Case Study: Card Sorting to Uncover Users' Mental Models for Better Information Architecture

Client:
HealthTrack, a startup providing an innovative mobile and web-based platform for managing patient health records, appointment schedules, and telemedicine services. The platform serves healthcare providers, patients, and administrative staff.
Objective:
To redesign HealthTrack’s information architecture (IA) in a way that aligns more closely with users’ mental models, ensuring better usability, easier navigation, and an intuitive user experience. The goal was to improve how information is categorized and accessed within the platform by conducting a card sorting exercise.

Background
HealthTrack had been experiencing a growing number of user complaints regarding the organization of information within its platform. Healthcare providers, in particular, had difficulty locating patient records, medical histories, and scheduling tools, which led to inefficiencies in daily workflows. Patients also found it challenging to navigate between their health data and appointment scheduling features.
HealthTrack’s existing IA was built around the platform’s features, with menu items grouped by functionality (e.g., Patient Records, Appointments, Billing, Reports). However, this categorization did not align with how users naturally thought about their tasks or the language they used in their workflows. The development team needed to understand users’ mental models better and re-structure the IA to create a more intuitive experience.

Challenges Identified
Misaligned Information Structure:
Users had difficulty finding what they needed because the IA was organized from a developer-centric perspective rather than a user-centric one. Features were grouped by technical categories, which did not match the way users expected to interact with them.
Varying User Needs:
Different user groups (doctors, patients, and administrators) had distinct workflows and requirements, making it harder to design a one-size-fits-all IA. For example, doctors were primarily focused on patient records and medical histories, while patients were more concerned with appointment booking and accessing their medical data.
Cognitive Overload:
The platform’s menu was dense with options, and the layout didn’t make it clear which categories users should prioritize. As a result, many users felt overwhelmed by the number of choices available and struggled to find what they needed.

Design Approach: Using Card Sorting to Uncover Users’ Mental Models
The design team chose to use card sorting, a popular UX research technique, to help uncover how users mentally categorize information. By conducting a card sorting exercise, we would be able to observe users’ natural groupings of information and adjust the IA to fit their mental models. This approach would allow us to create a more intuitive structure that felt familiar to users and reduced cognitive load.
Step 1: Defining the Scope of the Card Sorting Exercise
The design team decided to focus on the most crucial sections of the platform—those that experienced the most complaints and were most heavily used, such as:
Patient Records: including history, test results, prescriptions, and more.
Appointments: scheduling, upcoming appointments, and past visits.
Billing: patient payments, outstanding balances, and insurance details.
Reports: lab reports, diagnostic information, and prescription records.
We also made sure to include elements from the platform that were frequently accessed by all user groups.
Step 2: Designing the Card Sorting Exercise
We used an open card sorting method, which would allow participants to freely categorize the cards and create their own groupings. Each card represented a distinct element of the platform (e.g., Test Results, Current Medications, Patient History, etc.). Participants were asked to group these cards into categories that made the most sense to them.
Participants: We recruited a diverse group of users, including healthcare providers (doctors, nurses), patients, and administrative staff. We made sure to gather feedback from all user personas to get a comprehensive view of the mental models across different use cases.
Tools: We used an online card sorting tool (OptimalSort) that allowed participants to sort cards remotely, making it easy to conduct the study across multiple time zones and user environments.
Step 3: Analyzing the Results
Once the card sorting exercise was complete, we analyzed the groupings and identified patterns in how users categorized different elements of the platform. Some of the key findings included:
Grouping Patterns: Many healthcare providers grouped items related to patient history, test results, and prescriptions together under a category they labeled “Patient Data” or “Medical Information.” However, patients often categorized these elements separately (e.g., Appointments, Medical History, and Health Records).
Language Preferences: Different user groups used slightly different terminologies. For example, healthcare providers often referred to “test results” and “patient history,” while patients preferred simpler terms like “my records” or “health data.”
Task-Driven Grouping: Both healthcare providers and patients grouped information according to the tasks they needed to complete. Healthcare providers focused more on diagnostic data and treatment records, while patients focused on appointment scheduling and accessing their medical histories.
Step 4: Redesigning the Information Architecture
Based on the insights gathered from the card sorting exercise, we worked to redesign HealthTrack’s IA with the following goals in mind:
Task-Based Groupings: We aligned information based on how users think about their tasks. For example, Appointments was placed alongside Medical History for patients, while doctors had their own Patient Dashboard that included Clinical Records, Test Results, and Treatment History in a more streamlined structure.
Simplified Categories: Instead of cluttering the global navigation with too many options, we consolidated similar features. Patient Records was divided into subcategories that matched the users’ mental models: Overview, Medical History, and Prescriptions.
Language Adjustments: We adapted the language to match user preferences. For example, healthcare providers’ “patient data” became “clinical records” for clarity, while patients’ “health records” became “my medical history.”
Step 5: Prototyping and Usability Testing
After restructuring the IA, we created a clickable prototype of the new design, which included the updated global navigation and task-oriented categories. We conducted usability testing with the same group of users to ensure the new structure met their needs:
Task Completion Rates: Testing showed a 40% reduction in the time it took to find critical information, such as appointment schedules or patient records.
User Satisfaction: Post-test surveys revealed that 85% of users felt the new IA was more intuitive and aligned with their tasks.
Error Reduction: Participants made fewer mistakes when navigating between different sections of the platform, which translated into a smoother user experience.

Results: Improved User Experience and Greater Task Efficiency
After implementing the changes, the following outcomes were observed:
Increased User Adoption: The redesigned IA saw a 30% increase in daily active users as the platform became easier to navigate.
Higher User Engagement: Users spent 25% more time interacting with the platform, thanks to easier access to important data and tasks.
Reduced Support Requests: Support tickets related to navigation and task difficulty dropped by 40%, indicating that users were able to find what they needed more efficiently.

Conclusion: The Power of Card Sorting for Uncovering Mental Models
This case study demonstrates how card sorting can be a powerful tool for uncovering users’ mental models and improving information architecture. By aligning the platform’s organization with the way users think about and categorize information, HealthTrack was able to enhance the usability of its application. The process provided valuable insights into how different user groups approach tasks, allowing the team to design a more intuitive experience that reduced confusion, improved task efficiency, and ultimately led to greater user satisfaction.