DefinitionThe HTTP status code a server returns when a requested page does not exist. Too many 404s waste crawl budget and frustrate visitors; broken links to them should be fixed or redirected.
A 404 is the server’s way of saying “I looked, and there is nothing here”. It is a normal part of how the web works: pages get deleted, products sell out for good, people mistype URLs. The mistake is not having 404s, it is having 404s on URLs that still matter: pages with backlinks, pages in the navigation, pages that Google still sends visitors to. The second mistake is the opposite one: redirecting every dead URL to the homepage so that nothing ever returns a 404, which confuses Google and irritates visitors just as much.
Why 404s cost traffic and revenue
A 404 on a page nobody links to and nobody visits costs nothing. A 404 on a page that used to rank costs everything that page earned. A German seed webshop on PrestaShop that I audited had removed several hundred discontinued products over two years without redirects. Around forty of those URLs still had links from forum threads and gardening blogs, and a few still had rankings weeks after removal. All of that authority was leaking into nothing. Redirecting those forty URLs to the closest current product or category took an afternoon and recovered rankings on a handful of category pages within the following month.
There is also the visitor side. Someone clicking a Google result or an old email and landing on “Page not found” leaves. On a webshop that is a lost sale; on a B2B site it is a lost lead. And there is the crawler side: Googlebot has a limited appetite for any site, and every request spent on a dead URL is a request not spent on a page you want indexed. This is why 404 management is a standard part of technical SEO.
How to find the 404s that matter
Google Search Console is the first stop. The Pages report under Indexing lists URLs that returned “Not found (404)” when Google tried to crawl them. That list tells you what Google still knows about, which is more useful than what your own site links to. Export it and add two columns: does the URL have backlinks (check in Ahrefs Site Explorer under Best by links, filtered on 404 status), and did it have traffic before it disappeared (check in GA4 with a longer date range).
Then crawl the site with Screaming Frog. The Response Codes tab shows every internal link that points at a 404, and the Inlinks pane shows exactly which pages contain those links. These are the ones you can fix directly in the content or navigation. Also check that the 404 page really returns a 404 status code and not a 200: open the browser dev tools, go to the Network tab, load a nonsense URL and look at the status. A “soft 404” that returns 200 with a “not found” message is worse than a real one, because Google may keep it in the index.
Common mistakes with 404s
Redirecting everything to the homepage. Google treats a redirect to an unrelated page as a soft 404 anyway, and the visitor who wanted a specific product is now staring at a homepage. Redirect to the closest relevant page or let it 404.
Deleting products without a plan. Webshops remove thousands of products over time. Each one needs a decision: redirect to a replacement, redirect to the category, or 404 with a “no longer available” page that suggests alternatives.
Leaving dead URLs in the sitemap. An XML sitemap full of 404s tells Google the sitemap cannot be trusted. Sitemaps should only list live, indexable URLs.
A useless 404 page. A blank page with “Error 404” and nothing else. A good one has the site’s navigation, a search box and a few links to popular categories, so the visit is not automatically lost.
Confusing 404 with 410. A 410 (Gone) tells Google the removal is permanent and tends to get URLs dropped from the index faster. For large clean-ups of URLs that will never return, 410 is the cleaner choice.
How I handle 404s in client work
I do not try to reach zero 404s; that is not a useful target. I build a short list of dead URLs that have either backlinks, historical traffic or internal links pointing at them, and I fix those in order of value. Backlinked URLs get a 301 to the most relevant live page. Internally linked ones get the link fixed at the source. Everything else is allowed to return a proper 404 or 410 and drop out of the index on its own.
For webshops I also set up a process rather than a one-off fix: a rule in the platform or a monthly check that catches removed products before they become a problem, and a 404 page that shows alternatives from the same category. On sites with a lot of churn I add a Looker Studio report fed by Search Console so the client can see new 404s appear each month without needing me to run a crawl. Small, boring, and it prevents the kind of slow authority leak that is hard to spot until rankings have already dropped.
Frequently asked questions
Do 404 errors hurt my rankings?
Not directly. Google has said many times that 404s on a site do not lower the ranking of other pages. What hurts is the indirect effect: lost backlinks that used to pass authority, lost internal link paths, and visitors who leave. So a 404 on a forgotten blog post is harmless, and a 404 on your best-linked category page is expensive.
Should I redirect a discontinued product?
If there is a direct successor or a very similar product, redirect to it. If not, redirect to the parent category only when the product page has backlinks or traffic worth keeping. Otherwise a 404 or 410 with a helpful page showing alternatives is the honest option and the one Google prefers.
How quickly does Google remove 404 pages from the index?
Usually within a few weeks for pages that are crawled often, longer for deep pages that Google rarely visits. Google will also keep re-checking known 404s occasionally for months, which is normal and does not need fixing. A 410 status speeds the process up slightly, and removing the URLs from the sitemap and internal links helps more.