Case study · UX/UI take-home · Mobile app · 2026
one day, five places, a bus she has never taken.
A guide that plans the order, tells her when to walk and when to take the bus, and changes the plan when the day changes: a late bus, a longer visit, rain at dusk.
Jane, 30, solo, tech-savvy, one day, by bus. Start at the Sheraton, end at Vincom, three places in between. I split the brief into what is given, what is required and what is missing.
The brief asked for a bus app. Read closely, the real problem was a day: five places, a time limit, a bus network she has never used, and a plan that has to change while she is out in it.
Explain every suggestion, show when data is estimated, allow reordering. But tech-savvy does not mean she knows the local bus: validating a ride and knowing where to get off still need guidance.
She wants to make the most of the day and may change plans. The app has to say clearly when a plan is unrealistic, and re-plan fast.
No one reminds her of the time or where to get off. The app has to act as a companion: proactive, one-handed, still useful offline.
The brief left three things undefined, so I defined them.
“Overall travel experience” became: avoid closing times, avoid the midday heat, include a sensible lunch. “May not meet” the time limit meant alerts must be proactive, while planning and while travelling. “Alternative routes” meant the app must offer the next step, not just report a delay.
Start and end are fixed, so only three stops can move: six possible orders. That suggested a feature for comparing orders. Research later made it unnecessary.
I know the city; Jane does not. I wrote down what I believed, then checked each belief against official hours, news, traveller reviews, walking times and hands-on tests of local transit apps.
The whole route is about 47 minutes on foot; three of four legs are under 15. The bus only wins on two legs.
ConfirmedThe brief’s order already does this. So the value is not finding a hidden order but explaining this one, and re-planning it.
ConfirmedNeither place closes at midday. The real deadline is the Museum’s 17:00 ticket cutoff, and the Opera House is best at dusk.
RevisedBuses are free in 2026, but riders validate with a card tap, foreign visitors got stuck at that step, and buses often do not stop unless waved down.
Sharpened34–35° at noon; after 17:00, rain, rush hour and high tide can overlap. Far legs early, short walks in the evening.
ConfirmedVisceral and overwhelming. The Agent Orange room was confronting and the war crimes room was too much.Traveller reviews of the War Remnants Museum. The plan should leave a quiet moment after it, not schedule the next stop back to back.
The competitor review moved the problem twice.
First I looked at transit apps only, then at everything Jane might use for a day in a city: trip planners, guides, journals. Each owns one part of the day. Planners plan before, journals remember after, transit apps move you one leg at a time. None stays with you during the day and re-plans from what is actually happening.
Two claims I had in mind, “optimises the order” and “combines transport modes”, were already done by others. I dropped them rather than pretend.
Before any idea, I wrote what success means for her. Every later check was made against this list, not against my taste.
When I have only one day in an unfamiliar city, I want to visit the places I chose without wasting time or getting lost, so I can enjoy the day with confidence.Job to be done
Visits every place she chose, or knowingly drops one for a better alternative
Reaches the final stop within her time limit
Is never surprised by a closing time
Never stands at a stop unsure which bus to take or how to board
Can see what to do next at any moment
At any moment she sees what to do next without reading.
Every suggestion says why, in one line.
An alert always comes with an alternative she can take in one tap.
I went deep on the two that carry the most product thinking and the clearest gap: the stop order and the time limit. The rest were built to a solid, familiar standard.
The order is already clear, so the value is explaining it and re-planning fast when the day changes
Turns opening hours and the end time into early alerts she can act on
Needed for trust; focused on delays, waiting time and alternatives
The entry point; kept short because the stops are known
A familiar pattern, focused on “get off at the next stop”
Research said walking beats the bus on most legs. The brief said “by bus”. Forcing either would be wrong.
So the app compares both for every leg and recommends one from distance, heat, rain, the bus’s arrival and the time left. The bus is always shown with live times, and it is the default on long legs or in bad weather. For the comparison to be fair, a bus leg counts everything, including the wait:
Dotted is walking, solid is the bus, as on the app’s day line.
Lunch, rest from the heat and alternative plans were not added as new features. They live inside the five the brief asks for: lunch is a stop in the timeline, a cool place to sit appears next to “take the bus”, and a time alert offers to drop, shorten or swap a stop. Broad thinking, without diluting the product.
An ideal day never triggers the moments that make the app different. So the prototype follows a normal day with four disruptions, each testing a different part of the product, linked by cause and effect.
So the bus is suggested, with a guide to validating the free ride before boarding.
Live arrival, said as a time, not a warning. The leg time includes the wait.
The calm baseline: the plan on track, nothing asks for attention. The Museum stamp is stuck on the photo and franked 8:58.
The plan bends without blame: lunch first, the Palace after, still inside 19:00.
One optional tap on how it felt, then a quiet garden café four minutes away.
A nearby place to rest and cool down instead of another walk.
Walk now, 13 minutes, and arrive dry; the bus would take 15–25.
45 minutes inside the limit. The recap draws the day as she lived it.
The amber stops are the four disruptions. The line is dotted where she walks and solid on the bus. Every time, duration and price on the screens is read from this one timeline.
I logged every decision that changed, with what triggered it: 48 in total. The pattern taught me more than any single decision.
Most evidence-driven changes came early and cost a sentence. Most direction changes came late and cost a rebuild.
The day as chapters, one “Now” card with advice, an arrival guide. The list stays behind “Change the order”.
A whole day with a time limit and opening hours, re-planned as it happens. No app reviewed does all of it.
The alert names the real cost: lunch slips to 12:30, with no break after a heavy museum. A false alarm destroys trust in every alert after it.
Daylight on the planning screens, quiet at the heavy places. The fix was to scope the lesson, not reverse it. This page wears the same daylight.
Colour and illustration were chosen against an unstated target and redone once it became clear. Next time: a one-paragraph visual brief before any colour.
Two differentiators were named before the facts that decided them. Next time: distances, hours and competitor features first.
Both logic errors came from designing from intent. One shared timeline caught them; now every number on a screen comes from it.
A take-home leaves no time for usability tests. So every check is a structured self-review, labelled as such, and the riskiest assumptions are listed for the first real test.
23
WalkthroughEvery screen answers “what do I do next?”. A second pass against the timeline fixed five places where the prototype contradicted the day.
5/5
CriteriaClosing times, boarding and “what next” were only partly met until three missing states were added: no live data, a stop at risk, waiting out the rain.
412
Accessibility412 text segments on 23 screens, measured against what is actually behind them. One real failure, text on a bright photo, was fixed.
Can they find the stop, recognise the bus, wave it down and validate without help?
Do they understand the consequence, and trust the suggestion enough to take it?
Do they follow the recommendation, and does the reason change their choice?
The full screens, the design system and the clickable prototype live in the Figma file. In short:
ghé: to stop by.
Vietnamese readers see ghé, “stop by”; English readers see Ghé’s, the guide’s place. The letters are drawn as one road, and the pin is both the accent and the apostrophe.
Wave
Guide
Tip
Umbrella
GoodbyeHotelMuseum
Palace
LunchCoffee
Opera HouseVincom
Arrival is the moment the day is posted home. The place stamp is stuck on the photo’s corner like a postage stamp, a third off the image, and a postmark with the arrival time strikes over it.
My call, late in the project (Decision #48). It reversed an earlier rule, “a stamp never sits on a photo”. The postmark is decoration: the arrival time is always in text on the screen too.
+
The interesting part was never the route. It was the moment the day stops going to plan, and whether the app is still someone you trust.
Next
