Brilliant — Qiling (Riley) Chen

Overview

Leading research on a live B2C product.

I led UX research on a strategic redesign of Brilliant, a B2C SaaS platform for interactive STEM learning. The work added adaptive AI-assisted review and a real-time scaffolding system to cut the cognitive friction that loses non-technical learners.

Timeline: 1 month
Roles: UX Research
Visual Design
Prototyping
Team: Qiling Chen
Minwei Wu
Heer Patel
Noa Heilveil
Tools: Figma
Adobe XD
Google Form
After Effects

Research & Insights

Where the current product loses people.

Forum discussions and user feedback gave us two problems to test.

Scroll

Pain point: lack of autonomy and emotional buffers, with learner quotes about notification streaks over screenshots of the current app.

Pain point: inconvenience of reviewing, with learner quotes about recalling past mistakes over screenshots of the current app.

Planning our approach

Competitive audit — To map what the field already does well, and where every platform leaves the same gap.

Interview — To hear what learners feel, which the analytics never record.

Emotional journey map — To locate the point where that feeling turns into quitting.

Step 01

Competitive audit

I split the field into direct and indirect competitors, then scored every platform against the same criteria.

Step 02

Interview

12 university students, 218 data points. 15–30 minutes each. We performed thematic coding on the transcripts, tagging every sentence of individual user needs.

Step 03

Persona & Emotional Journey Map

Persona card for Luna, a senior digital art major: “I have the vision, but I lack the technical vocabulary to bring it to life.”

Emotional journey map, before and after: curiosity to frustration to disappointment, redrawn as excitement to reassurance to confidence.

Ideation

Sorting users’ needs.

Feasibility priority ranking retention, guidance, clarity and validation, beside a count of how often each need was mentioned.

How might we?

Brainstorming

Scroll

Prototype

Four features, answering the three questions.

The four features: Calendar Goal, Overview, Ask Koji, In Review — each with its component parts.

New features’ user flow

User flow from log-in through course, overview, in-review and notes, showing where Ask Koji and the calendar attach.

Sketch to mid-fidelity

Hand sketches exploring the calendar, reminders and review flows. — Hand sketches

Mid-fidelity screens for the pre-test, overview and in-review flows. — Mid-fidelity screens

High-fi demonstration

Each of the four features, running in the working prototype.

Feature 01

Calendar Goal & Reminder

Set a goal time in the calendar and get reminded, so study becomes a routine.

Feature 02

Overview & Pre-test

A knowledge pre-test and a clearly structured curriculum, so learners spend time on what is actually right for them.

Feature 03

Ask Koji, the in-course chatbot

A Q&A button and note-taking built into the lesson, so no question has to be left for later.

Feature 04

In Review: goals and mistake correction

Set goals for each day and week to establish a routine and earn rewards; collect and review mistakes at any time, so the effort of reviewing stops outweighing the benefit.

Test

Six users against four prompts.

Scale: 6 Users pilot tested our digital features against different information architecture.
Goal: To understand pain points of university students while using Brilliant and other e-learning platforms.

Test results: idea prompt 93% understandable, action prompt 70% successful, feedback prompt 76% satisfaction, and 10+ insightful design prompts received.

The redesign tested well on comprehension and less well on completion. Learners understood what each feature was asking of them and still stalled partway through. So the structure holds: the calendar, the pre-test, the chatbot and the review log each landed as intended. What the next round owes them is fewer steps between intent and done.

Reflection

What I learned, and what could be better.

This was my first opportunity to apply a structured research framework in a real design context. It strengthened my ability to translate qualitative insight into clear design direction, and working on an existing product taught me to identify gaps and prioritise within real constraints.

A compressed timeline limited cross-team coordination during prototyping, so the visual language was not fully standardised. Next time I would protect the final production stage and establish clearer design ownership earlier.