Good navigation helps people predict where an answer will be, recognize it when they arrive, and find it again. Start with a handful of real visitor tasks, then test whether your page groups and labels support them. You can do this with an existing site, a simple prototype, or a written outline.
Start with what visitors came to do
List five important tasks using customer questions, support requests, site-search logs, or conversations with actual users. Write the task as an outcome: “Find the replacement filter for a discontinued purifier.” Record the expected answer and the page that should provide it. An inventory of departments is useful internally; a navigation plan needs the visitor's vocabulary.
Separate a missing answer from an undiscoverable answer. If the replacement part is absent from the site, changing the menu cannot complete the task. If the information exists in three inconsistent places, appoint an owner and choose a primary page before testing its location.
Choose participants who resemble likely visitors, including people with different experience and access needs. The GOV.UK usability-testing guidance recommends believable tasks with clear goals and neutral wording. Obtain consent for research and any recording; use safe sample details when a prototype cannot protect personal information.
Group related answers and use specific labels
Write one content item on each card, then ask a few likely users to group the cards and name the groups. Listen for ambiguous items and competing interpretations. Use this exercise to propose a structure, then test whether people can find answers within it. Grouping preferences alone do not establish that navigation works.
Keep labels consistent from menu to destination heading. Distinguish “Delivery times” from “Track an order” when they answer different questions. Use descriptive links in the article itself where visitors need a prerequisite or next step.
W3C's menus tutorial explains meaningful menu structure, orientation, keyboard operation, and focus. A menu must remain usable beyond mouse hover. Check opening, moving through, and closing it with a keyboard, as well as its behavior with touch and screen magnification.
Observe a short, repeatable session
- Prepare the starting point. Use the same homepage or landing page for a given task. Verify that the prototype contains the expected answer.
- Read the task neutrally. Explain that the site is being evaluated. Ask the participant to describe what they expect as they proceed.
- Watch before helping. Record the first choice, wrong turns, uncertainty, and whether the answer is correct. Mark any help you give.
- Check understanding. Ask what they would do next. Reaching the right page while misunderstanding the answer is still a problem.
- Repeat after a specific revision. Keep the task and success definition consistent. Fresh participants reduce the effect of learning the old route.
Download the navigation test sheet to record the task, starting page, expected answer, observed route, assistance, and proposed repair. A small round reveals problems to investigate; report counts and examples without presenting them as a population-wide success rate.
Fix the obstacle that prevents the task
| Observation | Likely investigation | Next check |
|---|---|---|
| Repeated wrong first choice | Group names or overlapping categories | Test revised labels with the same task |
| Right page, wrong answer | Heading, explanation, model or date ambiguity | Ask the reader to explain the answer |
| Keyboard user cannot open menu | Interaction and focus behavior | Complete the route without a pointer |
| Answer found only through search | Missing contextual or category links | Test both browsing and search routes |
Repair blockers first, then frequent detours and confusing wording. Assign an owner and a verification task to each change. Review again when products, policies, audiences, or the menu structure change. For on-site query evaluation, continue with product and enterprise search testing.