Back to articles

Soft 404 in Search Console: What Triggers It and How to Fix It

SEOReport Team·
search-consolesoft-404page-indexinghttp-status-codesjavascript-seotechnical-seo

A soft 404 is a URL that answers 200 but reads to Google like an error or empty page. The 6 patterns behind it, how to confirm each, and the fix for each.

"Soft 404" in Search Console means your server answered a URL with a success status, 200, but Google read the page as an error page or an empty one and kept it out of Search. It is worth fixing in both directions: when the page should rank, its content or rendering needs repair; when the page is really gone, the server should say so with a 404 or 410.

What Google says "Soft 404" means

Google's Page Indexing report documentation (checked October 2026) describes the status as a page that returns "a user-friendly 'not found' message but not a 404 HTTP response code." Google's crawling troubleshooting documentation widens the definition: a soft 404 can also be "a page with no main content or empty page", and Google states that such pages "are excluded from Search."

No server sends a "soft 404"; the label is Google's judgment. Google classifies the page from its content as rendered. The older Index Coverage report showed sitemap URLs with this problem as "Submitted URL seems to be a Soft 404"; the Page Indexing report lists them all under "Soft 404". The troubleshooting documentation says an otherwise good page flagged this way probably "didn't load properly for Googlebot", was missing critical resources, or showed a prominent error message during rendering.

"Not found (404)" is the status for a URL where the server itself returned 404. A removed page there is correct. The same removed page in "Soft 404" means your server still claims the page exists.

What triggers a soft 404

1. An error template served with a 200 status. The site shows "page not found", but the server sends 200. Typical sources are catch-all application routes, a theme that renders its own error view, and server settings that hide the real code:

  • In Apache, ErrorDocument 404 pointing to a full URL (starting with http) sends a redirect instead of the 404; Apache's documentation warns the client then "will not receive the original error status code." The error page it redirects to usually answers 200.
  • In nginx, error_page 404 =200 /404.html; changes the code to 200, and error_page 404 = /404.php; returns whatever code the script sends.
  • A single-page app behind a fallback such as try_files $uri /index.html; answers 200 for every path, then draws its "not found" screen in the browser.

2. Empty listing pages. A category with no products, a search URL with 0 results, a filter that matches nothing, an empty tag archive: pages with no main content.

3. Out-of-stock or expired items stripped to one line. A product, job or property page that drops its description, photos and details when the item ends, leaving only "This item is no longer available".

4. Placeholder and thin pages. "Coming soon" pages, empty user profiles, location pages waiting for content, a template published before anyone filled it.

5. JavaScript that fails or shows an error state when Google renders it. The page works in your browser, but Google's rendered version is blank, half-built or shows an error. Google's troubleshooting documentation lists why resources fail to load: blocked by robots.txt, too many resources, server errors, and slow or very large resources. A data request that fails during Google's render has the same effect.

6. Removed URLs redirected to an unrelated page. Google's site-move documentation (checked October 2026) warns against redirecting many old URLs to one irrelevant destination such as the homepage, which "might be treated as a soft 404 error." The report lists the old URL, not the redirect target.

Pattern 5 is more likely when the first response carries little of the page. In 34 external websites where SEOReport compared the homepage's raw HTML with the rendered page between May and October 2026, 4 (12%) served raw HTML containing 2% or less of the words a visitor sees after rendering. On pages like those, any script or resource that fails during Google's render leaves close to an empty page behind.

Share of rendered words already in raw HTML, 34 sites, May–Oct 2026

Is a soft 404 a problem? Fix the status code or fix the content

Google's troubleshooting documentation frames the choice in 3 cases: the content is gone, it moved, or it still exists.

A page that no longer exists and has no replacementFix the status: return 404 or 410
A page whose content now lives at another URLFix the redirect: 301 to that exact replacement
A page you want indexed, flagged by mistakeFix the content or the render
An empty listing or search page you never wanted indexedReturn 404, or add noindex and stop linking to it
A product that is out of stock for nowKeep the page, with full content, marked out of stock
Old URLs redirected to the homepage in bulkRedirect each to its equivalent, or return 404 or 410
noyesnoyesnoyesnoyesURL listed as Soft 404Does the content still exist?Is there an exactreplacement?301 to the replacementReturn 404 or 410Live test screenshot showsthe full page?Fix blocked resources,failed API calls or the errorstateIs the main content morethan a line or an empty list?Add real content, ornoindex or 404 the pageValidate fix and monitor
Diagram source
flowchart TD
    A["URL listed as Soft 404"] --> B{"Does the content still exist?"}
    B -->|no| C{"Is there an exact replacement?"}
    C -->|yes| D["301 to the replacement"]
    C -->|no| E["Return 404 or 410"]
    B -->|yes| F{"Live test screenshot shows the full page?"}
    F -->|no| G["Fix blocked resources, failed API calls or the error state"]
    F -->|yes| H{"Is the main content more than a line or an empty list?"}
    H -->|no| I["Add real content, or noindex or 404 the page"]
    H -->|yes| J["Validate fix and monitor"]

How to confirm which cause applies

Group the examples by template first. Soft 404s usually come from a template or a server rule, so a long list often reduces to a few causes. Sort the examples by path (/search/, /category/, /products/) and test one URL from each group; the report lists at most 1,000 example URLs anyway. Working through flagged URLs one at a time, by hand or in an AI chat, fixes URLs and misses the template, the limit described in why one AI session cannot audit a whole website.

