Redesigning the Human Resources portal for the City of Saint Paul, improving navigation and user engagement through research-driven information architecture.
Client
City of Saint Paul
Role
UX/UI Designer & Researcher
Team
Hamza Mohammed
Duration
Sep 2023 – Dec 2023
The City of Saint Paul is a city that radiates with rich history, tradition, diversity, and economic vitality. The city has a strong commitment to preserving and enhancing the urban villages of its residents with the likes of the thriving Downtown area and its revitalized riverfront.
The City of Saint Paul plays an immense role as a commercial hub with many Fortune 500 companies such as 3M, Target, EcoLab and many more with numerous small businesses playing an immense role in the city's economy.
The City of Saint Paul is not only the seat of government for Minnesota but also a cultural heart and a city that many of its residents call home. The City of Saint Paul offers numerous opportunities for its residents such as Public Safety, Community Development, Educational Support, and much more.
When I arrived
When I joined the Human Resources team as a UX Research Intern, the team had already put in a lot of groundwork. There were organized user reviews of the existing HR pages and several rounds of usability tests that had been completed before I got there.
What I was handed
My role was to take that existing research and turn it into something actionable. The team needed a clear navigation strategy for the HR section, and after talking it through together, we landed on a Card Sorting Activity as the right method to move forward.
What was already in place
User Reviews
Collected feedback from residents and city employees on the existing HR pages, covering what worked and what felt confusing.
Usability Tests
Multiple rounds of testing had been run on the current HR pages, surfacing where users were getting stuck and losing confidence.
Organized Research
All findings had been documented and grouped by the team, giving me a solid foundation to build on rather than starting from scratch.
The task I was given
Create an efficient navigation
The core ask was to take the sprawling HR section and give it a structure that actually made sense to the people using it.
Run a Card Sorting Activity
We chose this method to let participants show us how they naturally group information rather than guessing what would work.
Build an IA and Site Map
The output of the card sort would feed directly into an Information Architecture and Site Map grounded in real user data.
The City of Saint Paul's Human Resources portal had grown organically over years. What started as a manageable set of pages had expanded into a fragmented, hard-to-navigate structure that left both city employees and residents struggling to find critical HR information.
Scattered Structure
Dozens of subpages lived in inconsistent locations with no clear hierarchy, making it nearly impossible for users to predict where information would be.
Unheard Users
Previous usability tests and user feedback had surfaced recurring pain points. No structured action had been taken yet to address them systematically.
No Mental Model
The existing navigation did not reflect how users actually thought about HR resources. Categories were defined by internal org structure rather than user needs.
Working Backwards from the Problem
Reviewed Existing User Feedback
Dug into the team's prior usability test recordings, survey responses, and annotated user reviews to identify the most repeated pain points across different user groups.
Identified Recurring Themes
Synthesized feedback into clusters. Users consistently reported confusion around benefits, job postings, and policy documents being buried under unrelated categories.
Defined the Target User Groups
Mapped out the two primary audiences: city employees seeking HR support and external residents exploring job opportunities. Each group had distinct navigation needs.
Designed the Card Sorting Activity
Built an Open Card Sorting study to let real Saint Paul residents reveal their mental model. The goal was to understand how they naturally think, not impose an internal org structure onto the navigation.
Built the IA from the Ground Up
Used card sort results to construct an Information Architecture and Site Map that reflected how users actually think, not how the city internally categorizes its departments.
“The navigation was not broken because of bad design. It was broken because it was never designed around the user. My job was to listen to what the data was already saying and build a structure that finally reflected it.”
— Hamza Mohammed, UX Research Intern
Conducted with 15 participants, the majority of whom live within Saint Paul, giving us direct insight from the people this portal is actually built for.
Overview
Card sorting is a user-centered research method used to design or evaluate information architecture. Participants are given a set of cards, each representing a piece of content or a feature, and asked to group them in a way that feels natural. The goal is not to test the participant but to understand how they think.
For this project, we chose an open card sort so participants could define their own category names. This gave us a direct window into how Saint Paul residents and city employees mentally organize HR information, without projecting any structure onto them. The goal was to identify natural mental models, inform navigation structure and labeling, and improve the overall findability of content across the portal.
Instructions given to participants
Group items naturally based on your intuition
You may modify or reorganize groups at any time
Create new or uncategorized groups if needed
Provide reasoning for your groupings
Ask questions and give feedback throughout
Process steps
Cards are distributed
Each participant receives a set of cards representing features, content types, or concepts from the HR portal.
