Development

Website migration checklist: move a site safely

Two groups of coloured tiles joined by a short bridge of three orange tiles

Migrations go wrong when somebody skips a step. The checklist we work through, in order, so you keep your rankings and your enquiries.

Most of the horror stories about website migrations come from the same cause: somebody changed the site and the addresses at the same time, without a list. Traffic falls off a cliff, nobody can say exactly what changed, and unpicking it takes longer than the build did.

A migration is not difficult. It is just unforgiving of skipped steps. This is the checklist we work through, in order.

First, know which kind of migration you are doing

People use one word for five different jobs, and the risk is wildly different between them.

TypeWhat changesRisk to rankings
New host, same siteWhere the files sitVery low
New design, same URLsHow it looksLow to moderate
New platformWhat runs the site, often the URLs tooModerate to high
New URL structureEvery address on the siteHigh. This is where sites get hurt
New domainThe address itselfHighest

Change one at a time where you can. Redesign and keep the URLs, launch, confirm it is stable, then change addresses if you still need to. Doing both at once means that if traffic drops you cannot tell which one caused it. We cover the rankings side of that in will a redesign hurt your Google rankings?

Before: the preparation that decides the outcome

1. Record where you are now

You cannot tell whether a migration went well without a before. Capture it while the old site is still up.

  • Export every URL on the current site.
  • Note current rankings for the terms you care about.
  • Screenshot Search Console: impressions, clicks and average position for the last three months.
  • Record monthly traffic and, more importantly, monthly enquiries.
  • List your top twenty pages by traffic. These are the ones that must not break.

2. Take a full backup, and test it

Not the automatic one you assume exists. Take a real backup of files and database, put it somewhere off the server, and confirm you could restore it. Our guide to backing up a WordPress site covers doing that properly.

3. Write the redirect map

This is the single most important artefact in a migration. Every old URL needs to point at its closest new equivalent, as a permanent 301 redirect.

  • Old URL to new URL, one row each, in a spreadsheet.
  • Map to the closest genuine match, not the homepage. Redirecting everything to the homepage is treated as a soft 404 and wastes the value of the old page.
  • Include pages you are retiring. Something has to happen to them.
  • Avoid chains: A to B to C. Point A straight at C.
  • Keep it. You will need it when something looks wrong in six weeks.

4. Check the technical carry-over

  • Titles and meta descriptions, page by page. These get lost in rebuilds constantly.
  • Heading structure. One H1 per page saying what it is.
  • Image alt text, which almost never survives a platform change.
  • Structured data, if you have it.
  • Canonical tags pointing where they should.
  • Your robots.txt. Staging sites block crawlers, and that block must not travel to live.

5. Sort the domain and hosting separately

Do not move the domain the same day you launch the site. See how to transfer a domain name, and while you are in the registrar, check the renewal date and whose name it is in.

Launch day

  1. Go live at a quiet time. Early morning midweek. Not Friday afternoon, and not the week of a campaign.
  2. Remove the noindex. The single most common migration disaster is a staging site going live with crawlers still blocked. Check it first, before anything else.
  3. Turn the redirects on and test them. Spot check your top twenty pages by hand.
  4. Check the security certificate. A migration is a classic moment for an SSL problem to appear.
  5. Submit the new sitemap in Search Console.
  6. Test every form. From a phone, on mobile data, and confirm the email actually arrives.
  7. Check analytics and tracking are firing on the new pages.
  8. Click your own menu and footer end to end.

The week after

  • Watch Search Console coverage daily for crawl errors and not found pages.
  • Every 404 in the logs is a redirect you missed. Add it and move on.
  • Check the site speed. New builds are often heavier than the thing they replaced, and images are usually why.
  • Expect some ranking movement. A dip in the first two to four weeks is normal while Google recrawls.
  • Do not panic and change things. Give it a month before drawing conclusions, unless something is plainly broken.

What actually goes wrong

In rough order of how often we see it.

