A museum app for every moment
Overview
TFAM's mobile guide had useful features, but key tasks were scattered across different touchpoints. I redesigned the experience around three moments: plan, wander, and remember.
Role
Product Designer
Scope
Research, product strategy, UX/UI, prototyping, testing
Timeline
Q2–Q3 2026
(10 weeks)
Validation Overview
Task completion
All 5 participants completed 4 out of 4 core tasks during usability testing.
Flow validation
Participants successfully completed key flows across discovery, booking, audio guidance, and navigation.
Issues identified
Testing surfaced 3 usability issues, which informed the final iteration.
CONTEXT
TFAM is Taiwan's first museum of modern and contemporary art, and it already had its own app. It looks polished, with strong branding and a clean interface, exactly like something made to show off the museum. I took it on as a solo redesign, from research through prototyping, to find out why so few people used it, and how to change that.
Project note
Independent concept project. Not affiliated with or commissioned by TFAM. Exhibition names are real; artwork imagery is created for demonstration purposes and is not official TFAM content.
SYMPTOMS
Built to look like art, not to be used
The app was not short on features. It had plenty. But it was built to look like the museum first, so the features visitors needed most were either hard to reach or missing. Four problems stood out.
Audio guide hard to start
The app's whole reason to exist, yet starting it meant turning on Bluetooth or typing in a code. No clear way in.
No way to book
Booking a tour or class meant leaving the app for email or a phone call.
Exhibition info was incomplete
The app only showed exhibitions that had an audio guide. For everything else, you went to the website.
Buttons don't look tappable
The interface was so clean, visitors couldn't tell what to press.
These weren't just my read. For an app this polished, the public reviews were only lukewarm.
“A curator's vanity project, not a useful app for the public.”
— TFAM app review, App Store
Heuristic Evaluation
Reviews pointed to recurring frustration. I used a heuristic evaluation to identify where the interface was creating friction and turn those complaints into specific design problems.
Homepage
Navigation is difficult to discoverThe main navigation is hidden behind an unlabeled symbol.
Visuals overpower navigationThe exhibition image dominates the screen, leaving little structure to guide visitors.
Interactive elements lack clear meaningThe marks at the bottom appear interactive, but their purpose and status are unclear.
Side Bar
Icons are hard to understandThe left sidebar uses six icons without labels, so visitors have to guess what each one means.
Visual hierarchy is inconsistentSpacing and alignment change throughout the sidebar, making related information feel disconnected and harder to scan.
The app unexpectedly sends users to the websiteWhen visitors tap Current, Upcoming, or Past, they are taken out of the app and into a browser instead of staying in the app.
Audio Guide
The back button is hard to recognizeThe back icon has low contrast and uses an unusual shape, making it harder to spot and understand.
The text is harder to readThe description uses wide spacing and full justification, creating uneven gaps between words on a small screen.
Code Input
The screen label does not match the task“Keyboard” describes the tool, not what visitors need to do.
The button label is misleadingThe button says “Send,” but visitors are starting an audio guide.
Homepage
Side Bar
Audio Guide
Code Input
DIAGNOSIS
Five visitors, one clear pattern
I used two research methods. First, five in-depth interviews with real visitors, both locals and tourists. Second, a four-lens audit of the live app: App Store reviews, TripAdvisor feedback, a heuristic review, and my own walkthrough inside the museum. The interviews showed me the pattern. The audit confirmed it.
Friction showed up at three moments of the visit
Plan (Before the Visit)
Exhibition details are hard to find, and booking means an email or a phone call.
Wander (During the Visit)
The audio guide has no clear way in. Visitors lean on staff and signage instead.
Remember (After the Visit)
Nothing carries the visit home. There's no way to keep the pieces you loved.
Plan (Before the Visit)
Exhibition details are hard to find, and booking means an email or a phone call.
Wander (During the Visit)
The audio guide has no clear way in. Visitors lean on staff and signage instead.
Remember (After the Visit)
Nothing carries the visit home. There's no way to keep the pieces you loved.
Two visitors, two different breaking points
Yu-Chen, 29
Marketing manager in Taipei
Goal
See what’s on and book a tour in a few taps.
Frustrations
Planning means juggling the website, social media, and a phone call.
Breaks
Before she leaves home. Language is never her problem, planning is.
Marco, 34
Architect from Milan
Goal
Find his way and understand the art without leaning on staff.