Participants review everything
Before grouping anything, participants take time to read through all the cards and get familiar with the content.
Grouping begins
Participants start placing cards into clusters that feel natural to them, based purely on their own intuition.
Groups get labeled
Once a cluster feels right, participants give it a name. In an open sort, they choose the label themselves.
Refinement happens
Participants revisit their groupings, move cards around, and adjust until the structure feels right to them.
Facilitator observes
Throughout the session, the facilitator watches for hesitation, questions, and moments of confusion without interfering.
Patterns are analyzed
After all sessions, groupings are compared across participants to find consistent mental models and naming conventions.
View the full card sorting board
All participant groupings and affinity clusters are documented in FigJam
Turning raw card sorting data into meaningful structure through affinity diagramming.
Context
After the open card sorting sessions wrapped up, we had raw groupings from 15 participants, each reflecting how they personally organized job-related information: reviews, benefits, career growth, employment details, and more.
The raw data alone was not the answer. The real work was in reading across all of it, finding where participants agreed without knowing it, and turning those patterns into a structure that could actually guide the navigation.
Transition to synthesis
No single participant's grouping was taken at face value. Instead, all sessions were analyzed collectively using affinity diagramming, a method that surfaces shared mental models by clustering similar responses from different people.
The goal was to move from individual opinions to collective patterns, and from patterns to a structure that felt intuitive to the broadest range of users.
Synthesis process
Aggregate the Data
All participant groupings were pulled into a single FigJam workspace. Every card and every category label from every session was laid out together so patterns could be seen across the full picture.
Identify Patterns
We looked for cards that consistently appeared together across different participants. Items that multiple people grouped the same way were strong signals. Outliers and inconsistencies were noted separately.
Cluster Similar Concepts
Related categories were merged into broader themes. Labels like "Job Reviews," "Company Reviews," and "Position Reviews" all pointed to the same mental model and were collapsed into one cluster.
Define and Label Themes
Each cluster was given a clear, user-centered label. The language came from participants themselves, not internal HR terminology, so the final labels would feel natural to the people using the portal.
Resolve Ambiguities
Some cards appeared in multiple groups across participants. These were examined carefully to decide whether they belonged in one primary category or needed to be cross-linked in the final architecture.
Prioritize and Structure
Clusters were ranked by frequency and importance. Primary categories were separated from secondary ones, and a rough hierarchy began to take shape that would feed directly into the IA.
Output
Four refined categories emerged from the synthesis. Each one reflects how participants actually think about HR content, not how the city internally organizes its departments.
Jobs / Careers
Job Postings
Open Positions
Promotions
Career Paths
Internships
Job Reviews
Performance Reviews
Feedback Forms
Evaluations
Goal Tracking
Annual Reports
Benefits
Health Insurance
Dental & Vision
Retirement Plans
Compensation
Employee Perks
Employment / HR
Leave of Absence
Policy Documents
Onboarding
HR Contacts
Crisis Support
Insights
Reviews drive decisions
Users strongly associate reviews with the decision to apply or stay. This category needed its own dedicated space, not buried under a general HR umbrella.
Career growth is distinct
Participants consistently separated career development from day-to-day HR tasks. Growth, learning, and promotions formed their own clear mental model.
Some categories need sub-structure
The reviews cluster in particular had enough depth that a flat list would not serve users well. Sub-categories were flagged for the IA phase.
Design implication
Informs the information architecture
The synthesized clusters fed directly into the platform's IA. Each category became a primary navigation node, grounded in real user data rather than internal assumptions.
Navigation reflects user mental models
Because the final structure mirrors how users think, not how the city is organized internally, content becomes easier to find and the overall experience feels more intuitive.
View the full affinity diagram
All synthesized clusters and participant groupings are documented in FigJam
A top-down hierarchy built directly from the synthesized card sorting clusters, reflecting how users naturally navigate job-related information.
About this diagram
This information architecture was built directly from the synthesized card sorting clusters. Each primary branch reflects a mental model that emerged from participants, organized into a hierarchy that separates job discovery, role reviews, benefits, and public safety into distinct, navigable sections.
Color key
Interactive prototype demonstrating the redesigned HR navigation structure and user flows.
View Interactive PrototypeThe redesigned Talent and Equity Resources page is now live on the City of Saint Paul's official website.
View Live SiteNext Project
TutorD
Prime Digital Academy · Product Design