What Is a URL?

URL stands for Uniform Resource Locator. It is the address typed into the browser bar, the address behind a link and the address Google stores in its index. Technically, a URL is a kind of URI (Uniform Resource Identifier), the format defined by the RFC 3986 standard: an identifier for a resource. A URL is the kind that also says where that resource can be found.

Three neighbouring terms often get mixed up. The domain is only one part of the URL, the name of the server (for example, zds.es). The slug is the last part of the path, the one that describes the page in words (for this entry, what-is-a-url). The URL is the whole thing: protocol, domain, path and, if present, parameters and fragment.

To a search engine, every different URL is at first a different document. Google says so explicitly in its guide to URL structure: it treats /APPLE and /apple as two separate addresses, each with its own content. Much of the trouble with duplicate content starts here, with the same page reachable under several URLs.

RFC 3986 describes the generic syntax in five components: scheme, authority, path, query and fragment. In https://zds.es/que-es-sitemap?utm_source=newsletter#faq, the scheme is https, the authority is zds.es, the path is /que-es-sitemap, the query is utm_source=newsletter and the fragment is faq. The scheme and the path are required; the rest is optional.

The parts do not all behave the same way. The scheme and the server name are not case-sensitive, although the standard recommends writing them in lower case. The path, on the other hand, is case-sensitive unless the server decides otherwise. Characters the standard does not allow as they are, such as a space or a "ü", are encoded with the percent sign: the "ü" travels as %C3%BC.

The fragment, the part after "#", never reaches the server. The browser uses it to jump to a section of the page. That is why Google says it generally does not support fragments and advises against using them to change a page's content; for that it recommends the JavaScript History API. Browsers, for their part, read URLs according to the WHATWG URL Standard, a living standard last updated on 7 October 2026.

A well-built URL helps two readers at once. A person who sees it in a search result or a link understands what the page is about before opening it. The search engine uses it as a stable identifier to crawl, index and collect signals. If the same page answers at several URLs, those signals are split, and the search engine has to choose which version is the main one.

The same holds for AI search. When an assistant cites a source, it links a specific URL. If the page exists in several versions, for instance with and without a trailing slash, or with campaign parameters, the citation can be spread across them or point to the version you did not want.

In this glossary we handle it like this, checked on 8 October 2026 with the sitemap entry: with a trailing slash, zds.es/que-es-sitemap/ redirects with a 301 to the version without one. With a campaign parameter, ?utm_source=…, the same page is served, but with a canonical tag pointing to the clean URL. And with the Spanish slug under the German folder, /de/que-es-sitemap, it redirects with a 301 to /de/was-ist-sitemap. Written in capitals, /Que-Es-Sitemap, it also redirects with a 301 to the lower-case form. Four variants, one URL that counts.

• Use readable words instead of long ID numbers, in the language of the people who will read the page.
• Separate words with hyphens, not underscores, as Google recommends.
• Keep paths in lower case throughout, and redirect upper-case variants with a 301 if the server accepts them.
• Pick one form, with or without a trailing slash, and redirect the other.
• Drop parameters that do not change the content, and for the ones that stay, use the usual encoding: "=" between key and value and "&" between parameters.
• Mark the main URL with a canonical tag when a page also has to exist with parameters.

• Changing URLs without redirecting the old ones, so links and citations end up on a 404.
• Using fragments ("#") to load different content that should be indexable.
• Mixing upper and lower case in internal links, so the same page exists twice.
• Creating new URLs for every filter combination in a shop or listing without controlling which ones get indexed.
• Translating a page's text but leaving the slug in another language, or the reverse, without keeping track of which versions belong together.