Relaunching a Website Without Ranking Loss

That takes time. Google writes that rankings can fluctuate after any major change to a website while Google crawls the pages again. For mid-sized websites, it can take several weeks or longer until Google shows the new addresses instead of the old ones, and longer still for larger ones. Over time, the rankings settle.

So fluctuations are to be expected. What you can influence is how clearly you show Google the path from the old address to the new one. That is what this article is about: inventory, mapping, test, launch day and monitoring.

The graphic shows the same old address with and without a redirect.

An inventory comes before the first draft. For the redirects in a website relaunch, you need a list of all old addresses. Google names four places to find them:

• the sitemap of your website, the page list for search engines
• Analytics, the visitor statistics, or the server logs for the most visited addresses
• the links to your website that Search Console shows
• the page list in your content management system

Images, videos, PDFs and downloads have addresses too and belong on the list. Which pages matter most for visibility is something you already clarify in the design briefing.

Two more things belong before launch. The website should be verified in Search Console, each time with all variants: with and without www, with http and with https. If the domain changes, this applies to the old and the new domain. The chosen verification method must also keep working on the new website.

Timing deserves thought too. If possible, Google advises moving during a period with fewer visitors. It makes sense to change only one thing at a time, for example the domain first, then the design. A relaunch usually changes plenty anyway, so it is worth not switching the domain as well.

Now every old address gets a target or a deliberate no. Google advises mapping the addresses of the old website to those of the new one, page by page. A simple redirect list is enough for this: old address, new address and a status. Status 301 stands for a permanent redirect, 410 for a page that is dropped.

Three rules help with the mapping:

• One matching target per page. Sending many old addresses to a single unsuitable target such as the homepage can confuse visitors and can be treated as a “soft 404”, meaning like a page that was not found. Merged content may point to its shared new page.
• A deliberate no where there is no successor. Pages that are dropped return “not found” (404) or “gone” (410). In the list, status 410 is enough.
• No detours. If A leads to B and B to C, A is better off redirecting straight to C. Google follows up to ten redirects in a row but advises going straight to the target.

For a relaunch with new addresses, permanent redirects are the right choice, meaning status code 301 or 308. The server sends the code along, and it decides what Google makes of it:

Google recommends setting them up on the server, where they have the highest chance of being understood correctly. They can be set up with your hosting provider, with the person who maintains the server, or via the redirect function of your content management system. A redirect via JavaScript is only a last resort.

Until launch day, the new website usually sits at a test address. To keep it out of search too early, Google names access restricted to certain IP addresses and the “noindex” rule as examples. With “noindex”, Google leaves a page out of search. What matters here is the robots.txt, the file in which a website tells search engines which areas not to crawl: “noindex” only works if the page is not blocked there as well. Otherwise Google never sees the rule, and the page can still appear in search, for instance when other pages link to it.

At launch, these blocks must come off, along with a robots.txt that excludes everything, if one was set. Google counts forgotten blocks among the common mistakes in site moves; they can keep the new website from being fully included in search. So it is worth preparing the robots.txt for launch and the list of pages with “noindex” in advance.

On the test address, a few more things are checked:

• Each page names its future address as the authoritative one with a “canonical” tag.
• On multilingual websites, the language tags (“hreflang”) point to the new addresses.
• Internal links lead straight to new addresses and not via a redirect first.
• For the time after launch: no firewall or overload protection blocks Googlebot, the program Google uses to crawl websites. Google points this out for hosting changes.

On launch day, these points are due, following Google's guide for moves with new addresses:

1. Redirects from the redirect list switched on at the server
2. Test-phase blocks removed: the access restriction, “noindex” and a robots.txt that excludes everything
3. Redirects checked, one at a time in the URL Inspection tool of Search Console, many at once by script
4. New sitemap submitted in Search Console
5. Internal links, links in profiles and the targets of running ads pointed to the new addresses

It is also worth trying to contact the operators of websites that link to you, starting with those whose link brings the most visitors. Google recommends that small and mid-sized websites move all addresses at the same time. For large ones, moving in sections is possible.

Several weeks are to be expected, longer for larger websites. These places show whether the move is working:

• Indexing in Search Console: The indexed pages of the old website go down, those of the new one go up. Error spikes show up here too.
• Sitemaps in Search Console: The new sitemap should be submitted. The old one can be removed or left submitted for a while for observation; Google mentions both. For the old one, the number of indexed pages falls, for the new one it rises. Warnings about redirects in the old sitemap are normal.
• Search queries in Search Console: Impressions and clicks of the new addresses appear here.
• Analytics and server logs: Traffic to the old addresses falls, traffic to the new ones rises.

If Search Console shows many “not found” reports, the redirect list deserves a look. A redirect may point to an address that does not exist, or a redirect may be missing. Reports about deliberately dropped addresses (404 or 410) are to be expected. If the domain changes, old addresses can also still appear in search from time to time. That counts as normal.

After the move, Google temporarily crawls the new website more heavily than usual. For very large websites, the hosting provider is informed before launch so that the server can handle the extra requests.

Redirects should stay in place as long as possible, usually for at least a year. That lets Google transfer all signals to the new address, including the links other websites have to yours. For visitors, it can make sense to keep them indefinitely.