Card Sorting for Information Architecture: Complete Guide

Information Architecture (IA) is the backbone of user experience. Card sorting is the most reliable method to build IA that matches how users think. Here's everything you need to know.

What is Information Architecture?

Information Architecture (IA) is the structure and organization of content, features, and navigation in digital products.

Good IA helps users:

✅ Find what they need quickly
✅ Understand where they are
✅ Predict where to find things
✅ Complete tasks efficiently

Bad IA causes:

❌ Frustration and confusion
❌ High bounce rates
❌ Low conversion
❌ Increased support costs

Card sorting is the #1 method for discovering user-centered IA.

Why Card Sorting for IA?

Traditional IA Methods (Problematic)

❌ Designer's intuition

Based on one person's mental model
Often doesn't match users
Hard to justify to stakeholders

❌ Stakeholder opinions

Reflects org chart, not user needs
Political compromises
Internal jargon

❌ Best practices from other sites

Your users might be different
Your content might be different
Context matters

Card Sorting Approach (Evidence-Based)

✅ User-centered

Based on actual user mental models
20-30 users reveal patterns
Removes designer bias

✅ Data-driven

Quantifiable results (similarity matrices, agreement scores)
Easy to present to stakeholders
Justifies design decisions

✅ Efficient

Run study in days, not weeks
Get clear patterns from 20+ participants
Cheaper than usability testing

The IA + Card Sorting Process

Phase 1: Prepare (1-2 days)

Step 1: Audit Your Content

List everything that needs organizing:

Pages or screens
Features or functions
Content pieces
Navigation items

Example (E-commerce site):

- Product pages (300+)
- Category pages (35)
- Support articles (50)
- Company info pages (10)
- Account features (15)
Total: 410 items to organize

Step 2: Select Representative Cards

You can't test 410 items. Select 30-50 representative cards:

Selection criteria:

High-traffic pages
Core features
Problem areas (analytics show issues)
Diverse content types

E-commerce example (reduced to 35 cards):

Products (sample from each category):
- Men's Running Shoes
- Women's Yoga Pants
- Kids' Winter Jackets
- Sports Equipment
- Outdoor Gear

Features:
- Shopping Cart
- Wishlist
- Order Tracking
- Returns Portal
- Size Guide

Support:
- Contact Us
- Shipping Info
- Return Policy
- FAQ
- Live Chat

Account:
- My Orders
- Account Settings
- Saved Addresses
- Payment Methods

Step 3: Write Clear Card Names

Good vs Bad card names:

❌ Bad (vague)	✅ Good (specific)
Resources	Video Tutorials
Solutions	Pricing Plans
Platform	Dashboard
More	Account Settings
Info	Shipping Information

Rules:

Use plain language (no jargon)
Be specific (avoid generic terms)
Keep it short (2-5 words)
Use consistent style
Avoid overlapping concepts

Phase 2: Run Card Sort (3-5 days)

Step 1: Choose Study Type

For IA, pick based on your situation:

Open Card Sort (discover new structure):

✅ Redesigning from scratch
✅ Don't know best categories
✅ Want unbiased user input
✅ Exploratory research

Closed Card Sort (validate existing):

✅ Testing proposed IA
✅ Have hypothesis to validate
✅ Comparing options A vs B
✅ Evaluative research

Hybrid Card Sort (test + discover):

✅ Have ideas but open to changes
✅ Want to validate & explore
✅ Iterative refinement
✅ Balanced approach

For most IA projects: Start with Open, then validate with Closed.

Step 2: Recruit Participants

How many:

Open sort: 20-30 participants
Closed sort: 30-40 participants
More is better, but diminishing returns after 30

Who to recruit:

✅ Target users (not colleagues!)
✅ Mix of experience levels
✅ Represent key user segments
✅ People who will actually use the product

Where to find them:

Existing customers (email list)
User research panels
Social media
UserTesting.com, Respondent, Prolific
Friends/family (last resort, less ideal)

Step 3: Set Up Study

Create your card sort study in 5 minutes →

Instructions template (for open sort):

We're redesigning [Product Name] to make it easier to use.

Please organize these [items/pages/features] into groups
that make sense to you. Create category names that describe
each group.

There are no right or wrong answers—we want to understand
how you think about these items.

This should take 10-12 minutes.

Thank you for helping us improve!

Step 4: Launch & Monitor

Send study link to participants
Monitor responses in real-time
Watch for confusing cards (low agreement)
Send reminder after 3 days
Close study when target reached

Phase 3: Analyze Results (1 day)

Step 1: Review Similarity Matrix

The similarity matrix shows how often cards were grouped together.

What to look for:

Dark clusters (over 70% agreement) = strong groupings
Light areas (under 40% agreement) = cards don't belong together
Isolated cards = may need rewording or different placement

Example:

Cards with 80%+ co-occurrence:
- "My Orders" + "Order Tracking" (89%)
- "Shopping Cart" + "Wishlist" (84%)
- "Contact Us" + "Live Chat" (81%)

→ These clearly belong in same category

Step 2: Identify Popular Groupings

Look at categories users created:

Example findings (E-commerce IA):

User-Created Categories (frequency):

1. "My Account" / "Account" / "My Profile" (78% of participants)
   → Strong consensus on this category

2. "Shopping" / "Shop Now" / "Products" (65%)
   → Multiple labels for same concept

3. "Help" / "Support" / "Customer Service" (82%)
   → Clear category, choose most common label

4. "Track Order" / "My Orders" (45%)
   → Split opinion: separate category or part of Account?

