Implementing Proper 404 Error Pages

Why the status code is the whole point

Search engines decide what to do with a URL largely from its HTTP status code, not from the words on the page. When a page no longer exists, the server must say so in the response header. A correct configuration returns 404 Not Found for a missing resource, or 410 Gone when the removal is deliberate and permanent. The design of the page a human sees is secondary. You can ship a beautiful not found template, but if the server answers 200 OK you have created a soft 404, and crawlers will keep requesting and trying to index a page that is not real.

Soft 404: the most common mistake

A soft 404 happens when a missing URL returns a success status. Typical causes include single page applications that render an error view client side while the server still sends 200, catch all CMS routes, and well meaning redirects that send every unknown URL to the homepage. Google flags these under Soft 404 in the Search Console Pages report. They waste crawl budget, keep dead URLs eligible for indexing, and hide genuinely broken links from your reports. Confirm the real status from the command line rather than trusting the browser view:

# Check the true status a URL returns
curl -I https://example.com/this-page-was-deleted

# Correct response for a missing page
HTTP/1.1 404 Not Found
Content-Type: text/html; charset=UTF-8
X-Robots-Tag: noindex

# A soft 404 looks like this and must be fixed
HTTP/1.1 200 OK

Status codes at a glance

Server configuration

Return the right status at the server layer, then point it at a friendly template. Examples for the three most common stacks:

# Apache .htaccess
ErrorDocument 404 /404.html

# Nginx
error_page 404 /404.html;
location = /404.html { internal; }

# WordPress: a 404.php template in the theme already
# returns the correct 404 header via the template hierarchy

For headless and single page apps, make sure the server or edge function inspects the route and sends a real 404 status for unknown paths before the client renders its error view. Sending the status from the edge is what stops the soft 404 pattern.

What a helpful 404 page should contain

Once the status is correct, the human facing page should reduce the chance a visitor leaves. Include a clear message that the page was not found, a site search box, links to popular or related sections, and a prominent link back to the homepage. Keep your global navigation and header so the visitor is not stranded. Do not auto redirect a 404 to the homepage: that recreates the soft 404 problem and confuses both users and crawlers, who see a success page where a missing one belongs.

Monitoring and crawl impact

Broken links that produce 404s waste crawl budget, so large sites should watch the volume. Track the Not found (404) and Soft 404 states in the Search Console Pages report, and periodically crawl your own site to catch internal links pointing at removed URLs. Because error responses consume requests, sites concerned about efficiency should read the guidance on crawl budget. When you are choosing between a 404 and a redirect, apply the same logic used for redirect handling: redirect only when there is a genuinely equivalent destination, otherwise return a clean 404 or 410. The X-Robots-Tag on error responses is part of broader HTTP header optimization that keeps low value responses out of the index.

Frequently asked questions