Design Website Navigation, Categories, URLs, and Breadcrumbs

Site structure determines whether readers get lost

Information architecture is not a pretty tree diagram. It is the consistency of names, locations, and routes. If the navigation says “Resources,” the page heading says “Knowledge Base,” and the URL starts with /solutions/, readers must guess again at every step. A search engine being able to fetch the pages does not mean their relationship is clear.

After choosing a platform, return to the content itself. Name public sections with words your readers would use, then map those names to internal fields. Your data model can be complex; public navigation should not be.

Put frequent, stable destinations in the main navigation

The header is not a complete directory. A content-led site often needs only four to six top-level entries: its main content areas, any genuinely useful tools or products, and perhaps an about or contact route. Privacy policies, login, and tag archives can live in the footer or appropriate context instead of crowding the top.

Test labels against tasks. A first-time visitor can predict what “Website Guides” contains; “Empowerment Center” gives little clue. Brand language can shape the visual identity without making the menu hard to understand. On mobile, check that the important destinations remain easy to reach without excessive scrolling.

Use categories for lasting themes and tags for details

An article category should be stable enough to last and broad enough to contain a meaningful sequence. Do not create a top-level category for every new search phrase. Making Caddy, Docker, and systemd separate primary categories can quickly leave you with several thin sections; they may work better as tags or topics within a deployment section.

Give each article one primary category so the same body is not published under several category URLs. Use internal links and tags when an article is relevant from multiple entry points. A category page should explain its scope and highlight a useful route, not merely list titles by date.

Keep URLs stable, readable, and restrained

A URL uniquely identifies a page and gives people a little context. Lowercase words separated by hyphens work well, for example /article/design-website-information-architecture. Dates, authors, and temporary category names are poor candidates for permanent URLs because they may change.

Choose one preferred URL for each piece of content. Decide how you handle filtered URLs with parameters, print versions, and case differences. Do not change an established URL every time you revise a keyword. An address that people have linked to and search engines have indexed is often more valuable than a theoretically neater replacement.

Breadcrumbs show location; body links show relationships

Breadcrumbs express hierarchy: Home / Website Guides / This Article. They help readers move up and can give search engines a structural signal. They cannot replace links to the previous or next task in a learning path, because task order is not always the same as category hierarchy.

Use the same labels as the actual sections, make parent levels clickable, and avoid linking the current page to itself. Structured data can help machines interpret the hierarchy, but visible navigation is the foundation. A JSON-LD breadcrumb alone does nothing for a reader who cannot see where they are.

Map old URLs before changing a live site

Export existing URLs before a redesign and mark each one: retain, merge, remove, or redirect. Retain a URL when you can. If two pages genuinely merge, use a permanent redirect to the closest replacement. If no relevant replacement exists, a clear 404 or 410 is better than sending every old address to the homepage; blanket redirects mislead visitors and hide migration mistakes.

After launch, check three things: old links reach the right destination, new pages have a single preferred canonical URL, and the sitemap contains only indexable preferred pages. Clean up redirects that bounce through several addresses.

Test the structure with a quick card sort

Put page names on cards and ask someone unfamiliar with the project to group and name them. Watch where they place “Backup and Restore,” “HTTPS Troubleshooting,” and “Server Monitoring” without telling them the intended answer. If people cannot predict what a section contains, revise its label or boundary.

Finally, follow three real routes: homepage navigation to a task, a search result into an article, and an article to its next useful page. The structure works only when readers can complete those journeys. Next, decide how to arrange the homepage's first screen, trust signals, and paths.