Information Architecture & Navigation Design

UI/UX Design Principles

Chapter 3 · Information Architecture & Navigation Design

Chapter 1 drew the line between UX and UI. Information architecture (IA) sits closer to the UX end of that line — it's the underlying structure and organization of a product's content, before a single button or color gets designed. Navigation is what a person actually uses to move through that structure. Get the architecture wrong, and no amount of good visual design in Chapter 5 will make the product easy to use.

A Real, Named Origin

Designer Richard Saul Wurman is credited with coining the term "information architecture" itself. His own book, Information Architects, helped establish it as a named discipline in the 1990s — a real, deliberate framing of organizing information as its own design problem, distinct from visual design or engineering.

Wurman's Real Organizing Principles

Several of Wurman's own real, named principles are directly useful for structuring a real product today:

Objects

Treat pieces of content as distinct, structured entities — not undifferentiated text on a page.

Choices

Offer meaningful navigation options — too few choices limits users, too many overwhelms them.

Disclosure

Reveal information progressively — enough to inform the next choice, not the entire structure at once.

Front Doors

Design multiple real entry points — most visitors won't arrive at your intended starting page.

Growth

Structure the system so it can genuinely expand later without a full redesign.

Navigation Patterns

Pattern	Structure	Best for
Hierarchical	A single tree — broad categories branching into narrower subcategories	Most content-heavy sites; matches how people already expect to browse
Hub-and-spoke	One central screen, each destination reached and returned to from there	Task-focused apps with genuinely separate, unrelated features
Sequential	A fixed, ordered path — step 1, then step 2, then step 3	Onboarding flows, checkout, anything with a real required order

The Real "Magical Number Seven" Correction

It's extremely common advice to keep a navigation menu to "7 items, plus or minus 2" — citing psychologist George Miller's famous 1956 paper. The real paper doesn't actually support that specific application.

What Miller Actually Found — and Said Himself

Miller's 1956 paper covered two genuinely different, only coincidentally similar-looking findings: how many distinct levels people can reliably tell apart on a single scale (roughly 4-8), and how many items people can hold in short-term memory right after seeing them (roughly 7). Miller himself stated plainly that the correspondence between these two numbers was "only a coincidence," and that there is "nothing magical" about the number seven — he used the phrase rhetorically, not as a hard cognitive limit.

Later research complicates the number further: memory span genuinely varies by what's being remembered (roughly 7 digits, but closer to 5 words), and psychologist Nelson Cowan has proposed working memory capacity is closer to 4 chunks, not 7. None of this research was ever about how many items a person can comfortably scan in a navigation menu — a completely different cognitive task from recalling a list from memory.

What This Actually Means for Navigation Design

A navigation menu isn't a memory-recall task — a user reads the menu, they don't memorize it first. The real design guidance isn't "cap it at 7 items" at all; it's to group related items sensibly, label them clearly, and let real usability testing (Chapter 2) — not a 1956 memory-span number — tell you whether a specific menu is actually too long.

The Three-Click Rule: A Real, Debunked Myth

A second widely-repeated rule — that users should be able to reach any content within three clicks, or they'll leave — was popularized by Jeffrey Zeldman's own 2001 book. Real usability research has since challenged it directly: studies have found the number of clicks needed affects neither user satisfaction nor task success rate. The Nielsen Norman Group has publicly labeled the rule false.

What Actually Matters Instead

What genuinely predicts whether a user finds what they need isn't the click count — it's whether each individual click is a confident, correctly-informed one. Five clicks where each choice is obvious beats two clicks where either one is a genuine guess. Rigidly enforcing "three clicks max" can actively hurt a real site — an e-commerce site forced into overly broad top-level categories to satisfy the rule makes browsing genuinely harder, not easier.

Card Sorting: Testing an IA With Real Users

Card sorting is a real, direct IA research method building on Chapter 2's own toolkit: real users are given cards labeled with real content topics and asked to group them in a way that makes sense to them (open card sort), or to sort them into categories you've already defined (closed card sort). The result is real evidence for how your actual users mentally organize the content — not a guess made in a planning meeting.

Hands-On Exercises

Exercise 1

A teammate insists a navigation menu must be trimmed to exactly 7 items because "that's the limit of human memory, per Miller's research." Explain, in your own words, what's wrong with this reasoning, citing what Miller's own paper actually found.

📄 View solution

Exercise 2

An e-commerce team forces every product into one of only three top-level categories, specifically to satisfy the three-click rule, even though several products genuinely don't fit any of the three well. Explain, in your own words, what real research says about this tradeoff, and what should guide the decision instead.

📄 View solution

Exercise 3

A team wants to figure out how their real users would naturally group 40 different content topics into categories, before designing any navigation menu. Explain, in your own words, which method from this chapter fits this need, and why it's more reliable than the team guessing themselves.

📄 View solution

Chapter 3 Quick Reference

Information architecture (term coined by Richard Saul Wurman) — the underlying structure of content, distinct from navigation and visual design
Navigation patterns: hierarchical (most content), hub-and-spoke (separate features), sequential (a required order)
Real correction: Miller's 1956 "7±2" finding was about memory recall, not menu scanning — Miller himself called the number "not magical," and it doesn't actually justify a hard 7-item menu cap
Real correction: the three-click rule is debunked — click count doesn't predict satisfaction or success; what matters is whether each click is a confident, well-informed one
Card sorting — a real IA research method for learning how actual users mentally group content