Step 3: Calculate Agreement Scores

Most card sort tools provide agreement metrics:

High agreement (over 70%):

Users have shared mental model
This structure will work well
Implement with confidence

Moderate agreement (50-70%):

Some variation in thinking
Consider user segments
Might need sub-categories or tagging

Low agreement (under 50%):

No consensus
Card may be unclear
Consider rewriting or removing
May need different approach (search, tagging)

Step 4: Look for Surprises

The most valuable insights are often unexpected:

Example surprise findings:

"Returns" grouped with "Shopping Cart" (not "Support") → Users think about returns while shopping
"API Docs" grouped with "Examples" (not "Technical") → Developers want practical examples, not just reference
"Pricing" grouped with "Features" (not "Company") → Users compare features + cost together

Don't ignore surprises—they reveal true user mental models!

Phase 4: Build IA (2-3 days)

Step 1: Create Category Structure

Based on card sort results, draft IA:

Example (from e-commerce study above):

Top-Level Categories:
├─ Shop (Products & Categories)
├─ My Account (Orders, Settings, Wishlist)
├─ Help (Support, Returns, Shipping Info)
└─ Company (About, Careers, Press)

Rules:

4-7 top-level categories (human working memory limit)
Clear, distinct categories (no overlap)
Logical hierarchy (broad to specific)
Consistent naming (parallel structure)

Step 2: Organize Content into Categories

Place each card into the structure:

Shop:
├─ Men's Clothing
├─ Women's Clothing
├─ Kids' Clothing
├─ Sports & Outdoors
└─ Sale

My Account:
├─ Orders & Tracking
├─ Wishlist & Saved Items
├─ Account Settings
└─ Payment & Addresses

Help:
├─ Contact Us & Live Chat
├─ Shipping & Returns
├─ Size Guides
└─ FAQ

Company:
├─ About Us
├─ Careers
└─ Press & Media

Step 3: Handle Edge Cases

Some cards don't fit cleanly:

Solution 1: Multiple Placements

Place in most common location
Add secondary access via search or links

Solution 2: Create "Other" Category

Only if truly necessary
Better: improve card names and re-sort

Solution 3: Breadcrumbs & Navigation Aids

Help users navigate between related areas
"Looking for X? Try Y section"

Step 4: Create IA Diagram

Visualize the structure:

Tools:

Lucidchart, Miro, Figma
PowerPoint or Google Slides
Even hand-drawn works!

Include:

All levels of hierarchy
Labels for each category
Number of items in each section
Navigation paths

Phase 5: Validate (1-2 days)

Don't skip validation! Card sorting shows groupings, but not findability.

Method 1: Tree Testing

Give users tasks, see if they can find items in your IA:

Example tasks:

"Where would you find your past orders?"
"Where would you go to return an item?"
"Where would you check shipping costs?"

Success: 80%+ users find items correctly

Method 2: Closed Card Sort

Run a closed card sort with your proposed structure:

Give users your category names
Ask them to place items
Check if placements match your IA

Success: 70% or higher agreement with your structure

Method 3: Prototype Testing

Build a clickable prototype and test with 5-8 users:

Can they complete key tasks?
Do they understand labels?
Any confusion or hesitation?

Method 4: A/B Testing (if possible)

Launch new IA to 50% of users, compare metrics:

Task completion rate
Time to find information
Pages per session
Bounce rate
Conversion rate

Real-World IA Examples

Example 1: University Website (150+ pages)

Before (department-centric):

├─ Admissions Office
├─ Registrar
├─ Student Services
├─ Academic Departments (25+ sub-pages)
└─ Administration

After card sort (task-centric):

├─ Apply & Admissions
│  ├─ Undergraduate
│  ├─ Graduate
│  └─ International Students
├─ Academics
│  ├─ Programs & Majors
│  ├─ Course Catalog
│  └─ Academic Calendar
├─ Student Life
│  ├─ Housing
│  ├─ Activities & Clubs
│  └─ Health & Wellness
└─ Current Students (Portal)
   ├─ Register for Classes
   ├─ Financial Aid
   ├─ Grades & Transcripts
   └─ Campus Resources

Result: 35% increase in task completion, 40% reduction in calls to admissions office

Example 2: SaaS Product (42 features)

Before (feature-centric):

├─ Data Management
├─ Analytics Engine
├─ Collaboration Tools
├─ Configuration
└─ APIs

After card sort (workflow-centric):

├─ Projects (where work happens)
│  ├─ Tasks & Boards
│  ├─ Files & Documents
│  └─ Team Discussion
├─ Insights (reporting)
│  ├─ Dashboards
│  ├─ Reports
│  └─ Data Export
├─ Team (people & permissions)
│  ├─ Members
│  ├─ Roles
│  └─ Activity Log
└─ Settings (configuration)
   ├─ Account
   ├─ Integrations
   └─ API Access

Result: 45% decrease in onboarding time, 28% increase in feature adoption

Common IA + Card Sorting Mistakes

Mistake #1: Testing with Internal Team

Wrong: "Let's card sort with our product team" Right: "Let's card sort with 25 target users"

Why: Internal teams have biased mental models based on how product is built, not how users think.

