Pranjal AgarwalWorkRoamer
UX Case Study · 2024

ROAMER

A travel discovery app for people who prefer getting lost. End-to-end UX from research to hi-fi — including a wrist companion.

Role

Solo UX Designer

Duration

12 Weeks

Platform

iOS · watchOS

Methods

Interviews · Lo-fi · Hi-fi · Testing

01The Question

Why do good places disappear?

"Build a travel app." Four words that pointed at a symptom and missed the disease underneath. Everyone has a place they loved and can no longer find — buried in a camera roll, a screenshot, a WhatsApp message. The brief pointed at features. I went looking for the feeling underneath.

Reframed as: help people stay lost, intentionally. A travel product for the moments between plans, not a routing engine optimising the path between them.

HMW make discovery feel accidental?HMW let memory live beyond the trip?HMW serve planners and wanderers both?HMW use language people actually speak?HMW make the wrist the primary surface?
02Research

What I *actually* observed.

01
A strong gut, no support

Users had rich internal decision logic — apps forced them into keyword searches that dismissed how they actually think about places.

02
Planner vs. wanderer: false split

No product bridged both modes. Planners got itinerary builders. Explorers got maps. Nobody got the moment between the two.

03
Memory is broken

Everyone described places loved and lost — buried in photos, old bookmarks, a WhatsApp message nobody can find. Four apps tested. All failed at the same point.

The solution to changing the product is a solution to changing the problem — and we still don't actually understand it.

14

User interviews across 3 archetypes

87%

Said apps "get in the way"

6

Competitor products audited

3

Prototype iterations before hi-fi

"Can't describe the vibe in keywords""Discovery feels like homework""Saved places go unvisited"
"Trusts friends over stars""Opens 4 apps for 1 decision""No memory of why something was saved"
"Would use wrist if it didn't need the phone""Re-discovers same place by accident""Doesn't trust recommendations"
03Wireframes & IA

I designed it *three times*. On purpose.

The original concept
Search-first home

First attempt prioritised a search-first model. Users hated it immediately. The keyword box felt like homework.

The first key insight
Mood over keyword

Removing the search bar and replacing it with a mood selector reduced decision anxiety. Users described it as 'finally feeling like someone understood how I travel.'

Final direction
Map + bottom sheet

Structure before style. Navigation logic validated before a single colour was chosen.

Information Architecture

The full map of Roamer

ROAMER
├── Onboarding  (zero onboarding — map opens immediately)
├── Home / Roam Mode
│   ├── Mood Filter  (Quiet · Alive · Golden · Outside · Sharp · Lost)
│   └── Map + Results  (place cards below)
├── Place Detail
│   ├── Description
│   ├── Your Memory  (personal note)
│   └── Navigate / Save
├── Memory Vault
│   ├── Saved  (with context)
│   ├── Visited
│   └── Shared
└── Watch Companion
    ├── Distance card
    ├── Place name
    └── Roam CTA
04Hi-Fi Screens

Low friction. *High intention.*

Every screen justified by a research finding. No feature added without a pain it resolved.

Roam Mode
Karim's Alley Café
127m · Quiet · Local
Opens 8am · Saved

Research: Users didn't want to search. They wanted to be shown something that matched their mood.

Mood Filter
How do you want
to feel right now?
Quiet / Alive / Golden
Outside / Sharp / Lost

Research: Atmosphere over taxonomy. 'Alive' returns different results than 'bar' — because that's how people actually talk.

Place Detail
Karim's Alley Café
Your Memory:
"Fixed my whole morning.
September, rainy Tuesday."

Research: Memory is a first-class feature. Not a pin. An experience.

Memory Vault
"Fixed my whole morning."
"Stumbled in during a storm."
12 saved · 8 visited

Research: Saves capture who you were with and why you loved it. Not just a location bookmark.

Apple Watch
127m away
Karim's Alley Café
Roam →
tap crown to dismiss

Research: Watch-first design forced brutal prioritisation. If it doesn't survive the wrist, it doesn't deserve a phone screen.

05Design Decisions

Designed from first principles, on purpose.

1
Language is the interface

Changing filter labels ('Quiet' vs. 'Relaxed') drove the biggest usability gain — bigger than any layout change. Words are product decisions.

2
Watch before phone

Designing for 44mm first forced brutal prioritisation. If it doesn't survive the wrist, it doesn't deserve a phone screen.

3
Memory as first-class

Saves capture context — who you were with, why you loved it — not just a pin on a map.

4
Mood over category

Discovery filtered by atmosphere, not taxonomy. 'Alive and buzzing' returns different results than 'bar.'

5
Zero onboarding

The first second of the app mirrors the first second of arriving somewhere new. No tutorial. Just a map and a mood.

6
Disappear into the place

If users are looking at their phone, the app has failed. If they're looking at the street, it worked.

06Usability Testing

I didn't just test usability. *I measured confidence.*

5 moderated + 3 unmoderated sessions. Tasks from first-time discovery to re-finding a saved memory. Success metric: did you feel more capable after?

94%

First-attempt task completion. No critical navigation errors recorded in hi-fi testing.

78%

Activated Roam Mode in session one, before being prompted. Feature sold itself.

91%

Correctly interpreted mood filters with zero explanation. Language design working.

8/8

Every tester said they'd replace their current travel app. "Finally made for me."

"Finally feels like it was made for me."

Participant 2

"I actually felt like I could get lost and be okay with it."

Participant 5

"The Watch companion made me stop looking at my phone. That's a first."

Participant 7

07Reflection

What this project *actually taught me.*

The brief is never the problem

Reframing the brief from 'travel app' to 'help people stay lost intentionally' changed every design decision that followed.

Constraints are creative fuel

Designing Watch-first made the phone app better. The most limited surface forces the most honest decisions.

Language is the interface

The biggest usability gain came from changing labels, not layouts. Words are product decisions.

Measure confidence, not completion

Task completion is a floor. The real question: 'After this, do you feel more capable?' That's the metric worth chasing.

"Good work isn't made from following a brief. It's made from the decision to keep being wrong, keep building, keep returning to the people you're building for."

Next Case Study →GenderverseMaking gender equality everyone's conversation.