Card Sorting for SaaS Products
SaaS navigation debt is real. Every sprint adds a feature. Every feature gets a nav item. After 18 months, you have a sidebar with 23 items, three different "settings" locations, and users filing support tickets asking where the reports went.
Card sorting is the most reliable way to understand how your users actually think about your product's capabilities — before you reorganize navigation, not after. This guide covers how SaaS teams run effective card sorts.
The SaaS Navigation Problem
SaaS products grow fast. Navigation structures designed for a 10-feature MVP don't scale to 50-feature platforms. The typical result:
- Features buried three levels deep because there was nowhere obvious to put them
- Multiple "settings" locations (global settings, workspace settings, user profile settings, integration settings)
- Power users who navigate via keyboard shortcuts or search because the nav is too slow
- New users who can't find core features during onboarding
- Support tickets that begin "I can't find..."
Card sorting doesn't fix these problems directly. It tells you how your users expect to find things — which is the foundation for fixing them.
What to Sort: Choosing Your Cards
The first challenge in a SaaS card sort is selecting the right cards. Your product has many features, but not all of them belong in a navigation study.
Include:
- Every item currently in your primary navigation
- Secondary navigation items users struggle to find (check support tickets)
- Features you're planning to add in the next roadmap cycle
- Settings and configuration options
- Reports and analytics views
Don't include:
- Transient UI elements (toasts, modals, tooltips)
- Features that are contextually surfaced (actions that appear when hovering over a row)
- Administrative functions only accessed during setup
Aim for 40–60 cards. More than 60 creates cognitive fatigue. Less than 30 won't reveal enough structure.
The Vocabulary Problem in SaaS
SaaS products often have internal feature names that don't match how users describe the same functionality. This is the single biggest source of card sorting insight.
Internal names vs. user language:
- "Automations" → "Workflows" / "Triggers" / "Rules"
- "Workspace" → "Organization" / "Account" / "Team"
- "Segments" → "Groups" / "Filters" / "Audiences"
- "Pipelines" → "Deals" / "Funnel" / "Sales process"
Check your support tickets, user interviews, and in-app search queries. The words users type when they can't find something tell you exactly what to write on the card.
Write cards using the language users use, not your product's internal naming. You can test whether your feature names match user vocabulary separately — but the card sort should reveal structure first.
Who to Recruit: The Two-Segment Strategy
SaaS products have a built-in user segmentation problem for card sorting: power users and new users have fundamentally different mental models.
Power users have learned your navigation through repetition. They know the product deeply but may have adapted to your current structure rather than revealing how they'd ideally organize it. They're good at validating that proposed structures don't break their existing workflows.
New users (under 30 days) approach your product with fresh mental models based on their prior experience with similar tools. Their sorts reveal how the category should be structured, not how it currently is.
The recommendation: run separate open card sorts with 10–12 participants from each segment. Compare the resulting clusters. Where they agree — that's your navigation backbone. Where they diverge — that's where you need additional context, onboarding guidance, or dual placement.
Setting Up Your SaaS Card Sort
Study type: Start with open card sorting. You want to discover emergent structure, not validate an existing one.
Study instructions: The framing matters. Try:
"Imagine you're new to [Product Name]. You're looking at a list of things the product can do. Please group these items the way you'd expect to find them in the app's navigation menu, and give each group a name."
This framing anchors participants to navigation (not conceptual categories) while keeping the exercise open-ended.
Time estimate: A 50-card open card sort takes participants 15–25 minutes. Acknowledge this in your recruitment message — SaaS users are busy and will abandon studies that take longer than expected.
Interpreting SaaS Card Sort Results
The Similarity Matrix
The similarity matrix shows how often each pair of cards was sorted together across all participants. For SaaS products, look for:
Strong clusters (70%+ co-occurrence): These items belong in the same navigation section. If "Integrations," "API Keys," and "Webhooks" are consistently co-sorted, they form a natural "Developer Settings" section.
Isolated items: Features that participants sort inconsistently — placed with different groups each time — are navigation liabilities. They likely need to be renamed, reconsidered, or given contextual placement instead of a fixed nav position.
Surprising groupings: Features you'd put in separate menus that users consistently group together. This is the most valuable finding — it reveals mental model mismatches you couldn't have discovered any other way.
Comparing Power User vs. New User Segments
After running both studies, compare similarity matrices side by side. Common patterns:
Strong agreement: Both power users and new users sort these features together. High-confidence navigation groupings.
Power user disagreement: Power users sort based on their learned workflow; new users sort based on conceptual categories. The new user pattern is usually more generalizable — power users can adapt, new users cannot.
New user confusion: Features that new users sort inconsistently but power users are consistent about. These need better labeling or progressive disclosure (hide advanced features until users reach them contextually).
Common SaaS Navigation Findings
Across SaaS card sorting research, several patterns appear repeatedly:
"Settings" fragments into multiple sections: Users don't treat all settings as one category. They separate "My Account" settings (personal preferences, notifications, profile) from "Product Configuration" settings (team permissions, integrations, data management). These should live in different places — one in user profile, one in workspace/org settings.
Reporting and analytics cluster with the features they report on: Users don't expect a global "Reports" section. They expect analytics about email campaigns inside the email tool, analytics about pipeline inside the CRM view. Context-specific analytics outperform consolidated reporting hubs in findability.
Collaboration features (sharing, permissions, comments) don't cluster cleanly: They're expected in multiple places depending on context. Rather than a single "Collaboration" nav section, surface these features contextually wherever the content lives.
Onboarding and "getting started" content clusters separately: New users sort setup guides, tutorials, and sample data import as a category. Power users ignore or exclude it. Consider a first-run experience layer separate from your permanent navigation.
Before and After: Measuring Navigation Improvement
Track these metrics before redesigning and again 30 days post-launch:
- Time-to-first-action for new user onboarding (how long to complete core workflow)
- Support ticket volume mentioning navigation or "can't find"
- In-app search usage (high search = navigation failure)
- Feature discovery rate for secondary features (% of accounts using the feature within 30 days of signup)
A well-executed navigation redesign based on card sorting typically improves new user task completion by 15–25% and reduces navigation-related support volume by 20–30%.
Running Your SaaS Card Sort Today
- Open freecardsort.com — no account or payment required
- Create an Open Card Sorting study
- Enter your feature list (40–60 items, user language)
- Share with 10–12 power users and 10–12 newer users separately
- Review similarity matrices for each segment
- Build your new navigation structure from the strongest clusters
The whole study setup takes less than 30 minutes. Results start coming in as soon as participants complete the sort.
Ready to find out how your users actually think about your product? Start a free card sort →