learn.web Back to the curriculum ↗

Research / Structure / Validation / 03

Make information findable

Model content, vocabulary, navigation, and search around how people think—not your org chart.

Time
70 min
Mode
Learn → Make → Check
Path
UX & Product Design

By the end, you can…

Information architecture is a prediction

Every category and navigation label predicts where a person will look. Start by inventorying the content and identifying entities, tasks, relationships, and lifecycle. An org chart is rarely a useful model for people outside the organization.

Labels carry more weight than icons. Use familiar, specific language and avoid forcing one item into several ambiguous categories. Search and navigation complement each other; neither repairs unclear content.

Test the structure without visual design

Open card sorting reveals how participants group information. Closed sorting tests an existing structure. Tree testing asks where people would look for specific items without interface decoration.

Measure first-click confidence, success, directness, and the language participants use. A wrong but popular location may signal that the model—not the user—needs to change.

Your studio task

Make — Card-sort a messy collection of at least 25 items and propose a navigation model.

  1. Inventory at least 25 content items.
  2. Identify the top tasks and vocabulary used by your audience.
  3. Run a small card sort or tree test.
  4. Revise labels and structure based on failed paths and hesitation.
Definition of done

Labels use the audience’s vocabulary and every high-priority item has an obvious home.

Open the interactive lesson with its workspace ↗

Knowledge check

What should primarily determine a navigation label?
  1. Internal department names
  2. Words the intended audience expects for the task or content
  3. The shortest possible abbreviation
Reveal answer (B)

Labels work when they match the audience’s information scent and vocabulary.

A “wrong but popular” location in a tree test most likely means…
  1. The participant is wrong
  2. The information model, not the user, needs to change
  3. The test is broken
Reveal answer (B)

Repeated expected-looking placement is a signal about the model, and the model is the thing to revise.