Card Sorting — Examples
Example 1 — Reading agreement from an open sort
An intranet team has run an open sort with twelve people and fifteen cards, and now has a pile of group names. Turn the results into a structure and pick the one label worth testing before the menu is built.
Participants: 12, each grouped the 15 cards alone and named the groups
Repeated names: "Pay & benefits" (9 people), "Policies" (7),
"IT help" (6)
The odd card: "Buying a laptop" went with Pay for 5 people and
with IT help for 6 people
Solution
Pair agreement (share of the 12 who grouped the pair together)
Payslip + Expenses 10/12 83% -> one category
Holiday + Sick leave 11/12 92% -> one category
Policies + Contracts 9/12 75% -> one category
Laptop + IT support 6/12 50% -> the label is the problem
Label to test: "Equipment and requests", in place of either the
Pay or the IT help wording around that card
Starting from pairs above 70% gives a structure most people already hold, while the evenly split pair sends you to the wording rather than to a fourth category. Running the twelve sessions separately is what makes those percentages worth reading, because a group session converges on whoever argues longest. The usual mistake is to design the tidy map one participant drew, and the second is to ship the chosen label without a closed sort to check it.