People finished tasks - then went back to check them twice.
Dense tables, no hierarchy, confirmations that came late or not at all. In port operations, “not sure if that saved” is a delayed shipment.
SlotXpert
Making port operations feel safe to use.
An internal B2B platform for the Port of Singapore Authority. Two user roles, one shared spine — and a redesign built around one idea: people should trust what they just did.
Case study · B2B operations≈ 4 min read
Impact · six months after launch
Fewer mistakes, faster adoption.
0135%
Fewer operational errors
0285%
Team adoption within six months
0380%
Retention versus the previous workflow
Measured by PSA’s internal QA team ✦Six-month post-launch review
The short version
01 · Problem
02 · What I did
03 · Result
Rebuilt the interaction layer around error prevention.
Role-based dashboards, one shared pattern language and confirmations you can’t miss — without retraining a single user.
35% fewer errors and 85% adoption in six months.
Users stopped repeating actions out of uncertainty. Both numbers come from PSA’s own QA review.
My role
Sole designer, research through handoff.
Embedded in PSA’s cross-functional team through the Metafic design studio. I owned the process end to end — no hand-offs between designers, one line of thinking from data to shipped UI.
- 1
Research & diagnosis
Mapped error-prone workflows from ~295 survey responses.
- 2
IA & flows
Role-specific IA with one shared pattern language.
- 3
Interaction design
Confirmation states, inline validation, progressive disclosure.
- 4
UI system
Components built around the backend and the shipping calendar.
- 5
Developer handoff
Stayed close to engineering all the way through implementation.
01Context & constraints
Redesign it without breaking anything people already knew.
SlotXpert is how PSA’s teams book and list container slots at one of the world’s busiest ports — huge volumes, every single day.
Five constraints shaped every decision
- 01Years of muscle memory
Any change had to feel like a better version of what users already knew.
- 02Massive tables
Cognitive load was already the limiting factor before I added anything.
- 03Low tolerance for change
Adoption asked for familiarity, not novelty.
- 04Live port operations
Delivery ran alongside operations — no quiet moment to roll out.
- 05Fixed backend
Designs had to route around the data structures, not through them.
02User research
~295 survey answers, three clear patterns.
PSA had surveyed both teams before I arrived. I took the raw data and turned a wall of freeform answers into something the team could act on.
Q.01
Task completion confidence
Could users complete key actions without hesitation?
Q.02
Error frequency
Where in the flow were errors clustering?
Q.03
System trust
Did users trust the system had recorded what they just did?
Source: PSA-provided survey summaries. I synthesised and visualised the findings; I did not administer the survey.
Completion wasn’t the problem. Confidence was.
Most users could technically finish their tasks — then went back to check them a second and third time. That framed the whole redesign: reduce errors, close the trust gap, don’t retrain anyone.
03The core problem
The platform never told people, clearly, what had just happened.
Missed confirmationsMisread rowsAmbiguous states
Careful users
slowed down.
They compensated for uncertainty by re-checking everything — safe, but slow.
Fast users
got things wrong.
They trusted a system that wasn’t telling them enough — quick, but error-prone.
The brief wasn’t visual. It was operational: make the system trustworthy enough that people act with confidence the first time.
04Information architecture & flows
Two roles, one shared spine.
Buyers and sellers run parallel workflows. They had to work side by side without drifting apart visually.
- 1Role routing at login
Buyers and sellers land in their own dashboards. The old shared dashboard mixed both roles — and quietly created most of the confusion.
- 2Primary actions up top
Book slot. List container. Surfaced at the top of navigation — no hunting.
- 3Same patterns, both flows
Learn one and you already know most of the other.
From the client’s map to my exploration
The client’s wireframes showed the existing mental model — a map, not a UI to copy. Wherever the friction data pointed to a broken interaction (slot booking for buyers, container listing for sellers), I rebuilt it.
05Key decisions & trade-offs
Safer beats cleaner in port operations.
Four trade-offs I made on purpose — each one chose predictability over polish.
- 01
Error prevention
visual minimalismA cleaner design would have removed some confirmation feedback. I didn’t. The screens are busier than a consumer product — they’re also safer.
- 02
Progressive disclosure
showing everythingHundreds of container records don’t need to be visible at once. Reveal them in context, where users are actually looking.
- 03
Familiar patterns
“better” new onesRetraining hundreds of trained users is slow and expensive, so core workflows stayed close to the existing mental model.
- 04
Shipping within backend limits
ideal solutionsSome ideal solutions didn’t fit the data structures. I documented every compromise so future iterations can revisit them.
06Final UI
A row’s colour always means the same thing.
Status colour and organisation identity stay visually separate. If you see amber, it’s amber for a reason — never because it’s a client’s brand colour.
- 1
Primary action first. Book slot and list container sit at the top of navigation.
- 2
Explicit confirmation. Every commit ends in a state users can’t miss or misread.
- 3
Colour = status only. Organisation identity never borrows a status colour.
Data is anonymised and product-identifying details removed for NDA compliance.
07Outcomes
Six months later, people trusted what they did.
From PSA’s six-month post-launch QA and stakeholder review.
35%
Reduction in user errors
85%
Adoption across buyer and seller teams
80%
Retention versus the previous workflow
↓
Time-on-task on both core flows
Users stopped repeating actions out of uncertainty.
The 35% didn’t come from one big idea. It came from the same error-prevention principles applied consistently across dozens of small interactions.
Metrics sourced from PSA’s internal QA and stakeholder review.
08Limitations & reflections
Predictability beats novelty.
People who use a system every day for high-stakes work don’t need it to look clean. They need it to act predictably enough to trust their own actions.
Test with real users at the wireframe stage.
Stakeholder review worked and the numbers say the direction was right — but moderated sessions with buyers and sellers before hi-fi would have de-risked interaction decisions earlier.
A proprietary system, shown with care.
SlotXpert is internal to PSA. Screens use anonymised data with product-identifying details removed.
