A 404 error is the HTTP response 404 Not Found. The server cannot provide the requested resource, or does not disclose that it exists. A missing page can correctly return 404; an important live page returning it needs investigation.

What is a 404 error?

The address in a link and the page that the server can serve are separate things. Someone can request a misspelled URL, follow an old bookmark, or visit a page you deleted. A 404 reports that the requested resource is unavailable. It does not say whether the situation is temporary or permanent.

HTTP groups 404 among client error responses, but that label does not establish who caused the problem. A deployment, routing rule or content deletion can break a valid address. The HTTP semantics standard defines the response; diagnosing its cause requires checking your site.

Why 404 errors matter for SEO

A normal missing-page response is not a sitewide ranking penalty. Google explained this in its 2011 guidance on 404 errors; current Search Console guidance also says that a removed page without a replacement can appropriately return 404. The useful question is which addresses are missing and whether people still need their content.

A broken booking link can stop a customer from making an appointment. An accidentally deleted service page can disappear from search once Google processes its response. These are specific losses of access and content, rather than evidence that every remaining page has been downgraded. There is no universal bounce-rate multiplier or removal deadline.

Do not assume that a long 404 list proves a crawl-budget shortage. Google's crawl-budget guidance mainly addresses very large or frequently changing sites. It recommends genuine 404 or 410 responses for retired pages and distinguishes them from soft 404s. Prioritize active customer journeys and unintended errors before chasing a zero-error dashboard.

How a 404 response works

This illustrative exchange uses an address that does not exist. The status line and the visible error message communicate different parts of the response.

GET /removed-guide/ HTTP/1.1
Host: example.com

HTTP/1.1 404 Not Found
Content-Type: text/html

<h1>We could not find this page</h1>

A branded error page can include useful navigation and still return 404. Changing its appearance does not require changing the HTTP status to 200. Google says it does not index pages returning 404 and removes previously indexed URLs after processing their unavailable status. The response does not promise when that processing will happen.

How a soft 404 differs

A soft 404 is Google's assessment that a page is missing or unhelpful despite a response that appears successful, commonly 200 OK. An error message, empty template or failed content load can produce this mismatch. Google can flag and exclude such a page; it is incorrect to say that the 200 status automatically gets an empty page indexed.

Inspect the rendered page as well as its response. If the URL should contain useful content, restore that content or fix the failed resource. If it really has no page, return a missing-page status. See the Page indexing report documentation for the distinction between Not found (404), Soft 404 and other exclusion reasons.

404, 410 and other responses compared

Choose a response that describes the actual resource
ResponseWhat it communicatesAppropriate situation
404 Not FoundNo current resource is provided; permanence is unspecified.A nonexistent address or removed page without a replacement.
410 GoneThe resource is unavailable and the condition is likely permanent.A deliberately retired resource when you want to communicate that distinction.
301 or 308The resource has moved permanently.An old address with a relevant equivalent at another address.
200 OKThe request succeeded.An actual page with its intended content, not a missing-page template.
503 Service UnavailableThe service is temporarily unavailable.A genuine temporary service outage, rather than intentional page deletion.

These are protocol meanings, summarized from RFC 9110. Google's crawler status-code guidance groups 404 and 410 together for Search handling. Do not select 410 solely to promise faster deindexing.

Illustrative 404 error examples

A renamed service page: a practice changes /services/consultation/ to /appointments/consultation/. If the old address has no rule, a bookmark can fail even though the service still exists. Redirect the old address to its genuine replacement and update links you control.

A retired guide: an obsolete guide has no useful equivalent. A real 404 or 410 is more truthful than redirecting it to an unrelated homepage. Remove its sitemap entry and correct any links that still promise the deleted information.

A campaign typo: an email links to /product-lauch/ instead of /product-launch/. Correct the campaign link where possible. A deliberate redirect from the mistaken address can help people who already received it, provided the destination is the intended page.

These are examples of decisions, not customer case studies or measured ranking results.

Use a permanent redirect when the old URL's content or purpose has an appropriate new home. Ask whether a visitor following the old link would find what they expected at the destination. A backlink count alone does not make an unrelated page a suitable replacement.

If an important page vanished accidentally, restoration may be the right fix. If it is intentionally gone with no replacement, a missing-page response is appropriate. Avoid a catch-all rule that sends every unknown address to the homepage: it hides routing mistakes and gives visitors no explanation.

A practical 404 review checklist

  1. Collect the exact URLs. Use the Page indexing report, a website crawl, relevant request logs and reports of broken customer journeys. Each has different coverage.
  2. Reproduce the response. Test the same hostname, path and query string. Inspect both the status and the visible content.
  3. Classify the cause. Separate a typo, deliberate retirement, move, broken deployment and temporary service failure.
  4. Choose the appropriate action. Restore intended content, redirect a genuine move, or keep a truthful missing-page response.
  5. Repair controlled links. Update navigation, article links and sitemap entries so they point directly to useful live content.
  6. Retest after release. Check the public response and destination, not only a preview. Record unresolved URLs and revisit Search Console after another crawl.

For a GET request that prints response headers without saving the body, try curl -sS -D - -o /dev/null https://example.com/removed-guide/. This is a syntax example, not a test of your site. curl -I uses HEAD; a differently configured server may answer HEAD and GET differently. Follow a suspected redirect chain separately and inspect the final page's relevance.

Common mistakes to avoid

A label on the page does not set its HTTP status

Do not treat a visible “404” heading as proof of a 404 response. Check the server response after publishing your custom error page. Likewise, a successful status does not establish that Google can render the intended content.

A robots.txt block prevents crawling; it does not let Google inspect a newly returned 404. Use the appropriate missing-page response for a retired URL rather than hiding it simply to clear an error list. Keep the error page's helpful links keyboard accessible and use a clear explanation instead of automatically moving the visitor somewhere else.

Frequently asked questions

Why does Search Console still show a 404 after I fixed it?

The report can describe Google's last crawl rather than the response you see now. Compare its crawl date with your release date, then inspect the live URL. A current successful response does not establish that Google has reprocessed or indexed the page.

Why does Google request a URL that never existed?

Google can discover an address through a link, including an incorrect one. A request does not prove that the page previously existed. If the address has no intended resource or relevant replacement, a genuine 404 is appropriate.

Read about 301 redirects for permanent moves, soft 404s for content/status mismatches, redirect chains for intermediate hops, and 500 errors for server failures. Crawl budget and link equity explain separate search concepts; neither supplies a universal numerical impact for a broken URL.

Sources and editorial note

  1. RFC 9110: 404 and related HTTP semantics
  2. Google: HTTP status codes and crawler handling
  3. Google Search Console: Page indexing report
  4. Google: Crawl-budget management
  5. Google: Do 404 errors hurt my site? (2011)

Revised with AI assistance against the primary references above, checked 12 October 2026. Examples are illustrative. This entry does not report customer results, a site audit or named expert approval. Its original publication date is preserved.

Explore this topic: 404 Errors and SEO: How to Find, Fix, and Prevent Them.