301 redirect

A 301 redirect is an instruction from your server saying that the page somebody asked for now lives at a different address, permanently. The browser follows it without the visitor noticing anything beyond the address changing, and search engines take it as an instruction to move whatever the old address had earned to the new one.

In short

• A permanent forward from an old address to a new one.
• No ranking value is lost in the hop, despite what you may have read.
• Redirecting to something unrelated wastes it entirely.
• Once created, leave it in place indefinitely.

When you need one

Addresses change more often than anybody plans for, and every change breaks every link that pointed at the old one.

• A site rebuild that reorganises the address structure.
• Moving to https, which needs an SSL certificate in place first, so the old insecure addresses lead to the secure ones.
• Choosing between www and the bare domain, so only one is live.
• Renaming a page or a service, where the subject survives under a new name.
• Merging two overlapping pages into one, which is the repair described under keyword cannibalisation.
• Retiring a product that has a direct successor.

In each case the alternative is a broken link. Somebody clicks something on another site, or in an email from two years ago, and lands on an error.

The tax that no longer exists

For years the accepted wisdom was that a redirect leaked a portion of a page’s accumulated value, often quoted at around fifteen per cent, and that chaining two together compounded the loss. This produced a lot of nervous architecture and a lot of consultants advising against changes that were otherwise sensible.

Gary Illyes of Google said plainly in 2016 that no PageRank is lost through a 30x redirect of any kind, and John Mueller confirmed it had already been true for some time. The figure was real once and has not been true for well over a decade.

Minimising redirects is still worth doing, for two reasons that have nothing to do with ranking. Each hop is a round trip the visitor waits through. And a site with redirect rules layered over four rebuilds becomes something nobody can reason about, which is how a rule that quietly breaks a section survives for months.

Permanent, temporary, and why it matters

The two codes say different things and are not interchangeable. A 301 tells a search engine that the new address replaces the old one and should be indexed instead. A 302 says the move is temporary and the original should be kept. Choosing 302 for a permanent move is a common accident, often because a platform defaults to it. In the short run Google keeps the retired address in its index and treats the new one as a stand-in. Left long enough it will conclude the move was permanent after all, so the mistake corrects itself slowly, which is not a reason to make it.

There are two newer codes and they are not a pair. A 308 is the permanent one, matching 301. A 307 is the temporary one, matching 302. Both preserve the details of the original request in a way the older codes did not, which matters to developers and rarely to anybody else. So a 308 offered where you expected a 301 is fine. A 307 in the same place is not.

Where they go wrong

Two failures account for most of the damage, and neither announces itself.

The first is redirecting everything to the homepage. It feels tidy and it defeats the purpose. Google recognises a redirect that bears no relation to what was requested and treats it as a page that no longer exists, so nothing carries over. The visitor, meanwhile, asked for a specific thing and got your front door with no explanation. If there is no equivalent page, a proper not found is the better answer.

The second is the chain. Address A points to B, B was later pointed at C, and nobody went back to update A. Googlebot will follow up to ten hops in one attempt before recording an error and coming back later, so a short chain is survivable and a long one is not, and every hop costs the visitor time regardless. Every time you add a redirect, check whether anything already points at the address you are moving, and repoint it at the final destination.

The migration checklist

Most redirect damage happens during a site rebuild, in the week nobody was watching. Before the switch, export every address currently getting search traffic and every address with links pointing at it. Map each one to its closest equivalent on the new site. Anything without an equivalent needs a decision made deliberately rather than by default.

After the switch, test the list rather than assuming. Load a sample of old addresses and confirm each lands where the map said it would, in one hop. Then regenerate the XML sitemap so it lists the new addresses rather than the old, then watch the not found report in Search Console for a month, because the addresses you forgot will announce themselves there and nowhere else.

Words you will hear

• 301. Permanent. The one you want almost always.
• 302, 307. Temporary. The original address stays the indexed one.
• 308. A permanent redirect that preserves the request method. Equivalent to a 301 for your purposes.
• 404. Not found. Appropriate when a page genuinely has no successor.
• 410. Gone, and deliberately so. Google processes it slightly faster than a 404 and the practical difference is small.
• Soft 404. A page that says not found while technically reporting success, or a redirect so unrelated that Google treats it as one.
• Redirect map. The spreadsheet pairing every old address with its replacement. The single most useful artefact of any migration.

Why they are never worth deleting

A redirect is the only mechanism that lets you change your mind about an address without punishing everybody who wrote the old one down. That includes other websites, old newsletters, printed cards and people’s bookmarks, none of which you control and none of which will be updated.

Which is why they are cheap to create, expensive to forget, and never worth deleting.