A blanket 404 redirect to homepage doesn’t help SEO, and on pages that had links or search traffic it loses what they carried. Google reads a redirect from a missing page to a homepage that doesn’t replace it as a soft 404: the old URL drops out of the index just as it would with a plain 404, the links pointing at it pass nothing to the homepage, and the visitor who clicked a link to a specific page lands somewhere that isn’t it. A 404 on its own doesn’t hurt a site. The decision is made URL by URL, and for most 404s the right answer is to leave them as they are.
Below: what Google does with a 404 redirect to the homepage, whether 404 errors hurt SEO at all, a table for deciding what each 404 should become, the redirects your CMS adds without asking, and what to do if a homepage redirect is already live.
What a 404 redirect to the homepage does in Google
Two terms first. A hard 404 is a missing page that says so, with a 404 or 410 status code. A soft 404 is a missing page that answers as if it were there, with a 200 or a redirect to somewhere irrelevant, and Google treats it as missing anyway. A 404 redirect to the homepage turns the first into the second.
A redirect tells Google that the page at the old address now lives at the new one. Googlebot follows it and Google checks the claim. When the target answers what the old page answered, the old URL is replaced in the index and its signals, links included, consolidate onto the target. When the target is the homepage of a shop, a blog or a service site, it almost never answers what a product page, an article or a city page answered, and Google treats the redirect as a soft 404, which is a page it has decided is gone whatever the status code says. John Mueller put it that way in 2017, and Martin Splitt repeated it in an SEO Office Hours short in March 2025: “If it moved somewhere else, use a redirect. If it’s gone, don’t redirect me to the homepage.”
The result in Search Console is a pair of things that don’t happen. The redirecting URLs show up under Soft 404 in the Pages report instead of Page with redirect, and the referring domains that pointed at them never appear against the homepage in the Links report. The homepage ranks as it did before, and the rankings the old pages had are gone. Splitt’s other complaint is the user one: a visitor who followed a link to a specific page and lands on the homepage starts the search again, and “it annoys me as a user.”
The argument for the redirect is link equity: the old page had backlinks, and sending it to the homepage is supposed to keep their value on the site. It doesn’t, because link equity moves through a redirect only when Google accepts the target as a replacement, and an irrelevant target is exactly what it rejects. The user experience argument fails the same way. A visitor who expected a product and got the homepage bounces, and a bounce from the homepage is harder to diagnose in analytics than a visit to a 404 page that logs the missing URL.
The plugin with 200,000 installs, and what it promises
The two results at the top of Google for this query are WordPress plugins that do exactly this. “All 404 Redirect to Homepage” has over 200,000 active installations, was updated in October 2026, sends every 404 to the homepage with a 301, and says in its description that it can “Improve your SEO rank” because “404 error pages hurts the rank of your site in search engines.” Both halves of that are wrong. A 404 doesn’t hurt the site’s rankings, and the redirect it replaces the 404 with passes nothing, so what the plugin changes is the report: the 404 count in Search Console falls, the soft 404 count rises, and the site is no better placed than before. The plugin is useful for one thing, which is catching a stream of 404s on a site nobody maintains. It shouldn’t be the rule on a site that someone does.
Do 404 errors hurt SEO?
No, as a status. “Google does not penalize for 404’s (those pages drop out of the index though),” Mueller said in April 2024, and the Pages report in Search Console lists Not found (404) as a normal state for URLs that no longer exist. A site with thousands of 404s from old campaigns, deleted products and mistyped links is a site that has been around for a while.
Two things about 404s surprise people. Google keeps requesting them for a long time: in March 2026 Mueller said such URLs will “be recrawled for potentially a long time, a 410 won’t change that,” and that the recrawling means “Google would be ok with picking up more content from your site.” And the crawl they cost is small. Since Google rewrote its crawl budget documentation in July 2026 to measure capacity as the time the server spends on connections, a fast 404 is about the cheapest response a server can give, and a redirect to a full homepage costs more than the 404 it replaced.
Where a 404 does cost something is on the few URLs that still carry value: external links from pages that rank, internal links from your own navigation and articles, and search traffic in the months before the page went. Those are the URLs the decision below is for.
What to do with each 404: the decision table
Four questions settle each URL. Does anything outside the site link to it? Did it get search clicks in the last 16 months? Does anything on your own site still link to it? And is there a page that does the same job now?
| What the URL has | A replacement exists | What to do |
|---|---|---|
| External links from pages that rank, or search clicks before it went | Yes | 301 to the replacement: the new product, the merged article, the page at its new address |
| External links or search clicks | No, but a parent covers the same need | 301 to the parent only if a visitor arriving there would find what the old page offered, such as a discontinued model’s category with its successors; otherwise leave the 404 |
| Internal links from your own pages | Either | Fix the links at the source: change them to the replacement or remove them. A redirect exists for links you don’t control |
| Nothing: no links, no clicks | Either | Leave the 404. This is most of the list |
| Content removed for good, by a decision you want stated | No | 410, or the 404 it already returns; Google treats them almost the same |
The homepage appears in none of the rows. It is the right target only for URLs that were themselves a version of the homepage, an old index.html or a previous domain’s root. The test in every row is whether the target is relevant to what the old URL offered: the closest relevant page gets the redirect, and where there is none the 404 stays.
