What Is Card Sorting in Validation?
Card sorting is a research method where participants group and label cards — each holding a concept, feature, or piece of content — to reveal how they mentally organize information. It surfaces the customer's own categories and priorities, which you use to shape navigation, information architecture, and product structure.
Quick Answer: Card sorting hands people a set of cards and asks them to sort them into groups that make sense to them. Watching how they cluster and name those groups exposes their mental model — the categories and language they already use — so you can design around real understanding instead of guessing.
How card sorting works: group and label cards
Card sorting works by making an invisible thing visible: the way a customer already organizes a topic in their head. You write one idea per card, hand the deck to a participant, and ask them to arrange the cards into groups that feel natural — then explain their reasoning.
The mechanics are deliberately simple. Every card carries a single concept, and the participant does the organizing:
- Prepare the deck. Put one feature, task, topic, or piece of content on each card — usually 20 to 60 cards.
- Ask for groups. Have the participant cluster cards that belong together, using whatever logic makes sense to them.
- Ask for labels. Once grouped, ask them to name each pile in their own words. The labels are as revealing as the groupings.
- Ask why. Probe the reasoning behind edge cases and hesitations — that talk-aloud is where the mental model shows.
The output is not a vote or a score. It is a picture of the participant's mental model: which things they see as related, what they call them, and what they treat as an afterthought. Testing Business Ideas by Strategyzer lists card sorting among its discovery techniques for exactly this reason — it is a way to understand how customers think before you commit to a structure.
That framing matters. Card sorting is a qualitative discovery method: it tells you how people organize and prioritize concepts, not whether they will buy. Treat it as a way to sharpen understanding, never as proof of demand.
Open vs closed card sorts: which to run
The two main variants differ by who creates the categories. In an open card sort, participants invent their own groups and names — best when you are still learning how customers think. In a closed card sort, you supply fixed categories and participants file cards into them — best when you want to test a structure you already have.
The table below compares the two by what each is built to learn.
| Dimension | Open card sort | Closed card sort |
|---|---|---|
| Who defines categories | The participant, from scratch | You define them in advance |
| Best used when | You are discovering mental models | You are validating an existing structure |
| What it reveals | Natural groupings and the words people use | Whether your labels fit how people think |
| Typical stage | Early, exploratory research | Later, once you have a draft architecture |
| Main risk | Messier, harder-to-compare results | You may miss categories you never offered |
Takeaway: Run an open sort when you do not yet know how customers carve up the space, and a closed sort when you have a proposed navigation or menu and want to check whether it matches their thinking. Many teams run open first to generate categories, then closed to confirm them.
A concrete card sorting example
Imagine a founder building a personal-finance app who cannot decide how to organize the main menu. Rather than argue internally, they run an open card sort with a handful of target users.
Each card names one feature: "set a savings goal," "see my spending by category," "pay a bill," "check my credit score," "get investing tips," and so on. One by one, participants sort the cards into piles and name them. What emerges is telling — several users cluster "savings goal" and "spending by category" together under a pile they call "Am I okay this month?", while "investing tips" lands in a separate, lower-priority pile most people label "someday."
The founder learns two things they could not have designed on a whiteboard. First, users think in terms of a monthly question, not the neat feature buckets the team had drawn. Second, investing — which the team saw as a headline feature — is peripheral in the customer's mind. The navigation that follows can now speak the customer's language.
What card sorting reveals for validation
Card sorting reveals understanding and priority, not willingness to pay. Used well, it tells you how customers relate concepts, which words they trust, and what they treat as central versus incidental — the raw material for information architecture and messaging.
Because it produces qualitative insight rather than a demand signal, card sorting is best treated as one instrument in a larger toolkit. It sits alongside a structured validation experiment that tests whether people actually want the thing, and it pairs with a buy-a-feature exercise when you need to know which features matter enough to trade for. Where buy-a-feature forces prioritization under a budget, card sorting explains the mental structure behind those choices.
Fit it into your research plan deliberately. The complete guide to customer research for founders shows where a discovery method like this belongs relative to interviews, surveys, and behavioral tests. Tools like Edmired help you keep the findings organized, but do not overclaim: a clean sort means you understand how people think — a foundation for good design, not evidence that they will pay.
Key Takeaways
- Card sorting reveals mental models. Participants group and label cards to expose how they already organize a topic, in their own words.
- It is a qualitative discovery method. It surfaces understanding and priority, not demand or willingness to pay — never treat a sort as market proof.
- Open sorts discover; closed sorts confirm. Let participants invent categories to learn how they think, or supply fixed ones to test a structure you already drafted.
- Labels are as valuable as groupings. The names people give their piles hand you the exact language for navigation and messaging.
- It informs information architecture. Use the results to design menus, navigation, and content structure around real mental models instead of internal logic.
- It complements, not replaces, demand tests. Card sorting is one instrument; pair it with experiments that measure what customers actually do.
Frequently Asked Questions
What is the difference between open and closed card sorting?
In an open card sort, participants create and name their own categories, which is ideal for discovering how customers naturally think about a topic. In a closed card sort, you provide fixed categories and participants sort cards into them, which tests whether an existing structure matches their mental model. Open explores; closed validates.
How many participants do you need for a card sort?
Card sorting is qualitative, so you learn a great deal from a small group of the right target users rather than a large sample. A handful of well-chosen participants usually surfaces the dominant mental models and recurring groupings. Because you are looking for patterns in how people think, add participants until new sorts stop revealing anything new.
Does card sorting prove customers will buy my product?
No. Card sorting reveals how people organize and prioritize concepts — their understanding and mental model — not whether they will pay. It is a discovery method for shaping information architecture and messaging. To test real demand, pair it with a validation experiment or a purchase-based test that measures actual customer behavior, not just how they group ideas.
When should I run a card sort during product development?
Run a card sort early, while you are still deciding how to structure navigation, features, or content and are unsure how customers think about them. An open sort suits the earliest exploratory stage; a closed sort fits later, once you have a draft architecture to validate. It is most useful before you commit engineering effort to a structure.