Mistake #2: Too Many Cards

Wrong: 80 cards = 45-minute study Right: 35 cards = 10-minute study

Why: Cognitive fatigue sets in after ~40 cards. Results become unreliable.

Mistake #3: Vague Card Names

Wrong: "Platform," "Solutions," "Resources" Right: "Dashboard," "Pricing Plans," "Video Tutorials"

Why: Vague cards get random groupings because users don't understand them.

Mistake #4: Skipping Validation

Wrong: "Card sort showed this structure, ship it!" Right: "Card sort suggests this. Let's validate with tree testing."

Why: Card sorting shows natural groupings, not necessarily findability. Validate before building.

Mistake #5: Ignoring Context

Wrong: "The card sort says X, so we must do X" Right: "Card sort says X, but our analytics show Y. Let's investigate."

Why: Card sorting is one data point. Combine with analytics, user interviews, business goals.

Mistake #6: Perfect Consensus Expectation

Wrong: "Only 60% agreement? The study failed!" Right: "60% agreement shows the primary pattern, 40% indicates sub-segments."

Why: Users are diverse. 70% or higher agreement is great. 50-70% is useful. under 50% needs investigation.

IA Card Sort Checklist

Before the Study

Audit content/features (list everything)
Select 30-50 representative cards
Write clear, specific card names
Choose study type (open/closed/hybrid)
Write clear instructions
Recruit 20-40 target users
Set up study in card sorting tool

During the Study

Send study link to participants
Monitor responses in real-time
Check for problematic cards (confusion)
Send reminders after 3-5 days
Aim for 20+ completions (open) or 30+ (closed)

After the Study

Review similarity matrix (look for clusters)
Identify popular groupings (over 70% patterns)
Note surprising findings
Draft IA structure based on results
Create 4-7 top-level categories
Organize all content into structure
Create IA diagram/sitemap
Validate with tree testing or prototype
Implement and measure impact

Advanced IA Techniques

Technique 1: Multi-Dimensional IA

Some content fits multiple categories.

Solution: Create multiple access paths

Primary navigation: Most common mental model
Secondary navigation: Alternative paths
Search & filters: For complex needs
Related links: Cross-connections

Technique 2: Personalized IA

Different user segments think differently.

Solution: Segment card sort results

Analyze beginners vs. experts separately
Create adaptive navigation
Offer "getting started" vs. "advanced" views

Technique 3: Faceted IA

Content has multiple attributes.

Example: E-commerce products can be organized by:

Category (shirts, pants)
Occasion (casual, formal)
Brand
Price range

Solution: Use filters and tags, not just categories

Technique 4: Task-Based IA

Organize by user goals, not content types.

Example: Government website

Not: "Department of Motor Vehicles" → "Forms" → "License Renewal"
Instead: "Renew Your License" (direct task access)

Technique 5: Progressive Disclosure

Don't show everything at once.

Solution:

Top nav: 4-7 main categories
Mega menus: Show sub-categories on hover
Contextual nav: Show related items based on current page

Tools & Resources

For Card Sorting

Free Card Sort (recommended)

Free plan: 3 studies, 50 responses each
Pro plan: $19/mo unlimited
Easy setup, great analysis
Start free →

For Tree Testing (Validation)

Optimal Workshop - Treejack Useberry UsabilityHub

For IA Diagrams

Lucidchart - IA diagrams and sitemaps Miro - Collaborative IA workshops Figma - Design and IA in one tool OmniGraffle - Professional IA documentation

Frequently Asked Questions

Q: How is IA different from UX or UI design? A: IA is the structure and organization (the skeleton). UX is the overall experience (how it feels). UI is the visual design (how it looks). IA comes first.

Q: Do I need card sorting if I already have analytics? A: Yes. Analytics show what users do, card sorting shows how they think. Both are valuable.

Q: Can I use card sorting for mobile app IA? A: Absolutely. Mobile IA is even more critical due to limited screen space.

Q: What if card sort results contradict business requirements? A: Find compromise. Use primary navigation for users, utility nav for business needs. Or educate stakeholders with data.

Q: How often should I redo IA research? A: Every major redesign, when adding significant new content, or if analytics show navigation problems.

Q: Can AI help with IA? A: AI can suggest structures, but real user card sorting data is far more reliable. Validate AI suggestions with users.

Next Steps

Audit your content - List everything that needs organizing
Create cards - Select 30-50 representative items
Run card sort - Start free study
Analyze results - Identify patterns and groupings
Build IA - Create structure based on findings
Validate - Test with tree testing or prototype
Implement - Build and launch
Measure - Track impact on user metrics

Related Resources

Ready to build better IA? → Start your free card sort study

Card Sorting for Information Architecture: Complete Guide (2025)