Information Architecture

011 min

Information Architecture at a glance

• What it is: How a product's content is grouped, labeled, navigated, and searched.
• Origin: Wurman named it in 1976; Rosenfeld and Morville wrote the web playbook in 1998.
• Apply it by: Testing category names and labels with real users before building screens.
• It backfires when: A structure ships before anyone tests whether users group things the same way.

021 min

The problem Information Architecture solves

A product with enough content eventually reaches a point where no single person remembers where everything lives. A support team keeps three separate FAQ pages because nobody noticed the other two existed. A shopper gives up on a category page and types the product name straight into search instead, because browsing the page never surfaced it — a discoverability failure, not a visual one.

None of this is caused by a wrong button or a wrong color. It is caused by content added over time with no shared rule for where a new page belongs, what to call it, or how someone would look for it later. Information architecture is the discipline that writes that rule down, on purpose, before the content grows past the point where anyone can fix it by rearranging a menu.

032 min

The four systems inside Information Architecture

Rosenfeld and Morville's book breaks information architecture into four working systems. A page can look finished and still fail if any one of the four was skipped.

Organization systems decide how content is grouped and in what order. A grocery site can group products by aisle, by task ("what I need for breakfast"), or alphabetically, and none of these is correct in general. The right scheme depends on whether the person arriving already knows what they want or is still figuring it out.

Labeling systems decide what each group is called. A label is the promise a link makes about what sits behind it: "Solutions" and "Resources" promise almost nothing, while "Pricing" and "API Docs" promise something specific. The same label has to work for a person who already uses the field's own vocabulary and a person who doesn't.

Navigation systems decide how a person moves between groups once they are already inside the structure: a top-level menu, links inside one section, breadcrumbs, related-item links. Navigation is the visible interface built on top of the organization scheme. It is a separate decision from the scheme itself, covered in full below.

Search systems decide how a person finds something directly, without browsing the structure at all: what gets indexed, how a misspelling or a synonym is handled, how results get ranked. A page with strong search and a weak organization scheme still fails anyone who wants to browse. A page with a strong scheme and weak search still fails anyone who already knows the exact word they want, an information scent problem search alone cannot fix.

Information architecture, in Wikipedia's summary, comprises the combination of organization, labeling, search and navigation systems within websites and intranets. Jorge Arango, co-author of the book's fourth edition, describes the same four systems from the user's side: people experience IA in the navigation, search, and labeling systems of websites and apps.

041 min

Where the name Information Architecture comes from

Richard Saul Wurman, an architect by training, introduced the term at the American Institute of Architecture conference in 1976, where he served as chair. Wurman argued that organizing information was a design discipline in its own right, on the same footing as organizing physical space, and developed the idea further in his 1996 book Information Architects.

Wurman's version stayed a general idea about clarity. Turning it into a method for building websites took Louis Rosenfeld and Peter Morville. Their book, Information Architecture for the World Wide Web, was first published in February 1998. Practitioners nicknamed it "the polar bear book," after the animal on its O'Reilly cover, and it went on to sell over 250,000 copies.

Peter Morville and Louis Rosenfeld were the book's co-authors from 1998 to 2015. The fourth edition, retitled Information Architecture: For the Web and Beyond and co-written with Jorge Arango, came out through the same publisher in 2015.

052 min

Information Architecture vs. navigation and sitemaps

Three words get used as if they mean the same thing, and they don't.

Nielsen Norman Group states the distinction directly: IA is the information backbone of the site, and navigation refers to those elements in the UI that allow users to reach specific information on the site. Navigation is shaped by the architecture underneath it, but information architecture reaches past navigation to cover search behavior and the content itself too.

A sitemap has an even narrower job. Nielsen Norman Group is direct that a sitemap is only one part of the site's information architecture. The same source puts the order plainly: the IA comes first, the sitemap follows — a sitemap is a diagram used to communicate a structure that was already decided, and it usually leaves out the labeling and search decisions the underlying architecture also has to make.

062 min

How teams actually build an Information Architecture

Getting a structure to match how people actually think is not something one meeting can settle. Three steps do most of the real work, usually in this order.

A content inventory comes first. Before anything gets organized, someone lists every page, document, or product that already exists, including the duplicate FAQ pages nobody remembered, so the team is organizing something real instead of guessing at what's there.

Card sorting comes next, while the structure is still undecided. Participants are given individually labeled cards and sort them into groups. Nielsen Norman Group defines it directly: card sorting is a research method in which study participants place individually labeled cards into groups according to criteria that make the most sense to them. In an open sort, no categories are given in advance, and the point is to see how people group things on their own. In a closed sort, the categories already exist, and the point is to check whether that grouping still holds up. Doing this early surfaces people's own mental models of the product, rather than the team's. Nielsen Norman Group notes that card sorting is used to discover users' mental models for how content should be grouped.

Tree testing comes after a structure exists, to check whether it actually works. Participants see a text-only version of the proposed hierarchy and are asked to find where a specific item would live. Nielsen Norman Group defines a tree test as an evaluation of a hierarchical category structure, or tree, by having users find the locations in the tree where specific resources or features can be found. Nielsen Norman Group is specific about what a tree test skips: to conduct one, you don't need to create any prototypes, design layouts, or visuals, nor do you have to write any content. Because the test strips out visual design entirely, a wrong answer points at the structure and the labels, not at a button someone missed.

The two methods answer different questions, in a fixed order. Nielsen Norman Group sums it up in one line: card sorting is used to generate ideas for an IA while tree testing is used to evaluate IA options. Running a tree test on a structure that was never card-sorted skips the step where the structure came from real behavior in the first place.

081 min

How to tell when an Information Architecture is broken

A structure is failing when the signs show up in how people actually use the product, not in how it looks.

• The same piece of content gets created more than once, in different places, because the people adding it didn't know the other copy existed.
• Search becomes the default way to find anything, including things that should be one click away, because browsing the structure doesn't work.
• Support tickets repeat the same question about where something is, even after the interface around it gets redesigned.
• A new page gets added to whichever section is easiest to reach in the content system, rather than the section a person would actually look in.
• Two people on the same team, asked to find the same item, take two different paths through the menu.

None of these get fixed by changing a color, a font, or a button's position. They get fixed by changing what a section is called, what it contains, or where it sits, which is what an information architecture review does and a visual redesign does not.

092 min

Frequently asked questions about Information Architecture

?8 questions

Questions people ask

§12 sources

Sources