A page can look perfectly fine in a browser and still be excluded from Google. One of the most common reasons we find in technical audits is the soft 404: a URL that tells visitors “nothing here” while telling Googlebot “200 OK, everything is fine.” The two messages contradict each other, and Google resolves the contradiction by trusting the content over the status code. The page drops out of the index, and the Page Indexing report in Search Console files it under “Soft 404.”
This post explains what a soft 404 is in Google’s own terms, how each HTTP status code changes what Googlebot does next, and which response a missing, moved or temporarily broken page should actually send.
What a soft 404 is
A soft 404 is a URL that returns an HTTP 200 (success) status code while its content says the page does not exist, or while it has no meaningful main content at all. Google’s Search Central documentation describes it as a page telling the user it does not exist while returning a 200, and adds that a page with no main content, or an empty page, can be classified the same way. When Google’s systems decide from the content that the page is really an error page, Search Console reports it as a soft 404 in the Page Indexing report and the URL is excluded from Search. The status code says one thing, the content says another, and Google goes with the content.
How Googlebot reads each status code
Google publishes a reference covering how HTTP status codes affect Google’s crawlers (last updated 4 February 2026 when we checked it on 29 October 2026). The short version, by status class:
- 2xx (success): the content is passed on for indexing, but a 2xx does not guarantee indexing, and content that looks like an error or an empty page is reported as a soft 404.
- 3xx (redirect): Googlebot follows up to 10 redirect hops for general web content; a 301 or 308 is a strong signal that the target should be processed, a 302 or 307 a weak one.
- 4xx (client error): content is ignored, new URLs are not indexed, already-indexed URLs are removed over time, and crawl frequency for those URLs gradually decreases. Google treats every 4xx the same way, with one exception.
- 429 (too many requests): the exception. Google treats it as a server-overload signal, grouped with server errors rather than with other 4xx codes.
- 5xx (server error): Googlebot temporarily slows its crawl rate in proportion to the number of URLs failing. Indexed URLs are kept at first, but URLs that persistently return server errors are eventually dropped.
Two details in that reference catch people out. First, 404 and 410 are handled the same way for Google Search; the documentation gives a 410 no special “remove it faster” treatment. Second, Google states that 401 and 403 have no effect on crawl rate. Returning 403 to Googlebot to “slow it down” does not slow it down; it tells Google those URLs do not exist for it.
Why sites produce soft 404s
Google’s crawl troubleshooting documentation lists missing server-side includes, broken database connections, empty internal search results and JavaScript files that fail to load as typical causes. In client work, the patterns we see most often fall into a handful of groups:
- Template fallbacks: a CMS theme that renders “No posts found” on an empty category, tag or author archive and still returns 200.
- Catch-all routing: a single-page app or a misconfigured rewrite rule that serves the homepage shell, with a 200, for any URL, including ones that never existed.
- Retired inventory: discontinued product, listing or location pages kept live with a “no longer available” message and nothing else.
- Bulk redirects to the homepage: removed pages redirected en masse to the home page rather than to a true replacement, which Google can treat as soft 404s because the destination is not equivalent.
- Rendering failures: a real page that depends on an API call or script Googlebot cannot load, so the rendered version Google sees is blank or shows an error message.
The last group is the one that hurts most, because the page works for users. Google’s documentation says an otherwise good page flagged as a soft 404 most likely did not load properly for Googlebot, was missing critical resources, or displayed a prominent error message during rendering. Resources blocked by robots.txt, too many resources, server errors and slow or very large resources are all named as reasons.
Which status code each situation should send
Google’s guidance maps cleanly onto three situations. Decide which one each flagged URL is in, then send the code that matches.
The content is gone and there is no replacement. Return 404 or 410. Keep a helpful custom 404 page with your normal navigation and links to popular sections, but make sure the server sends the 404 status with it. A friendly error page that returns 200 is exactly the soft 404 you are trying to remove.
The content moved or has a clear replacement. Return a 301 to the specific replacement, not to the homepage or a generic category. If you are consolidating several old pages, each should point to the page that actually answers the same need. We covered how Google weighs 301, 302 and meta refresh redirects separately.
The content still exists and was flagged by mistake. This is a rendering problem, not a status-code problem. Inspect the URL, look at what Google rendered, and fix whatever stopped the main content from loading: a blocked script, a failing API call, a timeout, or an error overlay that dominates the rendered page.
A fourth situation is temporary: the server is overloaded or down for maintenance. Here the right answer is 503 (or 429), and the documented window matters. Google’s troubleshooting guide says Googlebot will retry those URLs for about 2 days, and that returning 503 or 429 for more than 2 days will cause Google to drop those URLs from the index. A maintenance mode left switched on over a long weekend is enough to cross that line.
Check one URL today
You can find out what a page really sends in under a minute. From any terminal, run:
curl -s -o /dev/null -w "%{http_code}\n" https://www.example.com/a-page-that-does-not-exist-12345
Replace the domain with yours and keep the nonsense path. If the result is 200, your site is serving a soft 404 for every mistyped or deleted URL, and the fix belongs in your server or CMS routing, not on individual pages. Repeat the test on an empty category or tag archive, an empty site-search results page, and a discontinued product or service page.
Then, in Search Console, open the Page Indexing report, select the “Soft 404” reason, and sample the listed URLs with the URL Inspection tool. Run a live test and choose “View tested page” to see the screenshot Google produced. The Page Indexing report help page recommends exactly this, and it is the fastest way to separate “this page really is empty” from “this page did not render for Googlebot.”
Mistakes that make it worse
- Noindexing the error page instead of fixing the code: a noindex on a 200 page hides the symptom but leaves every bad URL crawlable and consuming requests.
- Redirecting every 404 to the homepage: a plugin setting that does this converts honest 404s into soft 404s in bulk.
- Using 403 or 401 to manage crawling: Google says these codes do not change crawl rate; the URLs are simply treated as missing.
- Leaving 503 in place too long: the documented 2-day window is short, so a maintenance page should come down as soon as the work is done.
- Treating 404s as a crisis: Search Console’s own help says 404s are not necessarily a problem when a page was removed with no replacement. Google will keep retrying those URLs less and less often, and that is fine.
Where this fits in a technical SEO review
Soft 404s are rarely one bad page. They are usually a rule: a template, a routing fallback or a redirect plugin applied across hundreds of URLs. That is why the fix usually lives with whoever controls the server and the theme, and why it belongs in the build stage of any website design or rebuild, where the 404 template, archive behavior and redirect map can be specified before launch rather than discovered afterward. In our SEO engagements, the status-code check above is one of the first things we run, because it takes a minute and tells you whether the site is giving Google honest answers.
The principle is simple: the status code should say the same thing as the page. When the page is gone, say 404. When it moved, say 301 to the real replacement. When it is down for a few hours, say 503 and bring it back well inside the 2-day window. When a live page is flagged anyway, look at what Google rendered, not at what your browser shows.