Where the four answers come from
- The list of 404s. Search Console, Pages report,
Not found (404), exported. A crawl of your own site adds the 404s Google hasn’t reached yet: in Screaming Frog, Response Codes → Client Error (4xx). - External links. In Ahrefs, Site Explorer → Best by links filtered to HTTP 404; in Semrush, the Indexed Pages report with the broken filter. Keep the URLs whose linking pages have search traffic of their own. A link from a page nobody visits and Google may not index changes nothing either way.
- Search history. Search Console’s Performance report, Pages tab, last 16 months, filtered to the URLs on the list. A page with clicks before it disappeared had demand that a replacement can pick up.
- Internal links. The Inlinks tab for each 404 in the crawl. These are the cheapest fix on the list, because the link lives on a page you edit.
On most sites this turns a list of thousands into a few dozen URLs that need a decision and a long remainder that needs none. For the external-link side of the same work, the backlink audit guide covers recovering broken links as part of reading the whole profile.
404 or 410
For Google the difference is small. Mueller in April 2024: “The difference in processing of 404 vs 410 is so minimal,” and “I can’t think of any time I’d prefer one over the other for SEO purposes.” In March 2026 he added that a 410 doesn’t stop Google recrawling the URL. A 410 is worth sending when you want the decision recorded in the response itself, for a section removed on purpose or content taken down for legal reasons. Changing thousands of existing 404s to 410s to speed anything up is work with nothing at the end of it.
Redirects your CMS adds without asking
Some 404s never reach a 404, because the platform redirects them first. WordPress does it in two ways. When a post’s slug changes, core stores the old one and sends visitors from it to the new address with a 301, which is what you want. When a URL matches nothing at all, core guesses: redirect_guess_404_permalink() looks for a published post whose slug starts with the requested one and sends a 301 there. For a typo that is helpful. For a deleted post whose slug is the beginning of another post’s slug it is a redirect to an unrelated page, which is the soft 404 from the first section, made by the CMS. Since WordPress 5.5 the guess can be switched off with the do_redirect_guess_404_permalink filter or limited to exact matches with strict_redirect_guess_404_permalink. Before deciding, request a few deleted URLs and see where they actually land.
Shopify asks. When a product’s URL handle changes, the editor offers “Create a URL redirect” from the old handle to the new one, and the box is ticked by default. Deleted products get no redirect at all, which on a catalogue is where most of the decision table’s first two rows come from. On an ecommerce site the commoner mistake comes earlier: a product that is temporarily out of stock should stay live with a 200 and say when it’s back, and only a discontinued one enters the table.
The status code matters as well. A redirect to a replacement should be a 301; a 302 tells Google the old URL is coming back and keeps it in the index, which is covered in our guide to 302 redirects and link equity.
A 404 page that returns 404
The custom 404 page is for the visitor, and it has one technical requirement: it has to return the 404 status code. A friendly “page not found” served with a 200 is a soft 404 of its own, and a page that shows the message and then forwards to the homepage after five seconds is the homepage redirect with a delay. What goes on it is plain: a sentence saying the page isn’t there, a search box, links to the main sections and the most visited pages, and nothing that moves the visitor without a click. It doesn’t need noindex, because a page that returns 404 isn’t indexed anyway.
If the homepage redirect is already live
- Export what it covers. Before switching the rule off, get the list of URLs it has been catching: the plugin’s log if it keeps one, the server logs, or Search Console’s
Soft 404list. - Run the decision table on that list. Most of it goes back to plain 404. The URLs with links or past traffic and a real replacement get individual 301s.
- Switch the blanket rule off and ship the individual redirects in the same release, so the URLs that should redirect never spend a day as 404s. The rule itself is usually one catch-all line, in the plugin’s settings, in .htaccess or in the server config, and it has to go from every layer that has one. Check the new redirects for chains while you’re there: an old URL that redirects to a page that itself redirects should point at the final address.
- Clean the sitemap and the internal links. URLs that now return 404 come out of the XML sitemap, and internal links to them get changed at the source, so Googlebot stops being sent to pages you’ve decided are gone.
- Read it again in four to eight weeks. The
Soft 404count should fall andNot found (404)rise, which is the report becoming honest. The redirected URLs should showPage with redirect, and their referring domains should start appearing against the replacements in the Links report.
If the redirects were part of a migration and the traffic fell after it, the redirect is usually one cause among several, and the others are in our article on the 301 redirect SEO penalty.
404 redirect FAQ
How many 404 errors are too many?
There’s no number. A site’s total 404 count says how old it is and how many links to it were mistyped. What matters is the share of 404s among URLs that had external links from ranking pages or search clicks, and on most sites that share is a few dozen URLs whatever the total.
Should a 404 redirect be a 301 or a 302?
A 301, when the page has a real replacement. A 302 passes PageRank too, but it tells Google the old URL is coming back, so the old address stays the one Google keeps.
Should I redirect 404s from mistyped URLs?
Only the ones that get real visits. A mistyped link on another site that sends traffic is worth a 301 to the page it meant; a mistyped URL nobody has requested twice isn’t worth a rule in the config.
What I don’t know is whether a homepage redirect left on for years costs a site anything beyond the signals of the URLs it caught. Google treats them as soft 404s, and I haven’t seen evidence either way that a large share of soft 404s changes how Google reads the rest of a site, so I don’t claim it does. If you’d rather have the 404 list read against links and traffic for your site, it is part of our SEO audit services.