YumGo - UX Case Study
Led a team of 4 from research to a tested prototype, and rebuilt a broken filter mid-study so testing could go on.
Overview
Led a team of 4, from research to the tested prototype
Providing comfort in unfamiliar places
YumGo is a food discovery app for tourists. When you're somewhere new, with dietary needs or a certain vibe in mind, it helps you find a place that feels right before you walk in.
- !The problem Finding restaurant details often means bouncing between apps, with missing menus and info you can't trust.
- Why it matters Decision-making quickly becomes stressful on top of other travel pressures.
- +The solution One app for the whole journey, from discovering to dining, so you know what to expect before you sit down.
Result
Designed my version of the final prototype
YumGo (Taylor's Version)
Everyone on the team finished their own final prototype. Mine reflects the fun of travelling, while bringing the comfort of apps you already know. Try it yourself!
Every screen, from discovering to dining
What came out of it
- Every feature tied to research Insights from interviews led to the menus that list allergens, notes on local customs, restaurant tags, and photos that showcase each place's vibe.
- Filters that finally worked Inconsistent filter selection stalled our first 2 test participants. After I rebuilt it, all 6 remaining participants finished the task without getting stuck.
- A shared brief for the team I wrote and maintained a shared brief so all 4 of us were working from the same findings and the same scope.
The process
How we got there
Problem
Eating while travelling should be good and easy. No ifs, ands, or buts.
Unfamiliarity is often what creates such novel and exciting travel experiences. It's all fun and games, until you're tired, short on time, juggling a group… and then hunger hits.
Now you're bouncing between apps to find the info you actually care about. Where do you start if you can't even think straight?
Problem Statement Tourists need a personalized and centralized way to discover and book restaurants while abroad, so that their travel experiences can be memorable, culturally-immersive, high-value, and easy.
Secondary research
Convinced the team to pivot from speed to authenticity
We (wrongfully) assumed it was all about time
We started out with the assumption that tourists wanted food ASAP so they could get back to sightseeing. Early research suggested otherwise, since wait times barely came up.
!Missing information Incomplete business and menu info
!Weak discovery & booking Apps didn't support the full journey
!Irrelevant content Full of noise and unwanted results
+Desire for authenticity! Tourists wanted local, quality meals
Not everyone was ready to let go of the original idea, but I emphasized how it's important to move in the direction that the research is pushing us.
The takeaway We shifted gears away from wait times and toward the authentic experiences that weren't showcased in current apps.
Interviews
Co-wrote the guides, moderated 2 of 8 interviews, and ran our affinity mapping
Travellers wanted to dine with certainty
We ran 8 interviews with recent tourists about travel, food, and the apps they use for both travel and food. I co-wrote the screener and interview guide, moderated 2 interviews, and took notes on 2 more. Then I ran our affinity mapping session to pull everything together.
On the surface, people talked about menus, reviews, and photos, but thematic analysis revealed that they all wanted a sense of familiarity in an unfamiliar place.
The takeaway People wanted to feel like they'd been somewhere before they walked in the door, so we focused on cultural connection and seeing the menu ahead of time.
Prototyping
Defined our main user path, built the main restaurant page, and kept teammates' screens consistent
So… what do we make?
We got a little too excited about features to add, so I defined our main user path early to keep us on track. Then I built the wireframe for the restaurant page, including its info, menu, and reviews, and revised teammate screens to keep a consistent visual language.
Our 2 rules moving forward
- Less is more! Fewer features to better support one complete journey
- Familiarity is the way Borrow from apps people already know
Testing
Rebuilt the broken filters between tests
Clarity had to be improved… a lot
I co-wrote our usability test guide, and 8 participants tried 2 tasks in our Figma prototype:
- Book a table somewhere dog-friendly with vegan and gluten-free options
- Write a review afterwards
MAYDAY
Task 1 sent people straight to the search filters, which was full of broken toggles, and both of our first 2 participants got stuck here. OH NO!
I knew that if we left it, a big portion of testing time would be wasted revealing something we already knew. Since I knew Figma's variables and components best out of the team, I tackled the problem in the couple of hours before the next session. I successfully rebuilt the filters' variables, components, and interactions so that every toggle responded as expected.
Left: the initial toggle, where not all filter buttons updated as expected. Right: icons are now properly selectable and deselectable, and their status is reflected in the restaurant list.
All 6 remaining participants used the filters without getting stuck, and the rest of the sessions surfaced things we didn't already know. I felt like a superhero. Phew!
What we discovered
!Filters were unclear People still didn't know what the icons meant
!Key info was easy to miss Too much text buried what mattered
The fix · Labels for every icon, and less text
From tested to final
Participants didn't understand the icon-only filters, so every icon got a label. Months later I restyled the whole app, checked it against WCAG AA and touch guidelines, and brought group voting and Discover Toronto into a Community tab. The filters work the same way, but I haven't tested this final version.
Design system
Set up the tokens, components and documentation
When I polished YumGo for my portfolio, I rebuilt the components and styles on a full design system.
- Reusable restaurant page for all restaurants I used variable modes to swap the content on the restaurant page, keeping consistency between 5 restaurants.
- 31 component sets Tab bars, headers, footers, filter chips, and more, each with descriptive variant states.
- AA contrast throughout Every text pairing is at least 4.5:1, and every icon has a text label.
| Tier | Tokens | Example |
|---|---|---|
| Primitive | 211, hidden | colours/brand/700 |
| Semantic | 117 | bg/brand |
| Component | 51 | button/primary/bg |
| Text ↓ Background → | canvas | surface | subtle | brand-subtle | brand | accent |
|---|---|---|---|---|---|---|
| fg/primary | 16.0 AAA | 15.1 AAA | 13.6 AAA | Not used | Not used | Not used |
| fg/body | 13.0 AAA | Not used | 11.0 AAA | Not used | Not used | Not used |
| fg/secondary | 6.7 AA | 6.3 AA | 5.7 AA | Not used | Not used | Not used |
| fg/link | 7.7 AAA | 7.3 AAA | 6.6 AA | 6.4 AA | Not used | Not used |
| fg/warning | Not used | Not used | 8.3 AAA | Not used | Not used | Not used |
| fg/inverse | Not used | Not used | Not used | Not used | 7.7 AAA | 7.2 AAA |
Reflection
What I learned
- Taking on too much work isn't leading Halfway through, I caught myself rewriting teammates' sections as we pulled things together, which made me the bottleneck. So I moved the work earlier: we agreed on direction, scope, and reasons before anyone started drafting, and I kept it all in one shared brief (our findings, what we were still working on, and what we'd dropped). Whenever the work started drifting off-scope, the brief brought us back in.
- Leadership is hard to judge from the inside While I tried my best to be a "good leader", there were times I felt like I wasn't leading the team well, like when we felt confused or took the project in different directions. At the end, one teammate told me unprompted that I was a good leader, which meant a lot. This taught me to check in more with my peers, not for reassurance, but to make sure they are getting what they need.
- Moving in reverse is still moving It was hard to backpedal on wait times, features, and stylistic directions. However, I learned to hold ideas loosely and see pivots for what they really are: improvements!