Linkbruno

Soft 404s on your site and in the links you point to

Updated 19 November 20268 min read

How a Linkbruno report shows it
Soft 404
Destination answers 200 but shows “not found”
partner.example/integrations/harbor
on /integrations, in the text
Check your site for this

Up to 25 pages and every link on them. No signup.

On this page
  1. What counts as a soft 404
  2. Where soft 404s come from on your site
  3. Check a URL with curl
  4. How to fix soft 404s on your site
  5. Submitted URL seems to be a Soft 404
  6. Cleaning up soft 404s at scale
  7. The other side: links you point to
  8. How soft 404 detection works, and where it goes wrong

A soft 404 is a URL that tells the reader the page doesn't exist but answers 200 OK instead of 404 Not Found. On your own site, it means the server is misreporting missing pages, and Google Search Console lists them under "Soft 404". In the links you point at other sites, a soft 404 is a dead link that every status-code checker reports as working. On your site, the fix is to send the right status code. For outbound links, it's to replace or remove the link once you've confirmed it's dead.

What counts as a soft 404

Google defines it like this (Google Search Central):

A soft 404 error is when a URL that returns a page telling the user that the page does not exist and also a 200 (success) status code.

In Search Console's Page indexing report, the Soft 404 reason means the page "returns a user-friendly 'not found' message but not a 404 HTTP response code" (Search Console Help). Google casts a wider net than error templates, though:

  • Empty or near-empty pages. "If the content suggests an error for Google Search, an empty page or an error message, Search Console will show a soft 404 error" (Google, HTTP status codes).
  • Many old URLs redirected to one unrelated page, usually the home page. Google's site-move guide says this "might be treated as a soft 404 error" (Google, site moves).
  • Real pages that didn't render properly for Googlebot, for example because a script or other resource failed to load (Google Search Central).

Where soft 404s come from on your site

Google lists four common causes: a missing server-side include file, a broken connection to the database, an empty internal search result page, and "an unloaded or otherwise missing JavaScript file" (Google Search Central). In practice they cluster around a few setups:

  • A catch-all route that renders a "Page not found" template without setting the status code. It's common in custom apps and in single-page apps, where the server sends the same HTML shell with a 200 for every URL and JavaScript draws the error (Google, JavaScript SEO).
  • Error pages configured with a full URL. In Apache, ErrorDocument 404 https://example.com/404.html makes the server send a redirect, so "the client will not receive the original error status code" (Apache docs). In nginx, error_page 404 https://... answers with a 302, and error_page 404 = /404.php; passes on whatever status the script returns, so a script that never sets 404 sends 200 (nginx docs). Use a local path instead: ErrorDocument 404 /404.html or error_page 404 /404.html;.
  • Empty category, tag, filter and internal search pages ("No products found").
  • Deleted products or listings kept as a shell that says "no longer available" and nothing else. Stores see this a lot; the Shopify guide covers what to do with sold-out and discontinued products.
  • A plugin or rule that sends every missing URL to the home page. It looks tidy, but it's one of the few kinds of broken link that costs you in search.

Check a URL with curl

Start by asking for a page that can't exist:

curl -s -o /dev/null -L -w '%{http_code} %{url_effective}\n' \
  "https://example.com/blog/no-such-page-$RANDOM$RANDOM"

404 followed by the same URL is correct. 200 with the same URL means your not-found template returns 200. 200 https://example.com/ means missing pages redirect to the home page. Repeat it in each section of the site (/blog/, /shop/, /docs/), because different sections are often served by different software.

For a specific URL, read the headers, then the page title:

curl -sI https://example.com/old-page
curl -sL https://example.com/old-page | grep -io '<title>[^<]*'

curl -sI sends a HEAD request, and some servers answer HEAD differently from GET. If the result looks odd, confirm with a GET: curl -s -o /dev/null -w '%{http_code}\n' https://example.com/old-page. A 200 with a title like "Page not found" or "Nothing here" is a soft 404.

How to fix soft 404s on your site

Situation Fix
Page removed, nothing replaces it Return 404 or 410. A custom error page is fine as long as the status code is 404
Page moved, or a clear replacement exists 301 to the replacement
Page is real but thin or empty Add the missing content, or merge it into a fuller page and 301
Page is real but didn't render for Google Run a live test in URL Inspection and fix the resources that failed
Empty filter, tag or search page Add <meta name="robots" content="noindex">, or return 404 when there's nothing to show
Single-page app error view Redirect to a URL that returns 404, or add a noindex robots tag from JavaScript
Catch-all redirect to the home page Remove it. Redirect moved pages one by one and let the rest return 404

Most of these come straight from Google: the troubleshooting guide for removed, moved and badly rendered pages, the Search Console help for thin pages (it recommends "adding more information on the page"), and the JavaScript guide for single-page apps. For a page Google flagged that looks fine to you, Search Console's help suggests running a live URL inspection test and clicking View tested page to see a screenshot of what Googlebot rendered. Once fixed, click Validate fix on the Soft 404 details page in the Page indexing report. Google says validation "typically takes up to about two weeks" (Search Console Help). Choosing between 404 and 410 is covered in fixing broken links.

