Hierarchical Card Sorting
Hierarchical Card Sorting (HCS) is a sorting interview that shows how a person or small group categorises and values a set of real cases, such as projects, partners or locations, by splitting the cards again and again into two piles according to the most significant difference between them. Rick Davies developed it in 1993 during PhD fieldwork in Bangladesh, drawing on Gregory Bateson's idea of information as a difference that makes a difference. After every split the respondent explains what that difference leads to, and the result is a tree of nested distinctions that can then be used to compare and rank the cases.
Facilitation script
- 1
Explain the purpose, agree the area of knowledge, and list the cases with the respondent. Write one per numbered card.
10 min - 2
Warm-up with all cards in one pile: ask what kinds of differences exist among the cases, without sorting yet.
5 min - 3
First split into two piles by the most significant difference. Ask what difference it makes and record the cards, the distinction and the consequence.
10 min - 4
Repeat the split, the question and the recording on each pile in turn until single cards remain or no meaningful difference is left.
30 min - 5
Draw the tree and read it back to the respondent. Correct labels that do not match what they meant.
10 min - 6
Work through the branches with an evaluative question, such as which side was more successful and why, and write down the resulting ranking.
15 min - 7
Review the result together: what surprised the respondent, which distinctions carry most weight, and how the findings will be used.
10 min
Tips
Do not suggest categories; the method only works if the distinctions are the respondent's own, including ones you find odd.
Write down the exact words used for each difference, because the wording is the finding.
Expect the first split to be easy and the later ones to slow down; that is where the less rehearsed criteria appear, so give them time.
Common pitfalls
Offering example categories to speed things up, so the tree reflects the interviewer's view of the portfolio
Accepting a split without asking what difference it makes, which leaves descriptive labels such as 'urban and rural' with no insight into why they matter
Starting with too many cards, so the respondent tires and the later splits become arbitrary
Recording only the piles and not the wording of each distinction, which makes the tree impossible to interpret afterwards
Presenting one person's tree as the organisation's view, although it shows only how that respondent sees the cases
Variations
The method is normally used with one respondent but adapts easily to a small group, which must then agree each split aloud; record the disagreements, since they show competing criteria. Running the same card set separately with different stakeholders, such as field staff, managers and funders, and comparing the trees shows how differently they see the same portfolio. Remote: use a shared spreadsheet or online board with one note per case and build the tree column by column. With about 25 cases or more the process becomes slow and tiring, so sample cases or sort in two sessions.
Where it fits
When to use it
An organisation runs many projects or partnerships and has no shared language for how they differ
Experienced staff judge cases well but cannot say what criteria they apply
An evaluation needs a ranking of cases together with the reasoning behind it
You suspect that head office and field staff see the same portfolio in different ways and want to make that visible
You need to pick a few contrasting cases for case studies and want the choice grounded in how insiders see them
When not to use it
The items are ideas or content to be grouped into themes, not real cases the respondent knows; use an Affinity Map or a standard Card Sort
Criteria are already agreed and you only need scores; a Decision Matrix is quicker
You have far more than 25 cases; the repeated splitting takes too long and tires people, so sample the cases first
The respondent knows only a few of the cases at first hand; the splits would rest on guesswork
A large group must take part at once; the method needs sustained attention on one tree, so use Dot Voting or Forced Ranking for a quick group prioritisation
Related methods
Frequently asked questions
What is Hierarchical Card Sorting?▾
Hierarchical Card Sorting is a participatory method developed by Rick Davies in 1993. A respondent repeatedly splits a set of cards, each naming a real case, into two piles by the most significant difference and explains what that difference leads to. The outcome is a tree of nested distinctions that shows how the person categorises and values the cases.
How is it different from an ordinary card sort?▾
A standard card sort lets people form any number of groups at once, usually to design a navigation or content structure. Hierarchical Card Sorting forces successive two-way splits and asks about the consequence of each one, which produces a hierarchy of criteria and, with an evaluative question, a ranking. It is a tool for evaluation and sense-making more than for information architecture.
How many cards should I use?▾
Keep it modest, especially the first time. Rick Davies warns that sets of 25 or more cases make the process time-consuming and risk boring both respondent and interviewer.
Can it be done with a group?▾
Yes. It is normally used with one respondent, but the same procedure works with a small group that agrees each split through discussion. Keep the group to about six so that everyone can see and handle the cards; that limit is our practical advice, not a figure from the source.
How long does a session take?▾
The sources give no duration. As an estimate, allow one to two hours for about 20 cases including the evaluative pass, and less for a smaller set.
Plan your next workshop with AI
Workshop Weaver helps you combine methods like Hierarchical Card Sorting into a complete, timed agenda in minutes.
Try it freeMethod descriptions on Workshop Weaver are original content written by our team, based on established facilitation practices. This method was inspired by work from Rick Davies.