By the end of this lesson you can
- Choose methods to validate an information architecture before and after launch.
- Set success measures for findability.
- Iterate on IA using evidence from tests and analytics.
The idea
An IA is a hypothesis about how people will find things. Validation checks that hypothesis before you invest in building it, and keeps checking after launch.
Before launch
- Tree testing: checks findability using the structure alone.
- First-click testing: shows a screen design and asks where people would click first to complete a task. A correct first click strongly predicts task success.
- Usability testing with prototypes: checks the IA in context with navigation and content.
After launch
- Analytics: paths through the site, pages where people go back and forth ("pogo-sticking"), and exit pages.
- Site search logs: frequent searches for things that are in the navigation suggest the labels or placement are not working.
- Support data: questions about where to find something.
Measures
Common measures are task success rate, directness (went straight there without backtracking) and time to find. Set targets before testing, such as 80% success on key tasks, so you know when the IA is good enough.
Why it matters in real projects
Restructuring after launch is expensive and disruptive for existing users. Validation catches problems while they are cheap to fix and gives the team confidence to proceed. Continued monitoring after launch shows when content growth has started to break the structure, which happens to every product over time.
A realistic example
A bank redesigns its help centre IA. A tree test with 60 participants and ten tasks shows 85% average success, but only 40% for "report a lost card", because people look under "Security" rather than "Cards".
The team adds a cross-link and moves the item. A first-click test on the new design confirms the improvement. Three months after launch, search logs show "lost card" is still a top search, so the team adds an urgent "Lost or stolen card" link on the help centre home page. Searches for it drop.
Step by step
Choose key tasks
Pick the 8 to 10 most important or frequent tasks.
Set targets
Decide what success rate and directness count as good enough.
Run a tree test
Test the structure without design.
Run a first-click test
Test designed screens to see whether people start in the right place.
Fix and retest
Change failing areas and test again.
Monitor after launch
Review analytics, search logs and support data regularly.
Common mistakes
Mistake Skipping validation because the structure "feels obvious".
Instead Test it. What feels obvious to the team often is not to users.
Mistake Only testing once.
Instead Iterate and retest, and keep monitoring after launch.
Mistake No targets.
Instead Set success thresholds before you test.
Check your understanding
Choose an answer, then check it. There is an explanation either way.
Practice exercise
Exercise · about 30 min
IA validation plan
Write a validation plan for an IA you created earlier in this stage.
Steps
- List eight key tasks.
- Set success and directness targets.
- Choose pre-launch methods and explain why.
- List three post-launch signals you would monitor.
Deliverable
A one-page validation plan.
Self-assessment
Tick each item when it is true of your exercise. Be honest; this is for you.
4 items
Summary
- An IA is a hypothesis that must be validated.
- Use tree tests, first-click tests and usability tests before launch.
- Use analytics, search logs and support data after launch.
- Set targets and iterate.
Further reading
Third-party sources, chosen for reliability. They open in a new tab.
- The importance of the first click (opens in a new tab)MeasuringU
- Tree testing part 2: interpreting the results (opens in a new tab)Nielsen Norman Group
Your notes
Marking complete updates your course and roadmap progress.