What Is a 404 Error? Fixing the Pages That Vanished

Jul 24, 2026 · SEO Glossary

Picture this: you send a newsletter to 3,000 subscribers with a link to your new case study. Forty minutes later, replies start arriving — "the link doesn't work." Someone renamed the page's URL during a last-minute edit, and every click from that email now lands on a page that says 404 Not Found.

If you've run a website for more than a month, some version of this has happened to you. So what is a 404 error, exactly? It's the HTTP status code a server returns when it's reachable and working but has no page at the requested address. The server is home; the room you asked for doesn't exist.

What is a 404 error in plain terms?

Every web request gets answered with a status code. Codes in the 200s mean success, 300s mean redirection, 400s mean the client asked for something wrong, and 500s mean the server broke. A 404 sits in that 400 family: the request was understood, the server searched, and nothing lives at that URL.

That's different from its neighbors, and the differences matter when you're debugging:

• 404 Not Found: no page at this address. Maybe there never was; maybe it was deleted or renamed.
• 410 Gone: a deliberate statement that the page existed and was intentionally removed. Google drops 410s from its index slightly faster than 404s, though the difference is marginal.
• 500 Internal Server Error: the page might exist, but the server crashed trying to build it. A code problem, not a missing-page problem.
• Soft 404: the sneaky one — a page that says "not found" to humans but returns a 200 OK code to machines. More on this below.

Where 404s actually come from

In the audits I've run, 404s trace back to a handful of causes, in roughly this order of frequency:

1. URL changes without redirects. Someone edits a slug, restructures categories, or migrates platforms, and the old addresses die. This is the big one — it's why 301 redirects exist.
2. Typos in links. A missing letter in an href, a stray space, a hardcoded staging URL like dev.example.com that shipped to production.
3. Deleted content. Discontinued products, expired job listings, retired landing pages.
4. External sites linking wrong. A blogger mistypes your URL, and you inherit a 404 you never created. Search Console's link report reveals these.
5. Case and trailing-slash mismatches. On some servers /About/ and /about/ are different URLs, and only one exists.

Do 404s hurt your SEO? The honest answer

Mostly no — with two important exceptions. Google has been clear that 404s are a normal part of the web and having them doesn't lower your site's quality score. Googlebot finds a dead URL, drops it from the index, and moves on. If a page genuinely has no replacement, a 404 is the correct response, not a failure.

The exceptions:

• 404s with backlinks or traffic. If an old URL has links from other sites pointing at it, every one of those links is a wasted vote. Redirect that URL to a relevant live page and reclaim the value. Same logic if analytics shows the dead URL still gets visits.
• Internal links to 404s. When your own menus and articles link to dead pages, you're burning crawl budget and telling users the site is unmaintained. These are 100% within your control and worth fixing every time — I cover the hunt in detail in the broken links guide.

Soft 404s: the sneakier cousin

A soft 404 happens when a missing page returns a 200 OK status while displaying a "not found" message — or a nearly empty page, or an unwanted redirect to the homepage. Users see an error; machines see a healthy page. Google's systems try to detect these and flag them in Search Console's page indexing report, but in the meantime you've got junk competing for crawl attention and, occasionally, thin "not found" pages getting indexed.

The most common cause I see: single-page apps and misconfigured CMS themes that render an error message inside a template that still returns 200. The fix is server configuration — make the error state actually send a 404 code. You can verify with curl -I or the Network tab in Chrome DevTools. The HTTP 404 status code reference covers the server-side details.

Build a 404 page that saves the visit

You can't prevent every 404, so make the landing soft. A good error page turns a dead end into a detour:

• Say what happened in human words. "We couldn't find that page" beats a bare "404 Not Found."
• Offer a search box. The visitor came looking for something specific — let them keep looking.
• Link to your most useful pages. Homepage, top categories, popular articles. Three to six links, not a sitemap dump.
• Keep your normal header and navigation. A stripped-down error page feels like falling off the site entirely.
• Skip the redirect-to-homepage reflex. Automatically bouncing every 404 to the homepage confuses users mid-task and creates soft 404s in Google's eyes.

One team I advised added a search box and four category links to a previously blank 404 page; their analytics showed the exit rate from that page drop from 89% to about 61%. Small effort, real recovery.

Find your 404s before your visitors do

Three sources, used together, catch nearly everything. Google Search Console's "Page indexing" report shows the 404s Google has stumbled into, including ones caused by external sites. Your server logs or analytics show which dead URLs real humans are hitting. And a crawler walks every link on your site and flags internal references to missing pages — scan your whole site and you'll have the complete list, with the linking source for each dead URL so you know exactly what to edit.

Then triage: redirect the dead URLs that have links or traffic, fix the internal links pointing at them, and let the rest 404 in peace. That's the entire playbook — no heroics required, just a check every month or two so the list never gets long.