Frustrations
The app and signage assume Mandarin, and the audio codes on the placards mean nothing to him.
Breaks
The moment he walks in. Language is the first wall he hits.
Each frustration turned into a question
How might we surface the audio guide the moment a visitor arrives?
How might we help first-timers navigate without staff or signage?
How might we help recurring visitors discover current and upcoming exhibitions?
How might we let visitors book in a few taps instead of emailing or calling?
How might we help visitors keep the works they loved?
The decision this led to
The personas pointed to one choice. Instead of redesigning a list of features, I anchored the whole app on three moments in a visit: plan, wander, remember. Every feature would live inside whichever moment it actually served.
Guiding principle
People take in information best right when they need it, not all at once on a home screen.
SOLUTION
One companion, three moments of the visit
From three moments to a five-tab app
The three moments became five tabs: Home, What's On, Map, Activities, and Settings. Every feature that was hidden in the old app now sits in the tab where a visitor would look for it. The audio guide, the old app's weakest point, can now be reached from Home, What's On, and the Map, so it's never more than one tap away.
Home
Arrival Screen
Audio Guide
What's On
Exhibition List
Exhibition Detail
Getting There
Map
Floor Map
Suggested Route
Exhibition Detail
Activities
Activities List
Date/Time Picker
Slot Picker
Confirm Screen
Booking Confirm
Settings
Notification
Interest Picker
Notification Toggle
Plan (Before the Visit)
What's On
Current and upcoming exhibitions in one place, so visitors stop hunting across social media, the website, and the front desk.
Pre-Book a Tour or Class
Reserve in a few taps, right in the app, instead of an email or a phone call.
What's On
Current and upcoming exhibitions in one place, so visitors stop hunting across social media, the website, and the front desk.
Pre-Book a Tour or Class
Reserve in a few taps, right in the app, instead of an email or a phone call.
Stay in the Loop
A notify-me toggle lets repeat visitors like Yu-Chen get told when a new exhibition or class opens, so their visits stay fresh.
Wander (During the Visit)
Arrival and Audio Guide
The app opens on a clear "Start audio guide" button. No digging through menus.
Offline Floor Map
Works without a connection, so first-timers can find their way without relying on staff or signage.
Arrival and Audio Guide
The app opens on a clear "Start audio guide" button. No digging through menus.
Offline Floor Map
Works without a connection, so first-timers can find their way without relying on staff or signage.
Two Languages
The whole app works in English as well as Mandarin, so a visitor like Marco isn't stuck at the first wall. The audio codes on the placards now map to something he can actually read and start.
Remember (After the Visit)
Save the pieces you loved during the visit. This was the lowest-friction moment of the three, so it was the right one to cut from this build while booking and wayfinding came first.
Note on the numbers below
This is a self-initiated passion project, so these are design targets I'd measure against, not results from real usage. I'm labeling them that way on purpose. Validating them against real visitor data is the natural next step once the museum's booking system exists.
+0%
Feature discovery
Target lift in audio guide use.
0%
Self-service booking
Target booking completion, in under 45 seconds.
0/5
Visitor confidence
Target confidence score, from a post-visit survey.
Try it yourself
This is a real, working prototype. Walk through the arrival screen, start the audio guide, and book a class, just like a visitor would. Please open it on your phone.
Open the PrototypeTESTING
It worked, but it wasn't finished
I built the first version as a full, working prototype and tested it with five visitors across the range TFAM serves, ages 25 to 58. All five finished the four core tasks (start the audio guide, find a gallery on the map, book a class, and check what's on), so the structure held. But passing the tasks only proved the app worked, not that it was finished. In fixing an app that was too pretty to use, I had overcorrected into one that was too plain to feel like a museum. The test showed me a good app has to do three things at once: be usable, feel like the brand, and be open to everyone. These are the problems that surfaced.
Finding 1: The arrival screen lost its context
Observed
One visitor wasn't sure what app she was even looking at. The arrival screen showed only a small logo, no museum name, nothing that felt like TFAM. In making it simple, I had stripped out the brand.
Existing App
Version 1
Version 2
Version 1
I put the audio guide at the top of the arrival screen. That made the audio guide easy to see, but the app lost its context. People had no idea what the app was. All that was left was the TFAM logo mark in the top-left corner, which most people didn't recognize.