MistakeWhat it costs
Noindex left on at launchTotal disappearance from search until someone notices
No redirects, or all to the homepageEvery ranking page starts from nothing
Titles and descriptions not carried overRankings and click-through both drop
Forms not testedEnquiries silently stop, sometimes for weeks
Internal links still pointing at old URLsEvery click goes through a redirect it should not need
Changing design and URLs togetherYou cannot diagnose the drop
No before dataYou cannot prove it went well or badly

One thing worth knowing about DNS

You will be told to allow up to 24 hours for changes to take effect. That figure is mostly cover. In practice new visitors usually see the new site within minutes.

What confuses people is that the owner is often the last to see it, because their own phone and browser still hold the old address in memory. If the site looks unchanged to you an hour later, check it on mobile data or on somebody else’s phone before assuming something has failed.

The condensed checklist

Before: record current URLs, rankings, traffic and enquiries. Full tested backup. Redirect map written. Titles, descriptions, alt text and structured data carried across. Domain and hosting handled separately.

Launch: quiet time. Remove noindex. Redirects on and spot checked. Certificate valid. Sitemap submitted. Forms tested from a phone. Analytics firing.

After: watch coverage daily. Fix every 404. Check speed. Expect a dip. Wait a month before concluding anything.

How we put this together

This is the order we actually work in, and the warnings are things we have either done ourselves or been called in to unpick afterwards.

The DNS section is the one we would underline. It comes from watching owners panic on launch day because their own phone still showed the old site while everyone else could already see the new one. Nothing had failed. Their browser was holding the old address in memory.

The point about redirects being the whole job comes from taking over sites where nobody wrote a redirect map, and the point about staging blocks travelling to live comes from it being the single most common launch fault we see, including on sites built by other people.

None of it is clever. It is just a list, and migrations go wrong when there isn’t one.

If you are moving a site and would rather not carry the risk yourself, it is the kind of job we do regularly, including moving shops between platforms with the addresses and rankings kept intact. Tell us what you are moving and we will tell you what is involved. Worth running the website audit checklist over the old site first as well, because a migration is a good moment to stop carrying problems across.

Paul Fletcher

Written by

Paul Fletcher

Web Developer

Paul builds fast, reliable websites and online shops, and makes sure they work well on every phone and laptop.

More guides by Paul Fletcher →

FAQs

An ordered list of what to record, carry across and test when a website changes host, design, platform, URL structure or domain. The essentials are a full tested backup, a redirect map from every old URL to its closest new equivalent, carrying titles, descriptions and alt text across, removing the staging noindex at launch, and watching Search Console daily afterwards.

It depends entirely on which kind of migration it is. Moving host with the same site is very low risk. Changing URL structure or domain is high risk and is where sites get hurt. A dip in the first two to four weeks is normal while Google recrawls. Change one thing at a time where you can, because if you redesign and change addresses together you cannot tell which caused a drop.

Leaving the noindex tag on at launch. Staging sites block search engines, and if that block travels to live the site disappears from search entirely until somebody notices. Check it before anything else on launch day. The next most common are missing redirects, or redirecting every old page to the homepage, which Google treats as a soft 404.

Yes, if any URL changes. Every old address needs a permanent 301 redirect to its closest genuine equivalent, not to the homepage. Avoid chains where A points to B which points to C, and keep the redirect map afterwards, because you will need it when something looks wrong weeks later.

You will be told up to 24 hours, but that figure is mostly cover. New visitors usually see the new site within minutes. The owner is often the last to see it because their own phone and browser still hold the old address in memory, so check on mobile data or somebody else's phone before assuming something has failed.

About a month, unless something is plainly broken. Watch Search Console coverage daily and fix every 404 that appears, but resist changing things in response to normal ranking movement. Without before data on rankings, traffic and enquiries you cannot tell whether it went well anyway, which is why recording that first matters.

Free quote

Want a website that works as hard as you do?

Tell us what you do and we will come back with clear, honest next steps.

Get in touch