404 errors are not bad for SEO in themselves: Google does not penalise a site for having them, and a 404 on one URL does not affect how your other pages rank. What costs you is what the 404 replaced: a page that ranked, had backlinks or was linked from your own menus. This guide explains what Google says about 404s, how to tell the harmful ones from the harmless ones, and how to work through Search Console's "Not found (404)" list without chasing every entry. For what a 404 is and how to fix one, see the 404 error; this page is part of the guide to finding and fixing broken links.
Quick answer
- No penalty. Google says 404s on some URLs do not hurt how the rest of the site ranks.
- The page drops out. A URL that answers 404 is removed from Google's index, so any rankings and traffic it had go with it.
- Fix 404s on URLs with backlinks, traffic or rankings, URLs in your own links or sitemap, and important pages that 404 by mistake.
- Leave 404s on URLs that never existed, spam and bot probes, and removed pages with no replacement and no links.
- Do not redirect everything to the home page. Google treats that as a soft 404.
- Triage the list: paste Search Console's 404 examples into the bulk URL checker, check backlinks and traffic, then fix from the top.
What Google says about 404s
Google has been consistent on this for over a decade:
- "Do 404s hurt my site?" In its 2011 Search Central blog post of that name, Google answered that 404s on URLs that no longer exist do not affect how the site's other URLs, the ones that answer 200, perform in search. 404s are a normal part of the web.
- 4xx pages are not indexed. Google's documentation on how HTTP status codes affect Search says its indexing pipeline does not index URLs that return 4xx, and removes URLs that are already indexed when they start returning one. Content served with a 404 is ignored.
- 404 is the right answer for a removed page. The same documentation, and Google's crawl budget guide, recommend returning 404 or 410 for pages that are permanently gone. Google says a 404 is a strong signal not to crawl the URL again, although it never fully forgets a URL it knows about and checks now and then.
- 4xx does not slow crawling. Except for 429 Too Many Requests, 4xx status codes have no effect on how fast Google crawls the site. 5xx errors do slow it.
So a 404 is not a signal about your site's quality. It is a statement that one page does not exist, and Google takes it at face value.
When 404 errors do hurt SEO
The harm comes from what the missing page used to do. Watch for these five situations:
| Situation | What you lose |
|---|---|
| A page that ranked now answers 404 | Its rankings and search traffic, as it drops out of the index |
| Other sites link to the URL | The value of those links, and the visitors who click them |
| Your own pages link to the URL | Internal link signals, crawl requests and visitors at a dead end |
| The URL is in your XML sitemap | Trust in your sitemap's hints |
| Hundreds of URLs 404 after a migration | All of the above, at scale; this is how migrations lose traffic |
A migration that loses rankings has not been punished for its 404s. It has removed pages Google was ranking, and the replacements are at new addresses Google has no reason to connect with the old ones. Redirects are what make that connection. The website migration guide covers the redirect map.
Soft 404s are worse than real ones
A page that says "not found" but answers 200 OK is a soft 404. Google has to guess that it is missing, keeps crawling it, and reports it separately in Search Console. The same happens when many removed pages redirect to the home page or a loosely related category: Google treats those redirects as soft 404s. A real 404 is clearer for everyone. Soft 404s shows how to find and fix them.
Which 404s to fix, and which to leave
Sort every 404 into one of these rows:
| The URL … | What to do |
|---|---|
| Had backlinks from real sites | 301 to the closest equivalent; see broken backlinks |
| Had search traffic or rankings | 301 to the closest equivalent, or restore the page |
| Is linked from your own pages | Fix the links; see broken internal links |
| Is listed in your sitemap | Remove it from the sitemap, or restore the page |
| Is an important page that broke by accident | Restore it now |
| Was removed, with no replacement and no links | Leave it 404, or use 410 Gone |
| Never existed: typos, guessed URLs, bot probes | Leave it 404 |
| Comes from spam links on other sites | Leave it 404 |
The last three rows are most of a typical 404 report. They need no action, and "fixing" them with redirects creates soft 404s. 404 vs 410 vs redirect has the decision rules in detail, with the server snippets.
