If Your Users Are Lost, It's Not Them — It's Your Information Architecture

When a user can't find what they're looking for on your website, the instinct is to blame the user. "They should have checked under Resources." "It was clearly labeled." "We've had that page for years."

But here's the uncomfortable truth: if users are consistently getting lost, the problem is not them. It's your information architecture.

Information architecture (IA) is the structural design of information — how content is organized, labeled, and navigated. It's largely invisible when it works, and painfully obvious when it doesn't. And because the people who design it are intimately familiar with the underlying structure, they're usually the last to notice when it fails.

Why Good IA Is So Hard

The fundamental challenge of information architecture is something called the curse of knowledge. Once you know how your product is organized, you can't unknow it. The labels that make perfect sense to you — "Resources," "Solutions," "Platform" — are deeply ambiguous to people encountering them for the first time.

Consider how many websites have a "Resources" section. Click into almost any of them and you'll find a wildly inconsistent mix: whitepapers, blog posts, case studies, videos, tools, calculators, templates. What makes these things "resources" rather than, say, "content" or "downloads" or "tools"? The label communicates nothing about what's inside.

Your users aren't failing to navigate your site. They're navigating it exactly as rationally as the evidence allows — and the evidence you've given them is ambiguous.

The Components of Information Architecture

IA is typically broken down into four interconnected systems:

Organization systems — How information is categorized. Are you organizing by topic? By audience? By format? By task? The choice matters enormously, and using the wrong one for your context is a common source of confusion.

Labeling systems — The words you use to describe categories, navigation items, and content types. Labels that work in internal documents often fail in user interfaces because they reflect your mental model, not your users'.

Navigation systems — How users move through the structure. Global nav, local nav, breadcrumbs, filters, search — these are the mechanisms through which your IA is experienced.

Search systems — How users find things directly rather than browsing. Good IA makes browsing work well. Good search makes it possible to bypass browsing when browsing fails.

All four systems need to work together. You can have a perfectly organized structure with confusing labels. You can have clear labels with navigation that doesn't let users follow their instincts. A strong IA addresses all four.

How to Diagnose an IA Problem

Tree testing

Tree testing is the most direct way to evaluate information architecture. You present users with a text-only representation of your navigation structure — no visual design, no content — and ask them to find specific things.

When I ran tree testing for the Society of Thoracic Surgeons redesign, we discovered that three of their primary navigation labels were understood differently by surgeons than by the team that created them. One label that the team considered intuitive was selected correctly by only 34% of participants. That's worse than random guessing.

Tree testing surfaces these problems at the structural level, before you've invested in visual design. It's one of the most cost-effective research methods in UX.

Card sorting

Card sorting asks users to group content items into categories — and then name those categories. The categories users create, and the labels they choose, reveal how they actually think about your content.

Open card sorting (where users create their own categories) is best for discovering mental models from scratch. Closed card sorting (where categories are predefined) is better for validating an existing structure.

First-click testing

Research consistently shows that if a user's first click in a navigation task is wrong, they will complete the task successfully only 19% of the time. If their first click is correct, they succeed 87% of the time.

First-click testing identifies where users go first when they're looking for something specific. If the majority are clicking in the wrong place, your labels or organization are misleading them.

Analytics and search data

Before any formal testing, look at your search logs. What are users searching for? If users are searching for things that exist in your navigation, it means they couldn't find them by browsing. High search volume for navigable content is a diagnostic signal.

Also look at exit pages. Where are users leaving without completing a task? High exit rates on pages that should be waypoints — not destinations — suggest users got there, didn't find what they needed, and gave up.

Principles for Better IA

Organize for users, not for your org chart. One of the most common IA anti-patterns is a website organized the same way the company is organized: Marketing, Product, Engineering, Legal. Users don't care about your internal structure. They care about their goals.

Label with your users' vocabulary. Run a few user interviews. Listen for the words they use to describe the things they're looking for. Use those words, not your internal jargon.

Limit top-level categories. Research on cognitive load consistently shows that more than 7 (±2) items in a group starts to degrade findability. If your top navigation has 12 items, something is wrong. Either categories need to be combined, or you have a strategy problem, not an IA problem.

Design for wayfinding, not just finding. Users need to know not just where they are, but where they've been and where they can go. Breadcrumbs, active navigation states, and consistent page headers are all tools for supporting wayfinding. Don't treat them as decoration.

Test early and often. IA is one of the few UX disciplines where simple, low-cost testing methods (tree testing, card sorting) can dramatically improve outcomes. There's no excuse for launching a new navigation structure without having tested it with at least five real users.

The STS Case Study

When the Society of Thoracic Surgeons came to us, their website had 340 pages organized into a structure that had grown organically over a decade. Clinical guidelines were buried four levels deep. Content existed in six separate sections with no clear distinction between them.

We ran Jobs-to-be-Done interviews to understand what members actually came to the site to accomplish. Then we ran card sorting to understand how surgeons categorized the content. Then we built a new IA informed by that research, and validated it with tree testing involving 50 surgeon participants.

The result: task success rate went from 51% to 87%. Page views increased 58% year-over-year. On-site interactions increased 94%.

None of that came from better visual design. It came from organizing information in a way that matched how surgeons actually think.

Where to Start

If you suspect your IA is letting users down, here's the minimum viable starting point:

1. Look at your search logs. What are users searching for that already exists in your navigation?
2. Ask five users to find something on your site while you watch. Don't help them. Just observe.
3. Map your current IA structure. Write it down. Does it reflect your org chart, or your users' goals?

You don't need a research budget or a six-month project to start improving. You need observation, honesty about what you find, and the willingness to reorganize around your users rather than your assumptions.

Your users aren't lost. Your IA just hasn't found them yet.