Rebuilding Information Architecture

Product

Internal P.E. Sharepoint Page

Timeline

2 weeks

Methods

Open Card Sort (unmoderated)

Context & Problem

Internal teams struggled with a department's Sharepoint page navigation due to content that grew organically over time, unclear structures, and inconsistent labeling. A card sort study had already been executed, but variability in setup required careful analysis to ensure reliable insights.

The core challenge: how do we deliver a defensible IA when the underlying data is inconsistent?

Research Goals

Understand how users mentally group platform content
Identify intuitive category structures
Surface gaps and misaligned terminology
Provide a clear foundation for restructuring the platform

My Role

Led analysis and delivery of IA recommendations

Conducted dendrogram analysis
Evaluated cluster strength and agreement
Translated findings into actionable IA

Impact

Improved category structure and labeling
Increased navigation clarity and findability

Example: High-agreement (75%) clusters informed clear groupings:

Old file names (Dev Team Roster & Org Functional Area) could be grouped under new category titles (Org or Team)

Outcome: Delivered a scalable, user-aligned IA structure.

Reflection

I stepped into this study after it had already been designed and launched, which meant working within an existing card sort setup that introduced variability in how participants grouped and labeled content.

To account for this, I took a more interpretive approach during analysis - looking beyond the dendrogram to assess cluster strength, areas of disagreement, and participant intent. This allowed me to focus on patterns that were meaningful and actionable despite inconsistencies in the data.

If I had been involved earlier in shaping this study, I would have approached the research design differently to better align with the system’s existing structure and reduce ambiguity in the data.

Method selection
- Since the platform already had an established hierarchy, a tree test would have been a more direct way to evaluate findability and validate the existing IA, rather than introducing additional variability through an open card sort.
Card sort design rigor
- If a card sort were still used, I would have prioritized clearer definition of cards through early stakeholder alignment - using workshops to ensure consistency in how content was represented.
Iterative validation
- I would have followed the card sort with a tree test to validate the navigation structure.

Takeaway: Strong research depends on both intentional study design and the ability to adapt when real-world constraints introduce variability.