The redesign
The work here was the information architecture, not the visual design. A screenshot can't show an IA, so this is the before and after side by side, with the tree-testing results attached to the tasks they changed.
6 → 5
top-level sections
32 → 18
leaf pages in the tree
36% → 85%
average task success
52 → 81
SUS, post-test
Before
About
- History
- Mission
- Facilities
- Contact
- Directions
Academics
- Undergraduate
- Graduate
- Minors
- Certificates
- Course catalog
- Advising
- Co-op
- Study abroad
People
- Faculty
- Staff
- Students
- Alumni
- Emeritus
Research
- Labs
- Publications
- Grants
- Seminars
News & Events
- News
- Events
- Calendar
- Newsletter archive
Resources
- Forms
- Room booking
- IT help
- Handbook
- Policies
- Committees
After
Programs
- Undergraduate
- Graduate
- Minors & certificates
- Compare programs
What you'd study
- Courses
- Co-op
- Labs & facilities
- Study abroad
Where it leads
- Outcomes & salaries
- Where graduates work
- Alumni stories
People
- Faculty
- Advising
- Contact
Staff & student tools
- Forms
- Room booking
- Handbook
- Committees
Tree testing, by task
Find out which degree covers accessibility
before34%
after89%
Find the co-op requirement for a BS
before41%
after92%
Find who to email about changing advisor
before22%
after78%
Find what graduates earn
before18%
after84%
Book a room for a student group
before63%
after81%
What the testing surfaced
Six top-level sections and 33 leaves meant nearly everything was three clicks deep and named after the department's own org chart.
Prospective students and current students were competing for the same navigation, and prospective students lost — the highest-traffic audience had the least direct path.
'Resources' was where anything without an obvious home went. Participants never guessed it contained room booking.
Nobody looked under 'Academics' for outcomes or salary information; they expected it near the programs themselves.
Note on the numbers: the direction and the conclusions are from the real project, but the per-task percentages here are illustrative sample data — the original study's raw results weren't available when this prototype was built.