I replaced my own WordPress site. This is what costs you your rankings

Replacing a site is the easy half technically. The hard half is that Google knows four years' worth of your pages and still expects to find them tomorrow. Get that wrong and you watch your organic traffic halve within two weeks, and it takes months to come back. I went through this on my own site.

Start with the old site's sitemap, not with the CMS

The temptation is to work through the page list in your WordPress admin. Do not. That list holds drafts, private pages and test pages that were never indexed, and it is missing exactly the URLs Google does know: categories, tags, custom post types. Open sitemap_index.xml on the live site instead. That is the list that counts.

For me that came to 87 URLs across seven sub-sitemaps. Seven of them I would never have found by hand: taxonomy pages like /website-type/webshop/ that WordPress had quietly created and that had picked up backlinks in the meantime.

Redirect to the subject, not to the homepage

The most common mistake: sending every old URL to the homepage in one rule. Google treats a 301 to a page that does not cover the old subject as a soft 404. The redirect is followed, but the authority is lost. You end up with no 404s at all and your rankings gone anyway.

Every old URL belongs on the nearest successor in substance. If that does not exist, that is a content decision, not a redirect decision.

Test against a real server, not against your config

A redirect map that looks right in your configuration file tells you nothing. Old URLs sit in Google with a trailing slash, and frameworks normalise that first, only then does your redirect get its turn. I requested all 87 URLs with a trailing slash against a production build and recorded the status code plus the final destination. Two redirects turned out to point at a page I had renamed along the way.

What you lose regardless

I had a page ranking on a local search term and a handful of WordPress how-tos that had been pulling traffic for years. I do not sell those subjects any more. No redirect fixes that: when a subject leaves your site, the ranking leaves with it. That is a perfectly good decision, as long as you make it deliberately instead of discovering it in your Search Console.

The checklist I work from now

• Pull the URL list from the live sitemap, not from the CMS
• Map every URL to the nearest successor in substance
• Request every old URL with a trailing slash against a production build and log the final destination
• Check that every redirect target actually exists as a route
• Submit the new sitemap and keep the old property in Search Console
• Deliberately accept which subjects you are dropping

It costs you a day. Climbing back up the search results over six months costs considerably more.