How to triage the 404s in Search Console
Search Console's Indexing → Pages report groups URLs Google could not index. The group Not found (404) lists URLs that answered 404 when Googlebot last tried them, with up to 1,000 examples you can export.
A workflow that takes about an hour for a few hundred URLs:
- Export the examples from the Not found (404) group.
- Check their current status in bulk. Paste the list into the bulk URL checker, up to 1,000 per run. Some will already answer 200 or redirect, because they were fixed since Google last looked; drop those.
- Find the ones with links. In Search Console's Links → Top linked pages, export your most-linked targets and match them against the list. Anything with external links goes to the top.
- Find the ones with traffic. In the Performance report, compare against pages that had clicks in the last 16 months, or check your analytics for visits to those URLs.
- Find where your own links point to them. Inspect a few URLs: URL Inspection often shows the referring page. Run the broken link checker on your home page and main templates.
- Act on the top rows of the triage table above, and leave the rest.
- Click Validate fix on the group once you are done. Search Console rechecks the URLs over the following weeks; ones you chose to leave as 404 simply stay in the report, which is fine.
Search Console keeps reporting URLs that answer 404, sometimes for months, because Googlebot rechecks old URLs occasionally. A URL staying in the report does not mean it still hurts.
How getReport helps you see the 404s that matter
The report checks the places where a 404 costs you most:
The broken link checker requests every link on a page, up to 100 on your host and 50 to other sites, and lists the ones that answer 404, 410, 5xx or nothing, with their anchor text.
The sitemap check samples up to 25 URLs from your XML sitemap and flags any that do not answer 200, since a sitemap should never list a 404.
And the report requests a random address on your site to confirm it answers a real 404 status with a helpful page, not a 200 or a redirect to the home page. Custom 404 pages that help covers what to put on that page.
Crawl budget: when many 404s matter
On most sites, 404s have no measurable effect on crawling. Google's crawl budget documentation is written for sites with roughly a million pages, or 10,000 pages that change daily. If you run a large shop or a site that generates URLs automatically, check Search Console's Settings → Crawl stats → By response: if a large share of Googlebot's requests end in 404, find what keeps generating those URLs, such as internal links, parameters or an old sitemap, and fix that source. The log analyser shows the 404s each bot keeps hitting from your access logs, in your browser. Crawl budget for sites under 10,000 pages explains why smaller sites rarely need to worry.
Common mistakes
- Redirecting every 404 to the home page to make the report go away. Google treats it as a soft 404 and nothing is recovered.
- Chasing every URL in the report, including typos and bot probes. They are correct 404s.
- Returning 200 on the "page not found" page, which turns every missing URL into a soft 404.
- Blocking 404 URLs in robots.txt to hide them. Google then cannot see the 404 and keeps the URL in its queue longer.
- Deleting redirects in a clean-up, which turns old, linked URLs back into 404s.
- Treating a migration traffic drop as a 404 penalty. It is missing redirects; fix the map.
Questions people ask
Are 404 errors bad for SEO?
Not in themselves. Google does not penalise a site for 404s, and a 404 on one URL does not affect how the site's other pages rank. They cost you when the missing URL had backlinks, search traffic or links from your own pages, because that value and those visitors are lost. Fix those with a 301 to the closest equivalent, and leave the rest as 404.
How many 404 errors are too many?
There is no number. Google does not judge a site by its count of 404s; a large site can have thousands for URLs that never existed without any harm. What matters is which URLs they are. One 404 on a page with backlinks or traffic matters more than a thousand on typos and bot probes. On very large sites, a high share of Googlebot's requests ending in 404 is worth investigating.
Should I redirect all 404 pages to my home page?
No. Google treats redirects from removed pages to the home page, or to any unrelated page, as soft 404s, so nothing is gained for search, and visitors land somewhere they did not ask for. Redirect only URLs that have a close equivalent, such as a successor product or a merged article, and let the others answer 404 or 410.
How long does it take Google to drop a 404 page from search?
Usually days to a few weeks, depending on how often Google crawls that URL. Once Googlebot sees the 404, it removes the page from the index. Popular pages are recrawled sooner. For an urgent removal, such as sensitive content, use the Removals tool in Search Console, which hides the URL for about six months while the 404 takes effect.