Your menu was probably organized by whoever built the site, around how the company is structured. Card sorting asks the customers how they would organize it instead.
What is card sorting?
Card sorting is a research method where people sort a set of labeled cards (one per product, service, or page) into groups that make sense to them. It used to be index cards on a table. Today it is usually done online, with tools built for it, so customers can do it from home in fifteen or twenty minutes. You end up with a picture of how your customers think your content fits together, and the words they use for each group.
What is the difference between an open and a closed card sort?
An open card sort lets people create and name their own groups, and a closed card sort gives them your group names and asks them to place each card.
- Open sort. Use it when you are starting over or do not know what the categories should be. You learn the customers' vocabulary and the groupings they expect.
- Closed sort. Use it when you already have categories, or a proposed set, and want to see whether people put things where you would. Cards that land in three different groups are the problem cards.
- Hybrid sort. Starts with your groups but lets people add their own. Useful when you think your categories are mostly right.
Most projects run an open sort first, draft a menu from it, then check the draft.
What is tree testing, and how does it fit?
Tree testing is the follow-up: people get a task and click through a text-only version of your proposed menu to find the answer. No pictures, no design, only the labels and the structure. If people go the wrong way on a task, the label or the grouping is wrong, and you fix it before a designer or developer spends a dollar. Card sorting builds the menu. Tree testing proves it works.
When should a business use card sorting?
Use it any time you are about to decide how things are grouped and named.
- A site redesign. Before anyone draws a page, settle the menu.
- A growing product catalog. When categories have multiplied and shoppers cannot find things.
- Merging websites or business units. When two companies or departments each bring their own names for the same thing.
- An intranet or help center. Employees and customers both get lost in folders named after the org chart.
- Renaming services. When sales calls keep starting with "what does that mean?"
Skip it when you have a handful of pages. A five-page site does not need a study.
What do I get at the end?
You get a recommended structure, the labels to use, and evidence for both. A good card sorting report shows which items people agreed on, which ones split them, what they named the groups, and a draft menu. Pair it with tree testing results and you have a structure you can hand to a designer with confidence that customers will find their way.
Has card sorting worked for a real company?
Yes. For HGTV Home by Sherwin-Williams I ran taxonomy research with real users: a baseline study, two rounds of card sorting that each tested a named theory, and a tree test to validate the final structure. We cut the categories from twelve to seven, and shopper confidence went up 13%. The full write-up is in the HGTV Home by Sherwin-Williams case study.
What does card sorting cost?
The online tools are inexpensive; check the vendor's current plan page. The cost that matters is the time to pick the cards, recruit the right people, and read the results without defending the current menu. A one-hour second opinion for $150 tells you whether you need a sort, a tree test, or neither. A full card sort and tree test is quoted as one fixed price on UX research and leadership.
Answered by Jared Lukes, Minneapolis. Twenty-five years in brand, UX, and technology, currently holding fractional AI and technology seats at three companies. About Jared.