Most e-learning platforms are adapted for mobile as an afterthought. UpGrade was designed mobile-first from day one — validated with real users before a single high-fidelity screen was made.
Most e-learning platforms are built web-first, then adapted for mobile as an afterthought. But research showed people don't learn that way anymore — they want to learn in 5–10 minute pockets, on their phone, without falling into a social media scroll instead.
Real finding from interviews: users specifically wanted shorter lessons "so I don't get stuck in social media," and said existing platforms "don't adapt well to the phone."
That gap — a course library built for desktop, used mostly on mobile — became the design problem.
Five in-depth interviews surfaced the pain points; a 28-person survey and a Kano pass turned them into a prioritized, buildable v1.
"When choosing a course, it's important to read reviews first. However, finding comprehensive reviews is often challenging, and only a basic estimate is available, which doesn't provide enough information."
"Not all well-known platforms have a good adaptation to the phone, which forces you to look for other alternatives. As well as bad design."
"Subtitles will be great to add. And shorter in time videos to choose from when you do not have much time, but want to learn rather than sit in the social media."
After research, I brainstormed every possible idea, then ran it through the Kano model to see which features were first-priority and which weren't worth v1's time.
After selecting the main features to implement, I mapped the full information architecture to structure everything — from onboarding through home, search, learn, and profile.
Predefining the sequence of steps for each core goal — starting a class, subscribing — kept problems to a minimum while working through the interface, and excluded every deadlock before it could be built.
A low-fidelity prototype let me test usability with real users via Maze before committing to visual design — surfacing pain points and navigation issues while they were still cheap to fix.
Testing showed most people completed each mission via the expected path, with good time-on-screen across missions and no screens flagged for rework — the signal to move on to high-fidelity design.
The Jobs-to-be-Done method kept focus on cause and effect — where, and under what conditions, someone would actually reach for the app.
When I choose a course to study, I want to know more about the course lecturer to check their qualifications.
When I have 5–10 minutes, I want to take some lectures so that I don't get stuck in social media.
When I'm learning in an app, I want to see my progress to know I'm understanding the material well.
When I struggle to maintain a regular study schedule, I want to track my progress to finish the course.
When I study a course, I want to study in my native language to better understand the material.
When I choose a course, I want to rely on reviews to quickly eliminate bad options.
Three calls shaped the product more than any single screen — each one a direct answer to something research surfaced, not a default.
Courses are structured around short, completable sessions rather than long-form video — directly answering the "I don't want to get stuck in social media" insight from interviews. Every lesson is built to fit inside a 5–10 minute pocket of time.
Testing showed users cared more about seeing progress than about gamified rewards, so the design leaned into visible progress — completion bars, scores, saved classes — instead of badges or streaks.
The Kano pass killed several "nice to have" ideas — offline mode, heavy social features — before they cost any design time, keeping v1 scope tight enough to hit a clean, testable release.
The full UI system — onboarding, discovery, learning, and account — built mobile-first and tested before a single screen went to high fidelity.
Ran 5 in-depth interviews plus a 28-person survey to validate the mobile/bite-size hypothesis before designing anything.
Cut anything that wasn't a must-have or high-leverage for a v1 — before it could cost design time.
Mapped the information architecture and user flow for a 60+ screen product, then tested a lo-fi prototype with 28 users before touching visual design.
Components, type, and grid, plus final screens for onboarding, course discovery, learning, and progress tracking.
Testing the lo-fi prototype early meant the real design work started from a proven direction, not a guess.