Open, closed and hybrid card sorts explained

Three variants, one method. The difference is who supplies the categories, and that single choice decides what you can learn.

Open sort: discover the structure

Participants create their own groups and name them.

You learn: what structure exists in customers' heads, and what they call things.

You give up: easy comparison. Everybody invents their own labels, so twenty people produce sixty group names for what might be four ideas. That is the cost of the method, not a flaw in it.

Use it when you have a long list and no structure yet, or when you suspect the structure your team built reflects how the company is organised rather than how customers think. Which it usually does.

The naming is worth as much as the grouping. When eight people independently call a pile "getting it back to you" and your website says "Returns and Exchanges Policy", you have learned something you cannot get from a closed sort.

Closed sort: validate a structure

You supply the categories. Participants file items into them.

You learn: whether your categories can hold your content, and which specific items keep landing in the wrong place.

You give up: the chance to discover a category you had not thought of. People will find a home for everything, because you have not given them the option not to.

Use it when the structure is agreed and you want evidence before building, or when you need to compare two candidate structures directly.

Hybrid sort: your categories, plus theirs

You supply categories and participants can add their own.

You learn: both, at a lower resolution than either. Where your categories work, people use them. Where they do not, people build something else, and the group they invent tells you what was missing.

Use it when you are fairly confident but suspect a gap, or when the list has grown and you want to know whether it still fits the shelves you have.

Which to pick

Your situation	Sort type
Long messy list, no structure	Open
Draft structure, want customer language for it	Open
Structure agreed, want evidence before building	Closed
Two candidate structures to compare	Closed, one each
Confident but suspect a gap	Hybrid
Content has grown, shelves may not fit	Hybrid

The standardisation problem, and how it is solved

Open sorts have one genuine practical difficulty. If fourteen people create groups called "Comfort", "comfort", "Fit and comfort" and "Comfy", your analysis sees four groups where there is one idea.

SortedResearch handles this with a standardisation grid: every raw name people typed, how often each was used, and a field to map them onto a single standardised name. Do that once and it applies everywhere else, to the clustering, to the popular placements, and to the report. It takes a few minutes and it is the difference between an open sort you can read and one you cannot.

A note on deck size

All three variants get harder as the deck grows, but not equally. A closed sort with 60 cards is tedious but manageable, because the categories stay on screen. An open sort with 60 cards asks somebody to hold an emerging structure in their head while the pile shrinks, which is real cognitive work.

Sixty cards is a sensible ceiling for an open sort, and it is where a lot of tools start to break down visually. If your list is longer than that, cut it first: remove near-duplicates, drop anything nobody would recognise, and prefer the customer's phrasing to the internal one.