HTTP 404 Not Found

404 says "not found", with no statement about the future. The URL may work tomorrow.

410 Gone says the absence is deliberate and expected to be permanent. Search engines treat it as a stronger signal to drop the URL, and tend to retry it less.

In practice: 404 for anything unknown, 410 when you deliberately removed something and want it gone from the index faster. Neither is required — 404 for everything is acceptable behaviour.

A soft 404 is a page that tells the visitor the content does not exist while returning status 200. Common in single-page applications, where the router renders a "not found" view without the server ever knowing.

It is worse than a real 404 because every automated consumer is misled: link checkers report the link as working, and search engines may index the page — though Google detects many of these and reports them as soft 404s in Search Console.

The fix is for the server to return the status, not only the view.

`curl -I https://example.com/does-not-exist` prints the status line. A correct configuration answers `HTTP/2 404`.

Test a URL you know is invalid on each major section of the site. Applications frequently return 404 correctly for static routes and 200 for application routes.

In Search Console, the Page indexing report separates "Not found (404)" from "Soft 404", which is the fastest way to discover the second.

Returns status 404, first and above all.

Says plainly that the page does not exist, and offers a way onward: search, or the main sections.

Does not redirect to the home page. A redirect turns "this does not exist" into "this moved here", which is false, and it hides broken links from everyone including you.

Typos and truncated URLs, from emails and chat clients that break long links across lines. These are unavoidable and entirely harmless.

Deleted content: a product retired, an article unpublished, a category merged. The 404 is correct, and the internal links still pointing at it are not.

Migrations. A move from one CMS to another almost always changes the URL pattern, and the redirect map is almost always incomplete. The 404s that appear a week after a relaunch are the map's missing rows.

Case sensitivity. Most web servers treat `/About` and `/about` as different paths, so a link written with the wrong capitalisation is a genuine 404 on Linux hosting and works fine on a developer's machine.

The distinction worth making in each case is whether anything on your own site links to the URL. A 404 nobody links to is the web working; a 404 reached from your own navigation is a defect.