UX Research · Information Architecture
Research-led IA redesign for one of America's 10 largest public library systems.
81 locations. Two top-level nav categories for everything. I led research and architecture — validated by 45 participants before a single wireframe.
The problem
The CPL site organized everything — printing, research help, language support, children's programs — under Browse and Events. The content existed. The structure didn't.
Before
After
Who I designed for
Personas developed from content inventory — their needs and frustrations anchored every structural decision.
What I did
Tested whether my proposed nav categories matched users' mental models. Strong alignment on Tech Services and Attend — with one unexpected pattern.
Users consistently grouped "Spanish language resources" under both Find and Help — revealing the language access ambiguity that later drove a dedicated nav decision.
Validated the proposed sitemap with persona-driven scenarios — before any visual design began.
75% directness on tasks that succeeded — users who found the right answer took the direct path. Failures were in labeling, not structure.
What the data showed
Each task mapped to a persona's core need. Every finding under 70% drove an architectural decision.
"You need help with your research paper."
Lowest success rate. Users looked under Help — not Resources. A labeling failure, not a structural one.
Decision → Elevated Research Assistance to top-level Help. Decision driven by data, not assumption.
"English isn't your first language. Find the Spanish version."
Language toggle was buried in My Account. Non-English speakers gave up at the homepage — a critical accessibility gap.
Decision → Surfaced language selector to homepage level — persistent and visible, not nested.
"Find a suitable event for your grandchild this weekend."
Kids' events mixed with all events — impossible to scan without filters.
Decision → Added Kids / Teens / Adults audience filters visible from the Attend top-level nav.
"You need to make copies of your resume."
Users found printing under Tech Services. Path was correct — label needed clarification.
Decision → Renamed to "Print & Copy" — intent over category name.
What changed
Each of the six categories maps to a user intent identified in research — not a content type, not an org chart. Amber items below were structurally changed based on task failures.
New top-level navigation
Full sitemap + wireframes on Behance
All 6 categories, sub-levels, and annotated wireframes — open in Behance →
53% success on Task 02 flagged a critical failure before a developer touched a line of code. That's why you run the study first.
Full deliverables — card sort matrix, tree testing paths, and final sitemap — on Behance.
"The problem wasn't missing content.
It was missing structure."