UX Foundations · Labs · 4 min read
Run a card sort
A guided lab: run an open card sort with paper cards. Choose 20 to 30 pieces of content, ask three to five people to group and name them, photograph the results, find the patterns and draft a simple structure. About an afternoon, with index cards and a pen.
The fastest way to stop arguing about the menu is to watch other people build it. In this lab you'll run a small open card sort with paper cards. You'll see people group things in ways your team never considered, and you'll collect their words for your labels. It's practice, not a final study.
What you'll do:
- Choose a product and 20 to 30 pieces of its content
- Make paper cards and run three to five short sessions
- Record the groups and the names people give them
- Find the patterns and sketch a first structure
You'll need: index cards (or paper cut into pieces), a pen, three to five people who could use the product, a table, a phone to take photos, about 20 minutes per person and an hour to analyse.
1. Choose the content
Pick a product you know but didn't design: a neighbourhood library, a gym, a local council, a food delivery app. List 20 to 30 things people come for: pages, services, products or tasks. Real studies often use 30 to 50 cards (NN/g, Card Sorting: Uncover Users' Mental Models); fewer is easier for a first try.
Example for a library: Renew a book · Opening hours · Children's story time · Wi-Fi password · Room booking · Lost card · Ebooks · Donate books.
2. Make the cards
- One item per card, written large and clearly. Number the back of each card (1, 2, 3...) so you can record results quickly.
- Plain words, the way a user would say it. "Renew a book", not "Loan extension module".
- No repeated words. If several cards start with the same word ("Online ebooks", "Online renewals"), people will group them by that word. Rephrase.
- No hints of a group on the card ("Kids: story time").
Shuffle the deck before each session.
3. Run each session (about 20 minutes)
- "I'm trying to understand how people group these things. There are no right or wrong answers."
- "Put cards that belong together in piles. Piles can be any size, and you can change your mind."
- "If you don't understand a card, put it to one side."
- "Please think aloud as you go." Then stay quiet and take notes on what they say.
- When they're done: "Now give each pile a name." Write each name on a blank card and put it on top.
- Ask: "Which cards were hard to place? Did any belong in two piles?"
Take a photo of the table before you pick anything up.
4. Record the results
For each person, write the groups using the card numbers:
| Person | Group name they gave | Cards |
|---|---|---|
| P1 | My account | 1, 6, 12, 15 |
| P1 | Visiting | 2, 4, 9 |
| P2 | Borrowing | 1, 7, 12 |
Also write down the cards they put aside, and one or two things they said.
5. Find the patterns
Put all the results side by side and look for:
- Strong pairs: cards that most people put together.
- Common group names: the words people used more than once.
- Problem cards: left aside, moved several times, or split between groups.
- "Other" piles: what ended up there, and why.
Then sketch a first structure: four to seven main groups, each with a label taken from people's words where possible, and the cards under each one. Mark the cards you're unsure about.
6. Check it quickly (optional)
Turn your structure into a simple paper tree: main groups on one sheet, subgroups on others. Ask two new people to find three things, with tasks that don't use your labels ("Your card is gone and you want to borrow a book today. Where would you go?"). That's a tiny tree test; see Card sorting and tree testing.
7. Check your work
Reflection questions: Which grouping surprised you? Which card did nobody know where to put, and what would you rename it to? With three to five people you're learning, not deciding: NN/g recommends at least 15 participants for a real card-sorting study. What would you want to confirm with more people?
Optional, with a tool: FigJam or any whiteboard tool works for remote sessions: one sticky note per card, and ask people to share their screen while they sort.
Related
Card sorting and tree testing · Information architecture · Test with three people · Content first