The new site goes live, and two weeks later organic traffic is cut in half. SEO loss during a website redesign or platform change usually comes not from the design but from changed URLs not being redirected, lost content and technical settings put live without testing. This guide explains how to plan a site migration from an SEO perspective, what to check on launch day and what to monitor afterwards.
Types of migration and their risks
What determines the risk is the scope of the change. A site where only the design changes and a site where both the domain and the URL structure change aren’t migrated with the same plan.
| Change | Risk |
|---|---|
| Design only (URLs and content unchanged) | Low–medium: heading structure, internal links and speed may change |
| Platform (e.g. custom CMS → WordPress or Next.js) | Medium–high: URL structure, metadata, rendering behavior |
| URL structure change | High: every URL needs a redirect |
| Domain change | High: all signals must move to the new domain |
| Merging multiple sites | High: content mapping and canonicalization |
Piling several changes onto the same day doesn’t add the risks, it multiplies them. If possible, split a domain change and a platform change into separate periods; that way, when traffic drops, it’s possible to tell why. A loss that can’t be isolated can’t be fixed.
Stage 1: Inventory
A redirect map is only as good as the inventory. Listing only the pages reachable from the menu and moving on leaves out old campaign pages that have gained backlinks over the years, PDF catalogs and image URLs.
- A full crawl of the current site: all URLs, status codes, titles, meta descriptions, canonicals.
- Pages and queries that received clicks over the last 12–16 months, from Search Console.
- The pages bringing the most traffic and conversions, from analytics.
- Pages with backlinks (link analysis tools or the Search Console links report).
- “Orphan” pages that are in sitemaps but don’t appear in the crawl; PDF and media files.
Server access logs are also part of the inventory. Extracting the URLs requested by Googlebot and real visitors from the last few months of logs reveals addresses the crawler can’t find but that still get traffic. Gather the inventory in a single table: old URL, status code, traffic, number of backlinks and a column for the new counterpart. This table becomes the source of the redirect map.
Stage 2: URL mapping and the 301 map
A new counterpart is defined for every valuable old URL. The rules:
- Where possible, URLs stay the same; if they must change, a single-hop permanent (301) redirect is set up.
- The redirect goes to the closest content equivalent; redirecting en masse to the home page can be treated as a soft 404.
- Redirect chains (A → B → C) and loops are avoided; old redirects are updated to point to the new target.
- Pages with no counterpart and no value can return a 404 or 410.
- Pattern-based rules (e.g.
/urunler/*) are used where possible, with exceptions defined one by one.
The chain we see most often in the field is protocol and domain redirects stacking on top of the URL redirect: http:// → https:// → www → new path. The user doesn’t notice. The crawler does. Write the redirect on the web server or CDN as a single rule that goes straight to the final target. The trailing slash (/), case differences and query parameters must also be handled separately in the mapping table; if the old platform uses parameterized addresses such as /Urun.aspx?id=42, each parameter value is a separate mapping row.
Don’t forget media files. When image URLs change, traffic from image search and backlinks to images are lost. When moving to WordPress, the new path structure under /wp-content/uploads/ must be mapped to the old image addresses.
Stage 3: Test environment checks
- The test environment is closed to search engines (password or noindex) — but this setting is written into the checklist so it isn’t carried over to live.
- A unique title, meta description, a single H1 and the correct canonical on every page.
- Structured data, hreflang (if any), Open Graph tags.
- Internal links point to the new URLs; no internal links go to redirects.
- Content comparison: text, images and alt text are complete on important pages.
- Performance measurement: the new site isn’t slower than the old one.
Another trace to check in the test environment is the test environment’s domain itself. If the test domain remains as an absolute URL in canonical tags, sitemap addresses, internal links and structured data, the search engine is directed to these addresses after launch. In WordPress, the address change in the database should be done with a tool that doesn’t break serialized data (e.g. wp search-replace). Also, on pages built with JavaScript, check whether the content is in the initial HTML response; content that isn’t visible without rendering may be indexed late or incompletely.
Stage 4: Launch day
- DNS TTL values are lowered a few days in advance.
- The content freeze period is announced; a final content sync is done.
- After launch, the 301 map is tested automatically (do all old URLs go to the expected target in a single hop?).
robots.txtand meta robots are checked; it’s verified that no noindex has been left over from the test environment.- The new sitemap is submitted to Search Console; if the domain changed, the Change of Address tool is used.
- Forms, analytics and conversion tags are tested.
Don’t test redirects by hand. Running the old URL list through curl -sI -o /dev/null -w '%{http_code} %{redirect_url}' in a loop gives the first response code and target for each address; the output is compared with the expected target in the mapping table. Every row that returns a 302, enters a chain or falls to a 404 is fixed on launch day. In WordPress, the “Discourage search engines from indexing this site” box should be one of the first lines on the launch-day checklist. Left ticked from the test environment, this box is the most expensive single-click mistake in the field.
Stage 5: Post-launch monitoring
For the first 8–12 weeks, 404 logs, Search Console coverage and page reports, and the ranking and click trends of the most important pages are monitored weekly. Missing redirects that are found are added immediately. Short-term fluctuation is normal; a lasting drop usually points to an identifiable technical cause.
Monitor by page group: products, categories, blog, corporate pages. If total traffic looks stable while one group drops completely, a pattern rule for that group isn’t working. Keep the old site’s crawl data and the inventory table after launch too; for a loss that surfaces months later, it’s the only source you can compare against.
Frequently asked questions
How long should 301 redirects be kept?
Google recommends keeping redirects for at least a year; for backlinks and bookmarks, it’s better to keep them as long as possible.
What should we do if traffic drops after migration?
First review the 404 and redirect logs, the Search Console coverage report and the pages that lost the most traffic. In most cases a missing redirect, a wrong canonical, a leftover noindex or lost content is found and fixed. Short-term fluctuation is normal.
What else needs to be done if the domain is also changing?
The old and new domains are verified in Search Console, URL-level 301s are set up and the Change of Address tool is used. The old domain must be renewed and the redirects kept in place.
For technical SEO checks, see our technical SEO and site architecture service; for the whole migration, see our modernization of legacy websites service. If you’re at the platform decision stage, the WordPress or Next.js guide may also help.