← Back to Case Studies
UX ResearchCard SortingInformation Architecture2026

Card Sorting —
structure before screens.

A deep-dive into card sorting as a foundational UX research method. How open and closed sorting sessions surface mental models, inform navigation taxonomy, and validate information architecture before a single wireframe is drawn.

Card Sorting case study cover
The Challenge

When the menu is the product.

Most digital products live or die by their navigation. Yet teams spend hours debating button labels and tab orders without ever asking the people who'll actually use them. The result is a structure that makes sense to the team — and confuses everyone else.

The brief on this engagement was simple: take a sprawling enterprise platform with seven years of feature accretion and find an information architecture that real users could move through without a manual.

Approach

Three sorts. Sixteen voices. One mental model.

Card sorting works because it puts users in the driver's seat — they group, name, and prioritise content the way it makes sense to them. I ran three layered sessions to triangulate findings and avoid biases that any single technique introduces.

  • Open sort. Participants grouped 60+ feature cards into clusters of their choosing, with no pre-existing categories. Surfaced raw mental models.
  • Closed sort. Same cards, but participants had to fit them into seven proposed top-level categories. Tested whether the team's draft IA matched user intuition.
  • Hybrid sort. A reverse pass — start from existing labels, then let users propose new groupings or rename what didn't land.

All sessions were moderated, recorded, and analysed for both the obvious patterns (which cards always cluster together) and the quiet ones (which labels are repeatedly hesitated over).

What surfaced

The team's labels were exporter language.

The biggest finding wasn't where the boxes went — it was how users renamed them. Internal terms like Configurations and Asset Library were quietly translated by users into Settings and My Files in nearly every session. The IA wasn't broken; the language was.

"Structure before screens. Card sorting is the fastest path to an IA that users actually understand — and the cheapest one to fix when you get it wrong."

The matrix analysis (which cards co-occur in user-formed groups) revealed three cluster pairs the team had never considered grouping. Two of those became new top-level sections in the redesigned navigation.

16+
Participants across two regions and three personas
3
Sort variations — open, closed, hybrid
60+
Feature cards tested per session
7→5
Top-level categories simplified after analysis
Outcome

Wireframes nobody had to defend.

The new IA shipped without the usual stakeholder pushback because every grouping was traceable to user evidence. Wireframe reviews shifted away from "what should this be called" debates to "does this flow work" — exactly where they should have been all along.

Card sorting didn't just produce a better menu. It produced shared language between research, design, and product — a small artifact that punches well above its weight.

From the gallery

Selected screens.

View full case study on Behance ↗

Like the way I think?
Let's talk about your product.

Get in touch →