404 error: what does it mean and should you fix it?

What does a 404 error mean?

A 404 error means that the server did receive the request, but found no resource at the requested address. “Not Found” means just that: nothing could be found there.

The number is read digit by digit. According to the same standard (section 15), the first digit of a status code gives its class, and the next two play no role in that classification. 4xx codes form the client error class: the request contains an error or cannot be fulfilled.

For a site owner, handling 404s is part of technical SEO: knowing which dead addresses deserve action.

404, 410, soft 404: three responses to tell apart

These three responses look alike to the visitor, not to a search engine.

The standard prefers 410 when the server knows the deletion is permanent, without requiring every vanished resource to be marked that way (RFC 9110, section 15.5.11). Google, for its part, treats all 4xx codes except 429 the same way, according to its documentation on HTTP status codes (updated on February 4, 2026). The idea that a 410 is taken into account faster than a 404 does not appear there.

Where do 404 errors come from?

A 404 error appears as soon as an address no longer matches, or never matched, a resource on the site. Common causes:

• Page deleted with no replacement content.
• Page moved or slug changed without a redirect.
• Typo in an internal link, in an external link or in the address typed by the user.
• URL generated by a bot: the Search Console help mentions addresses built by Googlebot when it follows JavaScript or other embedded content.
• Redesign without a redirect plan: during a site redesign, every old address that changes returns a 404.

A link that leads to a 404 is called a broken link, or dead link (MDN Web Docs, page last modified on July 8, 2026).

Does a 404 error hurt SEO?

No, as a general rule. The Search Console help page on 404 errors says so (undated page, accessed on October 19, 2026). 404 errors have no negative impact on a site’s performance in search results. Many of them can be ignored.

What Google does with a URL that returns a 404:

• it does not use its content;
• it removes it from the index if it was there;
• it does not process a new URL that returns a 404;
• it keeps requesting it for a while, less and less often, and there is no way to make it forget the URL for good.

404s have no effect on how fast the site is crawled. A dead address remains a URL that Googlebot revisits from time to time: it is a crawl budget matter, not a penalty.

Two things are lost, however. External links that pointed to the deleted page now lead nowhere. The visitor who arrives through such a link leaves without an answer.

How do you find the 404 errors on your site?

Cross-check Search Console, a crawler, the server logs and the HTTP headers: each one sees part of the 404s.

1. “Page indexing” report in Google Search Console, reason “Not found (404)”. It only shows URLs that returned a 404 during the previous month (help on page indexing).
2. “Crawl Stats” report, response type “Not found (404)”: the 404s received by Googlebot.
3. Crawler such as Screaming Frog: it spots the internal links that end in a 404.
4. Server logs: log file analysis shows the 404s actually requested by Googlebot and by visitors.
5. Checking the returned code: curl -s -D - -o /dev/null <address> displays the HTTP headers, and therefore the real code, whatever message appears on screen.

To place this check within a full audit, download the technical audit checklist as a PDF.

Should you fix a 404 error?

Not always. Google generally advises fixing only the 404 errors that the site itself links to or that appear in a sitemap. The rest is sorted according to where the URL comes from.

How to fix a 404 error, step by step

1. List the URLs returning a 404 from Search Console, the crawler and the logs.
2. Sort by source: internal link, sitemap, external link (backlink), unknown origin.
3. Act according to the table above: fix the link, set up a 301 redirect or leave the 404.
4. Update the sitemap so that it only lists valid URLs.
5. Check the code returned by the URL you dealt with, then request a new crawl if needed: see how to get a page indexed.
6. Give it time: according to Google’s help, the report should stop showing the error after about a month.

There is no point using the URL removal tool: a request does not remove the error from the report. It also only lasts about six months (Google help page “Remove a page hosted on your site from Google”, updated on December 10, 2025).

Another possible delay: caching. RFC 9110 declares a 404 response “heuristically cacheable” unless explicit cache controls say otherwise. One can infer that a fixed 404 is sometimes still served by a cache for a while.

Soft 404: when the error page does not say “404”

A soft 404 is a page that displays a “Page not found” type of message while returning a success code instead of a 404 code. Google reports it in Search Console and recommends a real 404 code for pages that cannot be found.

Three situations produce them:

• Error page served with a 200: the site says “not found”, the server answers “success”. For Google, a 2xx code does not guarantee indexing.
• Mass redirects to the home page: Google’s help counts this practice among soft 404s, which confuse users and search engines.
• Client-side routing in single-page applications: JavaScript displays the error, but the server has responded 200. Google offers two solutions in its JavaScript SEO basics (updated on March 4, 2026). The first: a JavaScript redirect to a URL where the server responds 404 (for example /not-found). The second: a <meta name="robots" content="noindex"> tag added to the error page. Details in our guide to JavaScript websites.

To check a doubtful page, run a live test in Search Console’s URL Inspection tool, then open “View tested page”.

The ErrorDocument trap on Apache

The directive can also be used in an .htaccess file.

404 error on WordPress

WordPress builds its error page from the theme. A classic theme uses the 404.php file, or index.php if there is none, according to the template hierarchy (updated on July 24, 2024). A block theme uses the templates/404.html template: this is the case for the Elev8 Lab theme.

Before displaying this page, WordPress tries to guess the intended address. By default, the redirect_guess_404_permalink() function looks for a post whose slug starts with the requested slug. The do_redirect_guess_404_permalink filter (since WordPress 5.5) turns this guessing off; the strict_redirect_guess_404_permalink filter makes it strict.

Observation on elev8-lab.fr, made on October 19, 2026 with curl: the address /seo/redirection-30/, with a truncated slug, responds with a 301 to /seo/redirection-301/, along with the X-Redirect-By: WordPress header. The address /redirection-301/, without the category, responds with a 404, by contrast. The guessing catches some incomplete links, not all of them: it does not replace a redirect set up by hand.

What should a useful 404 page contain?

A useful 404 page first returns a real 404 code. It then helps the visitor move on: a clear message, a search field, links to the home page and to the main sections.

MDN considers a custom 404 page “a good thing in moderation”. Humor is fine, as long as the visitor is not left in doubt about why the error occurred.

The Elev8 Lab 404 page follows this pattern. It shows “404”, the heading “This page does not exist (or no longer does)” and a sentence of explanation. It also offers the native WordPress search and two buttons: “Back to home” and “Browse topics”. Observation of October 19, 2026: a URL that does not exist on the site does return 404 Not Found, with a Cache-Control header that prevents caching. So the site does not serve soft 404s.