All work
Resume

Course project · A real brief from NTU PACE staff

NTU PACE

A course discovery redesign that lets learners search by goal instead of by NTU department.

Role
Product designer. Led the research and the presentation to NTU PACE executives
Scope
Redesign frontend to encourage course discovery
Timeline
3 months
Methods
Card sorting, usability testing, first-click test
Tools
Figma, Figma Make, Maze
View live prototype

TL;DR

NTU Academy for Professional and Continuing Education
UX Research Information Architecture Responsive Design

THE PROBLEM

NTU PACE attracts more than 268,000 visitors/year,
but 75% leave without exploring courses.
Role

Product Designer, led research and presentation

Tools

Figma, Figma Make, Card sorting, Maze

Deliverables

Presented findings & prototype to NTU PACE Executives

IMPACT

85% Improved task time
10% Reduced drop-off rate
200k - 1.2M est. recovered revenue

THE GAP

Learners search by goal. NTU organizes by department.

.

.

.

Confusing labels

11

12

Lost on “FlexiMasters,” “Stackable Certificates,” “Professional Executive Certificates”

"I don't know what all these categories mean."

– M, Aerospace Engineer

Poor Navigation

> 50%

Couldn't find courses by outcome

No Memory

No way to save and return

1.

Match structure to how people search, not how NTU is organized

  • Goal-based hero search replaces department categories
  • Users start exploring immediately in a way that matches how they think
493s→72s
Time taken to complete task
2.9→1.7/ 5
Difficulty rating

SOLUTION

2.

Remove the re-orientation cost

  • Sticky header: key sections stay visible
30%→53.3%
In Home page exploration

SOLUTION

3.

Decisions span sessions, not one visit

  • Save courses, no sign-in required
710 users
First click test
4.75
Helpful rating

SOLUTION

THE TRADE-OFF

Not every tested idea made it in.

Proposed: a preview overlay

A hover/tap preview so learners could see key course info at a glance, no new page or tab.

Cut after testing

Barely used in testing, and it added an extra click of its own, the opposite of what it was meant to solve.

Good user research cuts your own ideas too, not just the ones you're testing on someone else's behalf.

BUSINESS

IMPACT

Recovering5 - 10%of lost visitors=10,000 - 20,000learners
At1 - 3%conversion=100 - 600additional enrolments/year
Estimated revenue=$200k - 1.2M/year
View Live Prototype

WHAT I LEARNT

  • Prioritised structure over personalization — highest leverage, given timeline
  • Card sort exposed a mental-model gap the institutional team missed
  • Untested assumption: does smart search get used, or does everyone default to browsing?

Context

NTU PACE gets more than 268,000 users a year, and 75% leave without exploring a course.

NTU PACE staff asked students to tackle that problem as a real brief. Learners search by goal. The site was organised by department.

NTU PACE’s own 24/25 report set the baseline the redesign is measured against:

Metric24/25
Unique users268,000+ a year
Page views1.1m
Left without exploring courses75%
Homepage visitors who complete the enquiry form2.8%
Bounce rate34.92%
Average session72 seconds, 2.33 views
Course pages vs homepage3.5x longer sessions
Traffic58% organic search, 23% direct
Devices54% desktop, 44% mobile

Source: NTU PACE Google Analytics, Monsido and Acquire Optimize report, 24/25.

Research

Card sorting and usability tests showed 3 problems:

  • Confusing labels. 11 of 12 participants got lost on “FlexiMasters”, “Stackable Certificates” and “Professional Executive Certificates”.
  • Poor navigation. More than 50% couldn’t find courses by outcome.
  • No memory. No way to save a course and come back to it.

“I don’t know what all these categories mean.”

M, aerospace engineer

The card sort exposed a mental-model gap the institutional team had missed.

Decisions

3 changes, each tested against the old site.

  1. 01 · Search by goal

    Match how people search, not how NTU is organised

    A goal-based hero search replaces the department categories, so learners start exploring straight away.

    493s → 72s task time · 2.9 → 1.7 difficulty out of 5

    New homepage with a goal-based course search

  2. 02 · Sticky header

    Remove the re-orientation cost

    A sticky header keeps the key sections in view while learners scroll.

    30% → 53.3% home page exploration

    Homepage with a sticky section bar and recently viewed programmes

  3. 03 · Save courses

    Decisions span sessions, not 1 visit

    Learners save courses without signing in, then pick up where they left off.

    7 of 10 found it in a first-click test · 4.7 / 5 helpful

    My Favorites page with saved courses

Trade-off

Cut after testing

A preview overlay

I proposed a hover or tap preview so learners could see key course info without opening a new page. Testers barely used it, and it added a click of its own, the opposite of what it was for. I cut my own idea.

Outcomes

Measured in testing

Task time down 85%, from 493s to 72s. Drop-off down 10% against the 24/25 baseline, where 75% left without exploring courses. I presented the findings and prototype to NTU PACE executives.

85%faster task time
10%lower drop-off
$200k to 1.2Mest. revenue a year

How the estimate adds up

StepFigure
Lost visitors recovered5 to 10% = 10,000 to 20,000 learners
Conversion1 to 3% = 100 to 600 more enrolments a year
RevenueEst. $200k to 1.2M a year

The revenue uses about $2,000 per enrolment: a bottom-up average of the course fees on the NTU PACE site.

Reflection

  • I prioritised structure over personalisation. It had the most leverage for the timeline.
  • The card sort exposed a gap the institutional team missed.
  • Untested assumption: do learners use the smart search, or does everyone default to browsing?

Try the prototype. whole-metric-22560084.figma.site/

View live prototype