Submitted URL seems to be a Soft 404

That's how Search Console's older Coverage report described a soft 404 on a URL listed in your sitemap. The Page indexing report that replaced it shows the reason as just Soft 404. To see only the URLs from your sitemaps, change the filter above the chart from All known pages to All submitted pages.

A submitted soft 404 is a contradiction: your sitemap asks Google to index the page, and the page looks missing or empty. Decide which side is true.

  • The page is real. Fix what Google saw. Run a live test in URL Inspection, click View tested page, and check the screenshot and the HTML. Content loaded by a script that failed, a "no results" state, or a page that's mostly template are the usual causes.
  • The page shouldn't be there. Take it out of the sitemap and return 404 or 410, or 301 it to a replacement. Sitemap plugins often list empty tag, category and author archives; most have a setting to leave those out.

Cleaning up soft 404s at scale

When the report lists hundreds or thousands of soft 404s, they nearly always come from a handful of templates or rules. Fix them by pattern, not one URL at a time.

  1. Open the Soft 404 reason in the Page indexing report and export the examples (the export button is at the top right). The list is a sample, so it may not hold every affected URL.
  2. Group the URLs by pattern: /tag/, /search?, /product/, ?page=. In Google Sheets, =REGEXEXTRACT(A2, "^https?://[^/]+/[^/?]+") pulls out the first folder of each URL to sort on.
  3. Open two or three URLs from each group and pick the fix from the table above. One change to the template, the route or the redirect rule covers the whole group.
  4. After the fix, paste a sample from each group into the bulk URL checker. Removed pages should answer 404 or 410, moved ones should show 301 → 200, and none should come back as a soft 404.
  5. Click Validate fix in Search Console, as above.

Plenty of outbound links die as soft 404s instead of clean 404s. A site reorganises and every old article URL redirects to its home page. A shop removes a product and the URL shows "this item is no longer available" with a 200. A domain lapses and a parking page answers 200 for every path (links to expired domains). A checker that only reads status codes marks all of these as working, and Search Console won't mention them because they're on someone else's site.

It isn't a rare case. In a 2004 study of web decay (what's now usually called link rot), Bar-Yossef and colleagues found that soft 404s "account for more than 15% of the dead links" they measured (Bar-Yossef et al., WWW 2004). The study is old, so read that figure as a sign the pattern is common rather than as a current rate.

Signs an outbound link has gone soft:

  • It used to point at a specific article or product and now ends on the site's home page or a top-level section.
  • The final page's title or main heading says "not found", "404", "page doesn't exist" or "no longer available".
  • The final URL contains /404, /not-found or /search?.

When you find one, swap in the page's new address if it moved, use an archived copy if you were citing it, or remove the link. Fixing broken links covers each option, and finding broken links covers running the crawl. Linkbruno's crawl reports a 200 that shows a not-found page as a soft 404, and flags parked domains separately. To test a handful of URLs right now, the free redirect checker shows every hop and flags redirects that end on a home page.

How soft 404 detection works, and where it goes wrong

No tool can be certain that a 200 page is really a missing page, Google included, so every detector guesses. The common heuristics:

  1. Text matching. Look for "not found", "404" or "no longer available" in the title and main heading. It's cheap and often right. It misfires on pages about errors (this guide would trip it), on product names and song titles that contain those words, and it misses error pages in other languages or with custom wording like "Oops, nothing here".
  2. Comparison with a known-missing URL. Request a made-up address in the same directory and compare it with the real one: the same final URL after redirects, the same number of hops, or near-identical content means the server answers every unknown address that way. Bar-Yossef's team used 25 random letters in the same directory, because different directories on one host are often served by different software. It fails on sites that return a real 404 for random paths but keep a "product unavailable" shell for removed items, and the paper notes it can't decide for the home page of a parked domain.
  3. Redirect target. A deep link that now redirects to the site root is suspicious. It's wrong when the old page really was folded into the home page.
  4. Size. A page far shorter than the site's other pages. Thin but real articles trip it too.

Good detectors combine several of these and still produce false positives, so treat a soft 404 flag as "open it and look" before you delete a link. Your own pages are the easy half: when they send a real 404 or 410, Google and every link checker read them correctly without guessing.

See what’s broken on your site

The free scan reads up to 25 pages and checks every link on them. No signup.

Related