Version 2
I built a splash screen to carry the brand, then added a clear "Welcome to Taipei Fine Arts Museum" on the arrival screen. Now the app tells you what it is and where you are the moment it opens, and the working screens still stay clean and simple.
Observed
One visitor wasn't sure what app she was even looking at. The arrival screen showed only a small logo, no museum name, nothing that felt like TFAM. In making it simple, I had stripped out the brand.
Existing App
Version 1
I put the audio guide at the top of the arrival screen. That made the audio guide easy to see, but the app lost its context. People had no idea what the app was. All that was left was the TFAM logo mark in the top-left corner, which most people didn't recognize.
Version 2
I built a splash screen to carry the brand, then added a clear "Welcome to Taipei Fine Arts Museum" on the arrival screen. Now the app tells you what it is and where you are the moment it opens, and the working screens still stay clean and simple.
Finding 2: Designed for two hands, used with one
Observed
Visitors found it hard to type a code to start the audio guide. This is a situational disability, not a preference. In a museum full of families, many people have only one free hand, holding a child or a bag, and typing a number needs two hands and full attention.
Existing App
Version 1
Version 2
Version 1
Starting the audio guide meant typing a code on a keypad. It worked, but only if you had two free hands and could look at the screen the whole time. For a parent holding a child, or anyone with one hand full, that was hard to do.
Version 2
I made scanning a QR code the first option, since it takes just one tap, and kept typing the code as a backup for anyone who prefers it. Now most people can start the audio guide with one hand.
Observed
Visitors found it hard to type a code to start the audio guide. This is a situational disability, not a preference. In a museum full of families, many people have only one free hand, holding a child or a bag, and typing a number needs two hands and full attention.
Existing App
Version 1
Starting the audio guide meant typing a code on a keypad. It worked, but only if you had two free hands and could look at the screen the whole time. For a parent holding a child, or anyone with one hand full, that was hard to do.
Version 2
I made scanning a QR code the first option, since it takes just one tap, and kept typing the code as a backup for anyone who prefers it. Now most people can start the audio guide with one hand.
Finding 3: You could swipe, but not everyone can
Observed
The card sliders used a peeking card and dots to show they could be swiped. That signals it well, but swiping is still a gesture, and a gesture needs finger reach and flexibility. For a visitor using one thumb, knowing you can swipe doesn’t help if the swipe itself is hard to do.
Version 1
Version 2
Version 1
You moved between cards by swiping. The peeking card and the dots hinted that you could, but swiping was the only way, so if the gesture was hard for you, you were stuck.
Version 2
I added an arrow so you can move between cards with one tap, no swipe needed. Where a peeking card already hints there's more (What's On and Activities), the arrow shows up only when you start to drag, keeping the screen clean. Where there's just a row of dots (the full exhibition slider), the arrow stays visible, since without it people didn't realize the slider could move.
Observed
The card sliders used a peeking card and dots to show they could be swiped. That signals it well, but swiping is still a gesture, and a gesture needs finger reach and flexibility. For a visitor using one thumb, knowing you can swipe doesn’t help if the swipe itself is hard to do.
Version 1
You moved between cards by swiping. The peeking card and the dots hinted that you could, but swiping was the only way, so if the gesture was hard for you, you were stuck.
Version 2
I added an arrow so you can move between cards with one tap, no swipe needed. Where a peeking card already hints there's more (What's On and Activities), the arrow shows up only when you start to drag, keeping the screen clean. Where there's just a row of dots (the full exhibition slider), the arrow stays visible, since without it people didn't realize the slider could move.
REFLECTION
What I learned, and what comes next
What I learned
An app is not a landing page. I built the first version like a landing page: lead with the main action, keep it short. It was easy to use, but testing showed an app has to feel like a place, not just work like a funnel.
Simple and beautiful, not one or the other. Many polished apps I looked at were pretty but hard to use, the same trap as the old TFAM app. So I kept the simple, usable base and added TFAM's identity on top, mostly on the arrival and splash, not the working screens.
Micro-interactions are communication. Small motion in an app isn't just polish. It tells you a button works, a tap landed, or the app is loading. Most of my second round went into these details, and they mattered more than I expected.
What's next
Build the Remember phase and real booking. Both are designed and tested in the prototype, but need a backend, so they’re saved for a later build.
Do more research on the map experience. Right now it’s flat, just numbered steps. For first-timers like Marco, I want to test whether a more guided or interactive wayfinding experience actually helps, before deciding what that looks like.