Information architecture is the way content is grouped, labelled and connected so that people can find what they need and understand what they’ve found. Abby Covert puts it more broadly as the way the parts of something are arranged to make it understandable as a whole. It’s often mistaken for the navigation menu, but the menu is only the most visible part. A museum might organise the same collection by artist, period and material so visitors can explore by more than one route, and its floor plan, room names and object labels all belong to the same architecture. On a website, search, browsing and related links are different expressions of the structure, not substitutes for deciding how the information connects.
The term came from an architect. Richard Saul Wurman popularised it in his address to the American Institute of Architects conference in 1976, applying an architect’s concern with organised space to making complex information clear, in work closer to what’s now called information design. The web gave the term its working meaning: Louis Rosenfeld and Peter Morville, who came from library science, published Information Architecture for the World Wide Web in 1998 (the “polar bear book”, after its cover) and built the practice around organisation, labelling, navigation and search systems. Covert’s How to Make Sense of Any Mess (2014) took the idea back beyond websites to anyone arranging information for other people.
Why it matters for content
Content that can’t be found is, from the reader’s point of view, content that doesn’t exist. Information architecture is the difference between an estate people can navigate confidently and one they endure, and it decides whether related content behaves like a coherent body of knowledge or a set of unconnected pages that happen to share a logo. It also sets context before anyone reads a word of body text. As Jorge Arango points out, a navigation bar listing current accounts, savings, credit cards and mortgages tells you you’re on a bank’s website without saying so.
Good architecture uses the language of the people using it rather than the organisation’s internal vocabulary. Internal labels make readers translate the organisation’s structure before they can find an answer. Sometimes that structure is the subject, as when someone wants a particular department’s contact details, so the test is the need in front of you rather than a blanket rule against the organisation chart. During a redesign or migration, test the proposed grouping before moving everything into it. An organisational restructure is a reason to review the architecture; it isn’t automatically a reason to reproduce the new hierarchy on the site.
How AI changes the picture
AI hasn’t abolished the need for structure, despite hopeful claims that search and assistants make navigation irrelevant. It has changed where structure pays off. Clear hierarchies, consistent labels, sensible taxonomies and canonical pages all help AI systems retrieve the right material and represent it accurately, though a tidy hierarchy doesn’t guarantee a correct generated answer. For readers, search changes most, moving from lists of pages towards synthesised answers. Navigation changes less. A chat box assumes people know what to ask for and what things are called, which is often what they don’t know when they arrive, and a menu that rebuilt itself for every visitor would lose the shared context a stable one provides.
AI also makes the grouping work faster. It can cluster large content sets, propose candidate categories, spot orphan pages, suggest alternative labels and check whether existing labels match the words people search for. On large estates that’s a real gain, but the output is a set of hypotheses for testing rather than a finished structure.
AI can help compare the consequences of proposed structures, but it doesn’t carry the authority to settle organisational disputes. Information architecture decisions are frequently political: whose content appears in the top navigation, which team’s language wins, what gets demoted to a subsection. Dan Brown notes that every structural choice sets a precedent, which is why one team’s place in a “quick links” panel soon becomes every team’s request. Those decisions need people who can carry the consequences, ideally with evidence from card sorting or testing rather than seniority.
Questions people ask
- How do you test an information architecture?
Card sorting and tree testing are the standard methods, alongside watching people attempt real tasks. Card sorting explores how people group things and is most useful early, while tree testing checks whether they can find things in a proposed hierarchy before any design exists. Neither is automatically cheap, and both depend on well-designed tasks and a sensible participant sample.
- How is information architecture different from a taxonomy?
A taxonomy is one component of it: the controlled set of terms used to classify content. Information architecture also covers navigation, page relationships, labels, URLs and search, and the decisions about how all of these fit together.
- Do good search and AI assistants remove the need for structure?
They depend on it. Search works better on well-structured, well-labelled content, and assistants retrieve more reliably from it. Many people still browse to orient themselves before searching, particularly when they don’t yet know the right words.
- Should a website mirror the organisation chart?
Only where the organisation is what people are looking for, such as identifying who runs a service. For finding help by circumstance or topic, a different grouping usually works better. Test the routes instead of assuming one model fits every task.
Learn more
- A Brief History of Information Architecture – Andrea Resmini and Luca Rosati trace the term from 1960s IBM papers and Wurman to Rosenfeld and Morville, separating the information design and library science traditions behind it.
- How to Make Sense of Any Mess – the free online edition of Covert’s book, which treats information architecture as a skill for anyone and works through intent, structure and taxonomy in short lessons.
- Everyday Information Architecture – the site for Lisa Maria Marquis’s practical book (first published under the name Lisa Maria Martin), with its introduction and links to buy it in print or as an ebook. The book covers sitemaps, wayfinding and taxonomies for people who don’t call themselves information architects, with a strong thread on the ethics of how information is organised.
- Information Architecture Lenses – Dan Brown on questioning structural decisions from different angles, including the precedent each one sets and how navigation teaches people what a product does.
- The Impact of AI on Front-end Information Architecture – Jorge Arango on how AI is likely to change search, navigation and labelling, and why he’s wary of menus that rebuild themselves for each user.