Card Sorting for Government Websites

Government websites serve everyone: 18-year-olds filing for the first time, 75-year-olds managing benefits, new immigrants navigating unfamiliar systems, small business owners dealing with regulations. No other category of website has a broader, more varied, or more consequential audience.

When a commercial website fails a user, they leave frustrated. When a government website fails a citizen trying to file for unemployment benefits or find emergency housing assistance, the stakes are different.

Card sorting is one of the most effective tools for building government navigation that works across this full range of users. Here's how to do it.

The Fundamental Problem: Departments vs. Citizen Needs

The most universal problem in government website information architecture is the mismatch between how government is organized and how citizens think about their needs.

Government is organized by department:

• Department of Transportation
• Bureau of Motor Vehicles
• Department of Revenue
• Health and Human Services
• Department of Labor

Citizens think by task or life event:

• "I need to renew my driver's license"
• "I need to file my taxes"
• "I just had a baby and need to register the birth"
• "I lost my job and need to apply for unemployment"
• "I'm starting a business and need to register it"

These are different organizing principles. A citizen looking to "renew my license" has no particular reason to know that this involves the Bureau of Motor Vehicles under the Department of Transportation. They just know the task.

Card sorting reveals whether your navigation speaks in department-language or citizen-language — and points the way toward a task-oriented structure that works for people who don't know how government is organized.

Choosing Cards for a Government Card Sort

Anchor your card selection in your top citizen tasks. Most government analytics teams track "top tasks" — the things citizens most frequently try to accomplish. These become your most important cards.

Categories to include:

High-frequency citizen tasks

• Pay property taxes
• Renew vehicle registration
• Get a copy of my birth certificate
• Apply for a business license
• Report a pothole or street issue
• Pay a parking ticket
• Apply for a permit

Life event services

• Register a birth
• Marriage license
• Death certificate
• Moving — update my address
• Voter registration
• Starting a business

Benefits and assistance

• Apply for unemployment benefits
• Food assistance (SNAP)
• Medicaid / health coverage
• Housing assistance
• Disability services

Licensing and permits

• Driver's license
• Professional license
• Building permit
• Business license
• Contractor registration

Courts and legal

• Pay a court fine
• Jury duty
• Small claims court
• Public records request

Emergency and safety

• Report an emergency (non-911)
• Emergency preparedness
• Disaster assistance

Plain language rule: Never use internal department names as card labels. Not "Revenue Administration Filing" — write "Pay my taxes." Not "DMV Services" — write "Driver's license" or "Vehicle registration." The card must describe what the citizen is trying to do, not what department handles it.

Recruiting for Genuine Representativeness

Government services must work for everyone who is legally entitled to them. This creates a harder recruitment challenge than most commercial research.

Who you must include:

• Older adults (65+): Government websites serve a high proportion of older users. Navigation designed primarily for 25-40-year-olds systematically fails this population.
• Lower digital literacy users: People who primarily use the internet through smartphones, who aren't comfortable with complex navigation, who may need more time and clearer labels.
• Non-native English speakers (for English-language sites): If your jurisdiction has significant non-native speaker populations, their navigation expectations and vocabulary may differ.
• Lower income participants: Not only a demographic reality for many government services, but also important because this population often has the most urgent need for government services to work correctly.

What to avoid:

• Recruiting primarily from government employees or contractors — their familiarity with department structure biases results
• Only recruiting digitally fluent participants in the 25-40 age range
• Recruiting from your own professional network (tends toward higher education and digital literacy than your actual user population)

Aim for 20–25 participants — slightly more than commercial research projects — to capture the wider range of mental models in a general population.

Running the Open Card Sort

Framing for government users:

"Imagine you're visiting your local government's website. Here is a list of services and tasks available. Please group these items the way you'd expect to find them organized on the website, and give each group a name."

Practical considerations for wider user ranges:

Keep the study simple. Government card sorts should use shorter, simpler card labels than commercial research. If you're testing with lower-literacy users, consider a moderated format where a researcher is present to clarify card meanings if needed.

For unmoderated remote studies, add brief descriptions to cards where the label alone might be ambiguous: "Business License — Apply for permission to operate a business in the city."

Time estimate: Allow 20–30 minutes. Government card sorts with broad demographics take longer than commercial ones — participants who are less familiar with digital research tasks sort more carefully and take more time.

What Government Card Sorts Reveal

The Life Event Clustering Pattern

The most consistent finding in government card sorting research: citizens group services by life event, not by department.

"Having a baby" clusters: birth certificate, health insurance for children, school registration, childcare assistance.

"Moving" clusters: change of address, voter registration update, utility transfer contacts, vehicle registration update.

"Starting a business" clusters: business license, tax registration, zoning permit, employer identification number.

These clusters cut across multiple departments. Organizing navigation by these life-event patterns rather than department names typically improves task success rates by 30–50% for new or infrequent users of government websites.

The "Find vs. Apply" Distinction

Citizens often sort differently depending on whether a card represents information vs. action:

• "Information about unemployment benefits" vs. "Apply for unemployment benefits"
• "Find a licensed contractor" vs. "Get a contractor license"

Users expect information and transactional services to live in different navigation sections. "Learn about" content goes with reference material. "Apply for" content goes with task-oriented services.

If your current navigation doesn't distinguish between informational content and transactional forms, card sorting will surface this clearly — participants will sort them into different groups consistently.

Department Name Confusion

Cards labeled with department names (even when using friendly names) consistently show higher variance in card sorting — participants can't agree where they belong. Cards labeled with citizen tasks show lower variance — participants agree where they'd put them.

Track disagreement in your similarity matrix. High disagreement = navigation confusion in the real world.

Applying Results: The Task-Oriented Navigation Model

Based on government card sorting research, the most effective information architecture patterns for government websites are:

1. Top-level: Life events or service categories Rather than department names, organize top navigation around: Residents, Businesses, Visitors, About [City/County/Agency].

Within "Residents": life event subsections (Family & Birth, Moving & Property, Cars & Driving, Health & Benefits, Work & Business).

2. Primary search instead of navigation: For many government services, a prominent search box — optimized for common citizen queries — outperforms complex navigation for infrequent task types.

3. "Most requested" or "Top tasks" shortcut: Put the top 8–10 citizen tasks directly on the homepage regardless of where they live in the nav hierarchy. Citizens shouldn't need to understand navigation structure to complete the most common tasks.

4. Department directory as secondary navigation: Departmental organization is still useful for people who need to contact a specific office. Keep it, but don't make it the primary navigation framework.

Measuring Success

Before and after your navigation redesign:

Task success rate: Run usability testing on your top 10 citizen tasks. Measure completion rate and time-on-task before and after.

Search volume: Internal site search queries for top tasks should decrease after a successful navigation redesign — users who can find things through navigation don't need to search.

Call center volume: Track the volume of "can't find on website" calls to your service center. This is often the clearest signal of navigation failure and the most direct business case for investment.

Accessibility: Test with screen readers and keyboard navigation after restructuring. Navigation changes that improve findability for sighted users sometimes create barriers for assistive technology users. Test both.

Government navigation affects the lives of real people trying to access services they're entitled to. Card sorting is a rigorous, evidence-based foundation for getting it right.

Run your first government card sort for free. Create a study at freecardsort.com → No account or payment required.