Card Sorting for Healthcare Websites

When a patient can't find "make an appointment" on a hospital website, they call. When they can't find the pharmacy hours, they drive and discover it's closed. When they can't locate their test results, they worry. Healthcare navigation failures have costs that go beyond bounce rates — they create real-world friction for people dealing with health concerns.

Card sorting is one of the most effective tools for fixing healthcare website navigation, because it reveals the gap between how your organization names things and how patients think about them. This guide covers how to run it effectively.

Why Healthcare Navigation Is Uniquely Hard

Most websites have one audience with relatively similar mental models. Healthcare websites have at least five:

• Current patients managing ongoing care
• Prospective patients evaluating services
• Family members navigating on behalf of someone else
• Referring physicians looking for specialist contacts and protocols
• Health plan members checking coverage and providers

Each group uses different terminology, has different urgency, and arrives with different prior knowledge. A navigation structure optimized for referring physicians ("Cardiology Referral Network") is opaque to a patient searching for "heart doctor."

Card sorting helps you see how each audience group thinks — and where your current structure fails them.

The Healthcare Vocabulary Problem

The most common source of navigation failure in healthcare is vocabulary mismatch. Your organization thinks in departments and service lines. Patients think in symptoms, conditions, and life events.

Internal terminology (how healthcare organizations label things):

• Gastroenterology
• Ambulatory Care
• Women's Services
• Interventional Radiology

Patient terminology (how people search and think):

• Stomach problems / digestive issues
• Outpatient appointments
• Pregnancy / OB-GYN / gynecology
• Vein treatment / cancer treatment

Card sorting reveals this mismatch before it becomes a navigation problem. When you create cards using your internal labels and ask patients to sort them, you'll see exactly which terms cause confusion (cards sorted inconsistently or placed with uncertainty) versus which are clear (cards consistently grouped in the same cluster).

Selecting Cards for a Healthcare Card Sort

Choose 40–60 cards representing the most-accessed content on your site. Anchor your card selection in analytics: what do patients actually look for?

High-value card categories to include:

Service access

• Find a doctor
• Make an appointment
• Patient portal / MyChart login
• Get a referral
• Urgent care locations

Health information

• Conditions and treatments
• Health library
• Symptom checker
• After-visit instructions
• Preparing for a procedure

Administrative

• Pay my bill
• Insurance accepted
• Medical records
• Prescription refills
• Contact us

Location and access

• Parking
• Visitor hours and policies
• Directions
• Accessibility

Plain language principle: Write every card in the language a patient would use, not department names. Not "Otolaryngology" — write "Ear, nose, and throat." Not "Ambulatory Services" — write "Outpatient appointments."

If you're unsure which term to use, check your site search logs. The terms patients type when searching your site are exactly the terms they expect to see in your navigation.

Recruiting Healthcare Participants

This is where many healthcare card sorts go wrong: recruiting hospital staff, medical students, or health-literate colleagues as proxies for patients.

Staff have deeply ingrained mental models based on internal terminology. A nurse knows what "ambulatory care" means. A 65-year-old patient managing three chronic conditions may not.

Recruit for:

• Actual patients (past or current)
• Age range that matches your patient demographic (often skewing older)
• Mix of health literacy levels — don't only recruit people comfortable with technology
• Caregivers managing healthcare for a family member
• Geographic proximity to your facility or service area

Avoid recruiting:

• Hospital or clinic staff
• Medical students or nursing students
• People with healthcare backgrounds
• Exclusively tech-comfortable young adults

Aim for 15–20 participants. This gives you statistically meaningful pattern data while keeping the study manageable. If you're researching a patient portal specifically, add a screener question to ensure participants are existing patients.

Running the Study: Open First, Then Closed

Phase 1: Open Card Sort

Open card sorting asks participants to group cards into categories they create themselves and name. This reveals how patients naturally structure health content — without the influence of your existing navigation.

Setup in CardSort:

1. Create a new study and select Open Card Sorting
2. Enter your 40–60 content cards
3. Add a brief description: "You're looking for health information on a hospital website. Please group these items the way you'd expect to find them."
4. Leave the category creation open — don't suggest any category names
5. Share the link with your 15–20 participants

From the open card sort results, look for:

• Strong clusters: Cards that 60%+ of participants grouped together — these become your navigation categories
• Outliers: Cards consistently sorted alone or placed in many different groups — these may need to be renamed or reconsidered
• Surprising groupings: Cards you'd separate that participants consistently group together (reveals mental model differences)

Phase 2: Closed Card Sort

After the open sort, you have a proposed navigation structure based on how patients actually think. Use a closed card sort to validate it.

Setup:

1. Create a new study and select Closed Card Sorting
2. Enter your refined card list (you may trim or rename cards based on phase 1 findings)
3. Set the categories to match your proposed navigation structure
4. Ask participants to place each item where they'd expect to find it on a hospital website

A successful closed card sort shows 75%+ of participants placing high-priority items (appointment booking, patient portal, find a doctor) in the expected category. Items with lower agreement rates need clearer labeling or different placement.

Reading Healthcare Card Sort Results

Similarity matrix: Shows how often each pair of cards was sorted together. High similarity (80%+) means these items belong in the same navigation section. Low similarity for items in your current nav means your grouping doesn't match patient expectations.

Dendrogram: A tree diagram showing clusters. Read it from the bottom up — the strongest clusters (most consistent groupings) appear first. These become your primary navigation categories.

Category naming: The names participants gave to their groups in the open card sort are gold. If 12 out of 18 participants named a group "Find Care" or "Getting Appointments," that label works better than your internal term "Access Services."

Common Healthcare Navigation Fixes Card Sorting Reveals

Based on patterns across healthcare card sorting research:

"Find a Doctor" should be prominent and standalone: Patients consistently treat physician search as a primary task, not a subcategory of "Our Services." It almost always belongs in the top navigation.

Patient portal gets confused with patient education: "Patient Portal" (MyChart login) and "Patient Resources" (discharge instructions, health library) get conflated. Participants often sort them together but need them to behave differently. Consider renaming "Patient Portal" to "My [Health System Name]" or "Log In to Your Account."

Bill payment is isolated: Patients consistently put bill payment in its own category or near the top — not nested under "Patient Services" or "Administrative." The task has urgency that doesn't fit inside general service menus.

Urgent care vs. emergency room confusion: Participants frequently confuse these two services or expect them to be grouped together. A card sort can reveal whether your current labeling distinguishes them clearly.

Departments by condition, not specialty name: When given the choice, patients sort by condition ("Back pain," "Heart problems") rather than specialty ("Orthopedics," "Cardiology"). Consider condition-first navigation for large service menus.

After the Card Sort: Connecting to Outcomes

Card sorting findings should connect to measurable outcomes. Before deploying your new navigation:

Establish baseline metrics:

• Task success rate for top 5 patient tasks (measured by usability testing)
• Site search query volume for navigation-related searches
• Call center volume for navigation/location questions
• Patient satisfaction scores related to website usability

After deploying navigation changes, monitor the same metrics. Typical improvements in well-executed healthcare navigation redesigns:

• 20–35% reduction in "find a location" and "schedule appointment" search queries
• 10–15% reduction in call center contacts related to website navigation
• Improved task completion rates in usability testing

Card sorting is the foundation. Tree testing validates the resulting structure. Usability testing confirms real-world findability. All three are free on CardSort.

Running your first healthcare card sort? Start a free study → No credit card, no account required.