Personal Project
Mobile application design
Timeline
3 Week Sprint
Role
Product Designer & Product Manager
Team
Alisha Liu
Aspen Ma
Tools
Figma
User Testing
Notion
TL;DR
A three-week concept sprint for an REI app. I led the project across a team of three designers from research to hand-off. REI’s brief was broad, so our research narrowed it to one problem: climbers had route databases and challenge tracking, and no way to find partners they could trust. Testing returned 100% task completion and one finding that reshaped the core system: beginners couldn’t read the skill level tags, so we paired every grade with a plain-language level.
Research
A broad brief, narrowed by research
REI wanted to be the top brand in climbers’ eyes, with an accessible platform that keeps climbers informed whether they’re on a wall or not. Our market research narrowed it. Existing climbing apps, Mountain Project, Kaya Climb, and Vertical Life Climbing, all centred on route discovery and challenge tracking, with little to no social layer. Climbers wanted something the category wasn’t offering: a way to find compatible partners and share their climbing with other people.
In climbing, a partner question is a safety question. You’re trusting someone with a rope.
Problem
Two Personas and what we focused on
Research produced two: Climbing Calvin, the social climber looking for compatible partners, and Hardcore Henry, the driven climber focused on tracking progress. We prioritized Calvin because his persona matched the majority of the people we’d interviewed. The cost of that call was Henry, whose tracking needs became a minor feature instead of a core one.
That gave us the problem statement.
Process
Cutting onboarding to fund the flow that mattered
We sketched wireframes through a design studio, and the first pass focused on onboarding and profile creation. We cut it. None of it was directly relevant to the problem statement, and we spent the time on a dedicated find-a-climber flow instead. In a three-week sprint that trade is what kept the MVP focused.
We followed Jakob’s law throughout, borrowing familiarity from existing social platforms so the interaction model needed no teaching.
Testing with five climbers, and the tags that failed
We ran remote moderated usability tests with five participants, each given two tasks: find other climbers to connect with, and find a fun way to spend time bouldering with other people.
Every task completed. Four findings came out of it, two mild and two moderate. Users wanted the day of the week on events, not just the date. They expected event restriction information inside event descriptions. They prioritized knowing how far away other climbers were, more heavily than we’d weighted it. And beginners found the skill level tags confusing.
That last one mattered most. The whole product rests on climbers placing themselves accurately in a skill system, and a system beginners can’t read produces bad matches. In climbing, a bad match is a safety problem. We refined the classification into plain progressive language running from beginner to advanced.
Solution
The design that answered the findings
The high-fidelity design carried proximity information into user profiles and event details, climb logging with visibility into other climbers’ recent climbs, day-of-week indicators on events, and the refined skill classification.
Visually we used REI’s motif: neutral backgrounds with green reserved for calls to action, so the UI stays uncluttered and the brand appears at the moments of action. Climb, connect, conquer.
Reflections
What I learned
The taxonomy has to work for the least experienced person using it. We built a skill system that made sense to designers who understood climbing grades, and testing showed it failed the exact users whose safety depended on it. I now treat every classification and filtering system this way, including agricultural rating scales in my current work.
A strong MVP beats a pile of features. A targeted objective is what kept us on track over three weeks.
Design decisions need reasoning that lives outside the team. The more proof behind a call, the better it holds.
What I’d do differently
More validation. Another round of interviews and a second round of testing on the high-fidelity prototype would have cross-referenced our pain points and caught what we missed, especially around how users place themselves in the skill system. Five users is thin evidence for the mechanism the product depends on.
More delegation. I undertook too much of the work myself. Trusting my teammates further would have freed me to plan ahead and mind the overall quality.
Plain language solved the beginner problem and cost experienced climbers precision. Grades carry information that words don’t, so I’d explore showing the grade range beneath the plain-language label: beginners place themselves by the words, and climbers who read grades get the detail they’re used to.
Curious to know more?
I’m happy to chat more about my process over a call. Reach out to me at hello[dot]adidee[at]gmail[dot]com!
Next project
Designing intuitive wayfinding for a virtual reality classroom, grounded in how people orient in space.