301 redirect, 404 or 410: what an old page should answer, with a worked case

301, 302, 307 and 308: permanent and temporary moves

When a page changes its address, the old one still gets visits from bookmarks, other sites and search results that have not caught up. A 301 redirect sends those visits on. RFC 9110, the IETF standard for HTTP, says that 301 Moved Permanently means the resource has a new permanent address that future references ought to use. The server names that address in a Location header, and the browser follows it without the visitor noticing.

Google's page on redirects sorts them into two families. Permanent ones, 301 and 308, put the new address in search results, because the indexing pipeline takes the redirect as a signal that the target should be canonical. Temporary ones, 302, 303 and 307, keep the source page in the results. For the permanent kind its advice is short: use it when you are sure the redirect will not be reverted.

The codes 301 and 308 differ in one detail. RFC 9110 notes that, for historical reasons, a browser may turn a POST request into a GET after a 301, and points to 308 when that is unwanted; 307 plays the same role next to 302. Google's documentation on HTTP status codes lists 308 as equivalent to 301, yet asks owners to use the code that fits, so that other clients, such as other search engines, can benefit.

Where an old address should lead, and in how many hops

A redirect is only as good as its target: the page that now does the old one's job. Google's site-move guide warns against sending many old URLs to one irrelevant destination such as the home page. That can confuse users and might be treated as a soft 404 error, Google's name for an address that reports success but shows no real content. Merged pages are the exception: their old addresses may all lead to the new combined page.

The second rule is one hop. Google's crawlers follow up to 10 redirect hops by default, but the same guide advises redirecting to the final destination directly, or at least keeping a chain ideally to no more than 3 hops and fewer than 5. Google recommends a server-side permanent redirect whenever possible and keeps JavaScript redirects for sites that cannot do anything else, since rendering may fail.

404 or 410 for a page that is gone for good

Some pages have no successor: a finished promotion, a discontinued service. For them the honest answer is an error code. RFC 9110 defines 404 Not Found as the server finding no current version of the resource, without saying whether the absence is temporary or permanent. 410 Gone adds that the condition is likely to be permanent. The standard calls it a tool of web maintenance, telling the recipient that the resource is intentionally unavailable.

For Google the difference is small. Search Console Help says it in one line: "Currently Google treats 410s (Gone) the same as 404s (Not found)." Google's page on status codes adds that all 4xx errors except 429 are handled alike: a previously indexed URL is removed from the index, and crawling of it gradually slows. Yandex Webmaster help names 404, 403 and 410 together and says such a page leaves its search database within a week after the robot finds the instruction.

So a 410 is not a ranking trick; its value is precision for other clients and for whoever maintains the site later. The standard even leaves it to the owner whether to mark removed resources as gone at all. The real mistake lies elsewhere. Search Console Help warns that answering a missing page with any code other than 404 or 410, or redirecting it to the home page, can be problematic, and calls such pages soft 404s.

Noindex and canonical links when the page stays online

Some pages should leave search but stay on the site: a filter page, an empty category, a print version. Google's page on noindex describes a rule placed in a meta tag or an HTTP header. Once Googlebot reads it, Google drops the page from its results, whether or not other sites link to it. The rule works only if robots.txt does not block the page, otherwise the crawler never sees it; Yandex Webmaster help gives the same warning for its robot.

Duplicates need yet another answer. Google's guide to canonical URLs calls both a redirect and a rel="canonical" link strong signals, and a sitemap entry a weak one. It advises a permanent redirect only when you retire a duplicate, and prefers rel="canonical" to noindex for choosing a canonical within one site, since noindex blocks the page from Search completely. For pages in several languages it asks for a canonical page in the same language.

The viton13.com case: 673 old addresses answering 404

These choices are easiest to see on a site with many retired pages. The case is viton13.com, VITON13 Studio's own site and not a client commission. The starting point is the owner's Google Search Console export of 21 August 2026, when Google counted 738 pages as indexed and 2,572 as not indexed. Among the reasons Google listed, 673 addresses answered 404, that is, not found.

The 673 are the part this story is about. The task was to give Google a clear answer for every address it still knew. Removed pages were to be marked as gone for good, and moved pages sent to their new address. Empty pages were to stay out of the index, and no internal link was to lead to a page that does not exist.

How the studio's site now answers retired and moved addresses

Retired addresses on viton13.com now answer 410 Gone. That covers 1,008 exact VJOURNAL URLs and the addresses under a pattern rule, such as removed auto-generated archive briefs and English slugs under /ru and /es. It also covers 27 addresses from the site before 2026, among them old .html pages and /tproduct/ links. Another 74 retired stories redirect in one step to their surviving version.

Permanent redirects sit in one map of 141 rules: old paths such as /oldmoneymen, which now leads to the Old Money men's collection in the store, renamed service sections in all four languages and the www address, which leads to the main domain. Every request first passes the site's proxy, a layer that runs before any page is built. It checks the path against exact lists of retired and moved addresses, then against the pattern rules. It returns either a single 308 redirect or a 410 page with a noindex, nofollow header.

Topic pages with no stories carry noindex in each language where they are empty, and their hreflang tags list only the languages where the page can be indexed. Before going live, the deploy crawls the built release and refuses to publish it if any internal link fails. On 19 September 2026 the 673 baseline addresses were re-checked one at a time: 604 now answer 410 Gone, 67 redirect with a 308 to a live page and 2 still answer 404.

What the 604, 67 and 2 split does not prove

These numbers describe what the site answers, not what Google did with it. There is no Search Console data from after the fixes yet, so the case claims no change in indexing, rankings, clicks or traffic. The split describes Google's old 404 list on 19 September 2026, not a before-and-after of one release. The Search Console figures also cover only the main viton13.com property.

The honest reading is narrower and still useful. Since Google currently handles 410 and 404 alike, swapping one for the other is no reason in itself to expect a change in search. What changed is the kind of answer: apart from the 2 that still answer 404, the addresses on that list now say "gone" or "moved". Even a remaining 404 is not automatically a fault, because Search Console Help calls it the right answer for a deleted page with no replacement or equivalent.

Preparing a redirect map before you order the work

Whether you do it yourself or order it, the work starts with a list. Google's site-move guide suggests the sources: sitemaps, server logs or analytics, the links report in Search Console and the export of the content management system. Add the Search Console pages report with the addresses Google already sees as not found. Put everything into one table: the old address and its answer, a new address or the word "gone". Then open a sample with a tool that shows the raw status code and Location header, and repeat that check after every release.

On a small site whose hosting panel or plugin writes the rules, a dozen 301 redirect rules can be an hour of careful work. It becomes a project when there are hundreds of retired addresses, several languages, pattern rules and a release check that has to live in code. That is the kind of work VITON13 Studio did on its own site, and the studio's technical SEO implementation service offers the same for other sites. With the table, Search Console access and a note of what changed and when, the scope can be estimated before a rule is written.