Card Sorting for Product Managers
Every PM has been in this meeting: the design team wants features in one place, engineering wants them somewhere else, and the head of product has a strong opinion based on what a competitor does. Everyone has an argument. Nobody has data.
Card sorting ends these debates. It tells you where your users expect to find features — which is the only answer that actually matters for navigation decisions.
And here's the part product managers often don't know: you don't need a dedicated UX researcher to run one. The full process — from study creation to results — is something a PM can do in an afternoon.
When Product Managers Use Card Sorting
Card sorting is specifically useful for product decisions about where things go. It doesn't tell you whether to build something, or how to design it — it tells you where users would expect to find it.
Use it when you're deciding:
- Where to put a new feature in your existing navigation
- How to restructure navigation when your product has grown beyond its original IA
- Whether to split or merge menus (is "Billing" and "Subscriptions" one section or two?)
- What to name categories so users can find what they're looking for
- How to onboard new user segments who may have different mental models than your original users
Don't use it to decide whether to build a feature. Don't use it to test UI designs or specific workflows. Card sorting is specifically about navigation structure and findability.
The PM's 3-Hour Card Sort
You can complete a full card sorting study — design, collect responses, analyze, decide — in a business day. Here's how:
Hour 1: Study Setup (30 minutes)
Step 1: Define the question
Be specific. Not "where should features go?" but "should analytics features be under the Reports menu or under each individual feature page?" A narrow question gets you actionable data.
Step 2: Create your card list
List the features or content items you're testing — aim for 30–50 cards. Use the language your users actually use:
- Check your in-app search queries for the vocabulary users type when they can't find something
- Use exact terms from support tickets: what words do users use when they say they can't find a feature?
- Avoid internal product names that users never see
Step 3: Choose open or closed
- Open card sort: Users create their own category names. Best when you don't have a proposed structure yet and want to discover how users think.
- Closed card sort: Users sort into your predefined categories. Best when you have a proposed navigation structure and want to validate it.
For most product decisions, start with closed if you have a proposed structure to validate, or open if you're starting from scratch.
Step 4: Set up in CardSort
Go to freecardsort.com, create a study, add your cards, choose your type. The whole setup takes 10–15 minutes.
Hour 1.5: Recruitment (15 minutes to send, then wait)
Send the link to real users. As a PM, you have multiple channels:
- In-app messaging: A targeted message to a user segment using the relevant feature area
- Email: Send to a segment from your CRM (filter by feature usage if relevant)
- Customer Slack or Discord: Post in user community channels
- Existing research panel: If you have a user research list, this is what it's for
- Customer success team: Ask CSMs to forward to a handful of customers they're actively in contact with
You need 10–15 responses for a reliable card sort. Most PMs can get this in a day by tapping existing user relationships.
If you can't reach users directly, CardSort's AI participant simulation can generate instant responses to pilot your study before sending it to real users. This is useful for checking that your card labels make sense before collecting real data.
Hours 2-3: Analysis (1-2 hours)
CardSort generates a similarity matrix and dendrogram automatically. As a PM, the key things to look at:
Similarity matrix: Each cell shows what % of participants sorted two items together. High agreement (70%+) = reliable finding. Low agreement = ambiguity you need to address.
Dendrogram: A tree diagram showing which items cluster together. Read from the bottom — the first branches that form are the strongest groupings.
Outliers: Features sorted inconsistently by multiple participants are navigation problems. They either need to be renamed, live in multiple places, or be reconsidered.
Making the Decision (and Documenting It)
This is where card sorting pays off differently for PMs than for UX researchers. You're not writing a research report — you're making a product decision and documenting the evidence.
Decision document format:
Decision: [Feature X] will be placed in [Navigation Section Y]
Evidence: Card sort with 14 participants (link to study). 71% of participants placed Feature X in the equivalent of Section Y. The next most common placement was Section Z (21%). Agreement is strong enough to proceed with confidence.
Alternatives considered: Section Z placement, based on [team member's] recommendation. Card sort data does not support this — users don't associate Feature X with Section Z's other contents.
This document answers future questions, gives design and engineering clear direction, and creates an evidence trail that's useful for retrospectives and future navigation decisions.
Three Card Sorts Every PM Should Run
1. New Feature Placement (Before Building)
Before you spec where a new feature lives in navigation, run a closed card sort. Create a study with your current feature set plus the new feature (labeled without giving away what it does — just what users will call it). Ask users to place every item including the new one.
Where users put the new feature tells you where to put it. If they consistently sort it into a place you didn't expect, that's a finding worth understanding before you ship.
2. "Why Can't They Find It?" Audit
When a feature has low discoverability (low usage despite positive reception in testing, support tickets from users who didn't know it existed), run a closed card sort with users placing features into your current navigation structure.
Features with low placement agreement in the current structure are navigation problems. Features that users put in a different section than where they live are mislocated.
3. Navigation Restructure Before Development
Before investing in a navigation refactor, use open card sorting to validate that you've correctly understood how users would structure your product differently. An open card sort with 15–20 users takes a fraction of the time and cost of redesigning navigation based on assumptions.
Card Sorting vs. User Interviews for Product Decisions
Both are valuable but answer different questions:
User interviews tell you why users have the opinions they have, what jobs they're trying to do, and what frustrates them about your product.
Card sorting tells you specifically where users expect to find features and how they mentally organize your product's capabilities.
As a PM, you need both — but for navigation decisions specifically, card sorting gives you quantitative, decision-ready data in a format that's easier to act on than qualitative interview insights. "71% of users expect this feature in Settings > Notifications" is clearer direction than "users seem to think notifications are important."
Getting Started Today
- Go to freecardsort.com — no account required
- Create an open or closed card sorting study
- Enter your feature list (30–50 items in user language)
- Share with 10–15 users
- Make a documented, evidence-based placement decision
The data you collect will settle the meeting you've been dreading, give your design team clear direction, and create a record of why you made the decision you made.
Product decisions deserve user evidence. Start a free card sort →