Skip to content
search-engine-optimizedby Netsolution

Error pages and the soft 404

A 404 and a 410 are handled the same way: the crawler reports that the content does not exist and the address leaves the index. A 5xx error first preserves the address, then drops it. A soft 404 is a page answering 200 while showing an error, and it is the only genuinely ambiguous case.

What a response code says, and to whom

A page that no longer exists raises a question of protocol rather than of writing: the server answers with a number, and that number is the only thing the crawler reads before deciding. The published documentation lists the twenty most frequent codes and says, for each, what happens next.

The 4xx codes, except 429, are treated the same way: the crawler informs the next system that the content does not exist, the address is removed from the index if it was there, and a newly encountered 404 is not processed.

officialSource : Google Search Central, “HTTP status codes and network errors”

That sentence settles an argument that has run for years. The 404 and the 410 sit on the same line of the table, alongside 401, 403 and 411, and they get the same handling. No documentation promises a faster disappearance for one than for the other.

Choosing 410 over 404 has no documented consequence for indexing, since the published documentation describes them on the same line; it keeps a declarative value for the people who read server logs, not for the engine.

hypothesis

The status says what it is: a reading. The document does not compare the two codes explicitly, it handles them together, and we conclude there is nothing to gain by preferring one. What stays true lies elsewhere: a 410 declares a deliberate removal, and that declaration serves the people who will read the server logs six months from now.

The soft 404, the only genuinely ambiguous case

A 2xx response whose content suggests an error, an empty page or an error message causes a soft 404 error to be reported in Search Console.

officialSource : Google Search Central, “HTTP status codes and network errors”

This is the one situation in this guide where the site contradicts itself: the header announces success, the page shows an error. The usual causes all look alike — a template that always returns 200, a discontinued product routed to an empty page, an internal search filter that finds nothing and says so politely inside an otherwise normal page.

Your 404s are burning crawl budget and diluting your authority. Our fix redirects every removed page to the homepage.

The documentation writes that 4xx codes other than 429 have no effect on the crawl rate. And redirecting an address with no equivalent to the homepage produces exactly a soft 404: the page answers 200 without holding what was asked for.

Server errors, which are a different matter

5xx errors and the 429 code prompt crawlers to slow down; already indexed addresses are preserved at first and eventually dropped, and any content received from an address answering 5xx is ignored.

officialSource : Google Search Central, “HTTP status codes and network errors”

The distinction is worth holding on to during an incident. A 4xx error says the page does not exist; a 5xx error says the site cannot answer right now. In the second case the addresses stay in the index at first, which buys time to repair, but an outage that drags on ends in the same result as a deletion.

The crawl frequency of a 404 address gradually decreases, but 4xx codes other than 429 have no effect on the site's crawl rate, and the vendor explicitly advises against using 401 and 403 to limit it.

officialSource : Google Search Central, “HTTP status codes and network errors”

The advice attached to that sentence bears repeating because it circulates backwards: answering 401 or 403 to slow a crawler down slows nothing. Those codes say the content does not exist for the requester, and the only code described as an overload signal is 429.

The error page is a page

A custom error page is rated like any other page: the rater guidelines place it at the medium rating in general, and allow high ratings for one showing real effort and original content.

officialSource : Google, “General Guidelines”, version of 11 September 2025, section 9.2

This is the least known part of the subject, and it makes for cheerful reading. The document lays out a full scale for error pages: the medium rating for one that explains the problem and no more, a high rating for one that genuinely helps, and it goes all the way to the highest rating for a 404 page whose original content is a cartoon, a caption and a search box fitted to the site.

A broken page is generally rated low, but lowest where the fault is not temporary; the check offered is to open other pages on the same site, a widespread fault revealing a systemic lack of effort.

officialSource : Google, “General Guidelines”, version of 11 September 2025, section 9.2

The check offered is one you can run on your own site, and it beats a tool: open five pages at random. If the same fault shows up everywhere, it is not an incident, it is a characteristic.

Fixing soft 404s and making the error page useful

hypothesis

Effort
low
Time to effect
weeks
Risk
none
Durability
structural

The effort is the lowest on this site: a response code and a template. The delay follows the next crawl of the addresses concerned. The risk is nil, with one reservation worth naming — answering 404 on a page that really exists makes it disappear, and that is a common configuration mistake. Durability is excellent, because a status code does not go out of fashion.

What does not work, or no longer works

Bulk-redirecting removed addresses to the homepage, first. It preserves nothing and manufactures soft 404s, replacing a clear case with an ambiguous one.

Then treating the coverage report as a to-do list. An address appears there in 404 because it is a 404, which is sometimes exactly what was intended.

Finally, leaving the server’s default error page in place. Replacing it costs nothing, and the rater guidelines spell out what an error page that helps can earn.

Sources

  1. How HTTP status codes, and network and DNS errors, affect Google Search — Google Search Centralaccessed on
  2. General Guidelines, version of 11 September 2025, 182 pages — Googleaccessed on

Block revalidated on

Frequently asked questions

Should I answer 410 rather than 404 to remove a page faster?
The documentation describes both codes on the same line, with the same handling: the content is declared non-existent and the address leaves the index. No different delay is published. A 410 keeps its meaning for whoever reads the server logs, because it states an intention, but it promises the engine nothing more.
What exactly is a soft 404?
An address that answers 200, meaning success, while displaying an empty page or an error message. The engine sees the contradiction between code and content, and Search Console reports it. It is the only case in this guide where the site says one thing and shows another, which is why it gets flagged.
Does a 404 page consume crawl budget?
The visit frequency of a 404 address gradually decreases, and the documentation writes that 4xx codes other than 429 have no effect on the site's crawl rate. Cleaning up broken internal links remains useful for readers, but it is not a crawl budget operation.
What happens during a server outage?
5xx errors slow crawling down. Addresses already indexed are preserved at first, then eventually dropped if the outage lasts, and the content received is ignored. The crawl rate climbs back gradually once the server answers normally again.
Is it worth polishing the 404 page for search?
For the reader first, and the rater guidelines confirm it: the error page is rated like the others, the medium rating going to one that explains the problem, high ratings to one that genuinely helps. A search box and a few useful links are enough to change category.