Website migration checklist with every SEO step until after go-live

Why do migrations cost rankings?

A website migration changes templates, URLs, navigation and often the CMS at the same time. Each of these changes can remove signals that Google has collected over years. That is why Google's site move guide advises changing only one thing at a time. In practice, this is rarely possible. All the more important to know the typical causes. All nine are explained in detail with examples in the guide SEO risks in a website migration.

Phase 1 · Before the migration

Anchor SEO early in the project

Migrations usually start because of design, conversion or a new shop system. SEO often only comes in once the templates are finished. So assess at the start how much is at stake. A website without SEO basics can gain from a migration. A website with strong rankings mainly has to hold its ground.

Using data from the last 12 to 18 months, work through three scenarios: status quo, growth, and the case where the migration ignores SEO. The difference to the worst case shows project management, IT and management why SEO needs budget and time. Also plan SEO and IT capacity for the weeks after go-live. How to calculate the scenarios is explained in the guide Website migration traffic forecast.

Securing important content and URLs

• Export all URLs with clicks from Search Console, for the longest available period.
• Mark pages with many impressions and positions just behind the first page. They have potential and should stay.
• Collect URLs with backlinks from Search Console and your link tool.
• Keep PDFs and images that have clicks or links. Ideally, image file names stay the same.
• Add URLs from sitemaps, server logs, analytics and the CMS to the list. Google names exactly these sources for the URL list.
• From a crawl of the live website, save the title, meta description, H1, alt text and the anchor texts of internal links for every URL. This is your reference for the comparison on staging.

The individual steps, including data sources, PDFs and images, are covered in the guide Important content in a website migration.

Redirect mapping

Every old URL gets exactly one target with the same or better content. Sort the list by clicks so that the most important pages are checked first. Existing redirects from earlier changes also belong in the mapping. Their target is updated to the new URL so that no chains form. Content without a successor returns 404 or 410.

If the old website still runs on HTTP, the switch to HTTPS belongs in the mapping too. If the server first redirects from http to https and then to the new URL, every old URL ends up in a chain. Plan one direct rule to the final target for every variant of the old URL.

Phase 2 · Checking staging and DEV

On staging, every error is still cheap. Protect the environment with HTTP authentication and deploy the real robots.txt early. Why a complete robots.txt block on staging is risky is explained in the guide Staging environment and SEO.

Matching old and new per URL

Crawl the live website and the staging environment and compare every old URL with its successor. These are the key questions.

1. Is the keyword the old page ranks for still in the title, the H1 and the text?
2. Have the title and meta description been carried over or improved?
3. Is the content still there, or are sections, product texts or FAQs missing?
4. Does the page still get a similar number of internal links, including from the navigation and breadcrumbs?
5. Do the internal links still carry the same anchor texts? A new label in the main navigation changes the anchor text on every page.
6. Is the page still at the same click depth, or has it slipped further back in the navigation?
7. Are there roughly as many indexable pages per page type as before, such as categories, products and guides?
8. Are canonical, hreflang and structured data correct?
9. Does the rendered page show the same content and links as the HTML?

Weight the differences by the clicks of the old URL. Five missing product texts on pages without traffic weigh less than one missing category text on the strongest landing page.

Breadcrumbs carry a large share of these links. If the breadcrumb navigation disappears, parent pages lose the links from all their subpages at once. Replace it, or count how many internal links the most important categories still get afterwards.

Checking templates and technical setup

Some errors sit in a template or in the server configuration and then hit every page of that type. Check these points once per template and then by sampling in the crawl.

Testing redirects on staging

1. First check whether staging returns a real 404 for unknown URLs. Otherwise missing redirects go unnoticed.
2. In the old URL list, replace the domain with the staging domain and crawl the list.
3. Every old URL must point to its target from the mapping with a single 301, and the target responds with 200.
4. Fix and crawl again until the list is clean.

How to find chains, loops and wrong targets in detail is covered in the guide Testing redirects.

Phase 3 · Go-live

• Check the robots.txt of the live version. Google caches the file for up to 24 hours, so an accidental block keeps having an effect.
• Remove noindex, X-Robots-Tag and password protection from development.
• Serve redirects server-side as 301 or 308. Why 302 and 307 don't fit here is explained in the guide 301 and 302 redirects.
• Submit the new XML sitemap in Search Console.
• Return status 503 during maintenance windows, ideally with a Retry-After header.
• Only for a domain or subdomain change, report the move, to Google in the Search Console Change of Address tool and to Bing in the Site Move tool of Bing Webmaster Tools.

Directly after the switch, crawl the complete old URL list. This also works without a tool.

Every line with a 404, a 302 or a wrong target goes on the to-do list, sorted by clicks.

Phase 4 · After go-live

According to its documentation, Google temporarily crawls the new website more heavily after a move. For medium-sized websites, Google says it takes a few weeks for most new URLs to replace the old ones in search. For large websites it takes longer.

For monitoring, submit two sitemaps from the mapping, one with the old and one with the new URLs. In the page indexing report, filtered by each sitemap, the number of indexed old URLs drops and the number of new ones rises. Warnings about redirects in the sitemap with the old URLs are normal. In the performance report, compare clicks and impressions of the new URLs with the period before the migration.

Can you limit the crawl rate after go-live?

Older checklists recommend throttling the crawl rate in Search Console. Google switched that setting off on 8 January 2024. Since then, Googlebot sets its own pace and automatically crawls more slowly when server errors pile up or response times rise sharply.

Enough server capacity for the weeks after go-live therefore matters more. For large websites, Google recommends telling your hosting provider about the move in advance. In an emergency, crawling can be slowed down for a few hours, at most one to two days. Googlebot then gets status 500, 503 or 429 instead of 200, ideally with a Retry-After header for 503 and 429. If Googlebot sees these codes on the same URL for several days, the URL may drop out of the index. If your server can't return error codes, Google offers a form for a special request.

Special case: domain change

The Change of Address tool requires that you have verified the old and the new property with the same account and that the old homepage redirects with a 301. According to Search Console Help, the notice is in effect for 180 days, and the redirects should stay at least that long. Google's general site move guide says at least one year. Stick to the year and keep the old domain active, including its SSL certificate.

For Bing, report the move in the Site Move tool of Bing Webmaster Tools. The website has to be verified there first. Bing advises keeping the redirects for at least one to two years. Verify the new domain in both tools before go-live so that data comes in from day one.

Common mistakes

• All old URLs redirect to the homepage.
• The staging robots.txt with Disallow: / goes live.
• Old redirects keep pointing to earlier targets and form chains.
• The new navigation takes internal links away from categories.
• The breadcrumbs disappear without a replacement.
• An X-Robots-Tag: noindex in the HTTP header is left over from development.
• Error pages return status 200.
• Redirects are removed again after a few months.

FAQ