarounds into a clearer self-service platform.

Read the next case study