404 Not Found is an HTTP response that says the requested URL has no current resource, so users and crawlers cannot retrieve the expected page.
What 404 Not Found means
A real 404 is a response status, not simply a page whose design says “not found.” It is appropriate when no relevant replacement exists or when a URL was never valid. It differs from a redirect, which sends a visitor to a specific successor, and from a soft 404, where a server returns success while showing an error-like page.
Decision rule and evidence
Map a retired URL only when a close, useful replacement exists. Otherwise return a genuine 404 and remove internal links to it. Do not route every missing URL to the home page: that makes the original request harder to resolve and can create soft-404 signals.
Practical workflow
curl --head https://example.test/retired-page
# Expect: HTTP/... 404, not 200 and not a blanket home-page redirect.For a retired product or article, compare the old title, purpose and inbound links with the proposed destination. A redirect is defensible only when the destination answers substantially the same need; otherwise preserve the 404 and repair the internal link that created the dead path.
- Export internal links and response codes from a bounded crawl.
- Separate typo URLs, deleted content, and URLs with a clear successor.
- Restore an accidentally removed page; otherwise add one direct redirect only for the successor case.
- Make the custom 404 page helpful while keeping the HTTP response at 404.
- Re-crawl the former source pages and sample the response in a browser and an HTTP client.
Verification and common mistakes
Check the requested URL itself, not only the visual error template. Record the final status, the referring internal URL, and whether a replacement truly serves the same intent.