INFORMATION ARCHITECTURE
Flat, no hierarchy, bespoke one-off layouts — and every new tool launch added more of them the same way.
Every page on the site is exactly one of these. Adding a tool becomes a fixed recipe — no new patterns invented, ever.
CASE STUDY 04 · ENTERPRISE INFORMATION ARCHITECTURE
Improving the information architecture of an internal AI learning hub
AbbVie’s internal AI learning hub brings together resources, tools, and guidance to help employees learn about and apply AI in their work. As the content expanded, the existing structure became harder to navigate and users did not always know where to begin or which resources were relevant to them. My role was to evaluate the existing information architecture and recommend a clearer way to organize the experience using user research, content analysis, and stakeholder input.
ROLE
UX Researcher
TOOLS
Claude Code
Mural
Dovetail
OUTCOME
Recommended a task-based
navigation model
AI resources were difficult to find and navigate.
A site audit showed that resources were distributed across multiple pages, navigation paths, and content categories. Similar information appeared in different locations, making it difficult for users to understand where to begin or which resource was most relevant.
THE TWO DIFFICULT QUESTIONS
These two questions were hard to answer based on the structure of the Learning Hub.
“What can AI do for my job?”
“Which tool do I use for what?”
Multiple navigation paths created overlap.
Users could reach similar resources through different sections, making the overall structure harder to understand.
Content categories did not consistently match user goals.
Resources were often grouped around internal categories rather than what employees were trying to accomplish.
Users needed a clearer starting point.
The experience needed to help employees quickly identify where to begin and which resources were relevant.
I compared persona-based and task-based navigation.
The initial direction organized content around user personas. I explored a task-based alternative and evaluated both approaches against user research, existing content, and common employee goals.
Every new tool multiplies across every role section. Content duplicates, migration cost compounds, and the same page gets maintained five times.
Adding a tool is a fixed recipe: one capability page, optional learning path / prompt guide / events page, tag the training.
RECOMMENDATION
Use tasks as the primary navigation structure.
Research showed that users across personas often shared the same goals. Organizing the experience around those goals created clearer paths through the content while reducing unnecessary duplication.
Personas could still inform recommendations, examples, and contextual content without determining the site’s primary structure.
Why task-based navigation worked better
13 user interviews taken from Dovetail
Personalization by function and skill level, through filters and a guided assistant — not five static role pages.
A team card-sort exercise
Personas were already being framed as curated “AI for [role]” pages, never as navigation.
WHAT I HANDED OVER
WHAT I LEARNED
Users described goals more consistently than roles.
Participants often approached AI resources based on what they wanted to accomplish rather than which employee persona they belonged to.
The structure needed to support future content.
Task-based categories provided a clearer framework for adding new tools, resources, and learning content over time.
The recommendation balanced user and business needs.
Personas could support relevant recommendations and examples without defining the primary navigation.
THE OTHER CASE STUDIES
← BACK TO ALL WORK