Check the status code your server really sends from a terminal:

bash
curl -sIL https://www.example.com/products/blue-widget | grep -iE '^(HTTP/|location)'

A redirect followed by 200 on an unrelated URL points to pattern 6 or an ErrorDocument with a full URL. A single 200 on a page reading "not found" is pattern 1.

Request a URL that cannot exist. This shows whether your server can return 404 at all:

bash
curl -s -o /dev/null -w '%{http_code}\n' https://www.example.com/this-page-does-not-exist-7Q4

A 200, or a redirect to your homepage, makes every mistyped link and removed page on your site a candidate for this report.

Use URL Inspection's live test for rendering. Inspect a flagged URL, click "Test live URL", then "View tested page". Google's URL Inspection documentation says the rendered screenshot is available only in a live test. The same panel shows the JavaScript console output and the page resources, with the resource list under "More info". A blank or half-drawn screenshot of a page that looks complete in your browser is pattern 5; the JavaScript rendering guide covers reading raw HTML against the rendered page.

Scan your sitemap for error wording on a 200. This loop prints sitemap URLs that answer 200 but contain common error phrases in their raw HTML:

bash
curl -s https://www.example.com/sitemap.xml \
| grep -o '<loc>[^<]*</loc>' | sed -e 's/<loc>//' -e 's/<\/loc>//' \
| while read -r url; do
code=$(curl -s -o /tmp/page.html -w '%{http_code}' "$url")
if [ "$code" = "200" ] && grep -qiE 'page not found|no longer available|no results found|0 results' /tmp/page.html; then
echo "check: $url"
fi
done

Adjust the phrases to your own error copy and languages, and if your sitemap is an index of other sitemaps, run it on each one. It reads raw HTML only, so an error drawn by JavaScript still needs the live test.

How to fix each cause

Error template with a 200 status. Send 404 or 410 for missing content and keep your helpful error page as the body. Google's documentation recommends exactly that: a 404 page with your normal navigation and links to popular pages, served with a real 404 status:

text
# Apache: a local path keeps the 404 status
ErrorDocument 404 /not-found.html
# nginx: serve the error page without changing the code
error_page 404 /404.html;

For a single-page app whose server cannot know a route is invalid, Google's JavaScript SEO documentation gives 2 strategies: redirect with JavaScript to a URL that answers 404 (for example /not-found), or add <meta name="robots" content="noindex"> to the error view. The noindex route moves the URL to the status covered in URL marked 'noindex'.

Empty listings and internal search. A category that will refill should keep a real description and links to related categories, so it has main content while empty. A category that is gone should return 404 or 410. Internal search results rarely belong in Google's index: add noindex to the search template and stop linking to search URLs.

Out-of-stock and expired items. Google's guidance on pausing an online business (checked October 2026) says to keep a temporarily unavailable product page, mark it out of stock, and keep Product structured data in line with current availability. Keep the details visible. For an item gone for good, return 404 or 410, or 301 to a clear replacement.

Placeholder and thin pages. Publish the page when it has content, or keep it at 404 until then; if it must stay reachable, add noindex. Thin pages Google does not classify as soft 404s can appear under Crawled - currently not indexed instead; the content health guide covers finding them.

Rendering failures. Unblock CSS and JavaScript that robots.txt disallows, make the data request that fills the page reliable, and put the main content in the server's first response where you can.

Bulk redirects to the homepage. Map each old URL to its real equivalent, or return 404 or 410. Page with redirect covers checking the redirects your own sitemap and links still use.

How to verify the fix

  1. Rerun the tests above. The made-up URL returns 404, the sitemap loop prints nothing, and the live test screenshot shows the full page.
  2. Click "Validate fix" on the Soft 404 issue in the Page Indexing report once every group is fixed. Google's documentation says validation typically takes up to about 2 weeks and in some cases much longer.
  3. Expect removed URLs to move, not disappear. URLs that now return 404 or 410 leave "Soft 404" and are reported under a not-found or other 4xx status, the correct end state for a page that is gone. Pages you kept should show as indexed in URL Inspection after Google recrawls them.

A full SEOReport diagnosis flags pages your sitemap lists that answer 200 with almost nothing to read, a homepage that answers 200 but reads like an error page, and a homepage whose content only appears after JavaScript runs. It does not check interior pages for "not found" wording or follow where removed URLs redirect, so the status and sitemap tests above remain the way to catch those.

Common questions

Does a soft 404 hurt the rest of my site?

The flagged URL is excluded from Search; it is not a manual action against the site. The wider cost is crawling: Google's crawl budget documentation says soft 404 pages "will continue to be crawled, and waste your budget," which matters most on large sites.

Should I return 404 or 410? Are 404s bad?

Either works, and neither is bad when the page is really gone: Google's Page Indexing documentation says 404s are not necessarily a problem if a page was removed without a replacement. Google's documentation on HTTP status codes says all 4xx errors except 429 are treated the same, and URLs returning them are dropped from the index. Use 410 to state that a removal is permanent; the result in Search is the same.

My page has real content but is still flagged. Why?

Google most likely rendered a different page from the one you see. Compare the live test screenshot with your browser; a blank area or an error message usually traces to a blocked resource, a failed data request or a slow, very large resource.

Get the complete diagnosis of your site

An evidence-backed report and a prioritized action plan, on a plan with monthly credits.