Migrating blog posts, categories and URLs from an AI site to WordPress

Your blog is probably your biggest asset

When owners of AI-built sites talk about migration, they talk about the homepage design. When I look at their analytics, the story is different: often sixty to eighty percent of organic traffic lands on old blog posts. Lose those URLs or those rankings and the migration is a failure, no matter how pretty the new homepage is.

That is why I always start a blog migration with an inventory, not with design. Export or crawl the old site and list every post URL, its title, its publication date, its categories or tags, and its meta description. This spreadsheet becomes the checklist you verify against after the move. Without it you are migrating blind.

Keep slugs identical wherever possible

The safest migration changes nothing about the URLs. If the old post lived at /blog/how-to-price-a-project, the WordPress post should live at exactly that address. Set the permalink structure before importing anything, because changing it later means mass redirects.

Sometimes the AI builder forces a pattern you cannot keep, for example a language prefix or a generated ID in the path. In that case, create a single 301 redirect per changed URL, from the old address straight to the final new address. Never chain redirects: old to intermediate to new costs crawl budget and a bit of ranking signal at every hop. I keep the redirect list in the same inventory spreadsheet, with a column where I tick off each redirect after testing it in a browser.

Dates, authors and categories matter more than you think

Three details are forgotten in almost every blog migration I audit. First, publication dates: if every post suddenly shows today's date, you lose the freshness signal and returning readers get confused. Set the original publication date on each post. Second, authors: if the old site credited a person, create that author in WordPress instead of dumping everything under 'admin'. Third, category and tag pages: these often rank themselves, so give them the same URLs and, if the old site had them, the same intro texts.

Also check the internal links between posts. If post A linked to post B using the old absolute URL, that link now points at a site that may soon go offline. A search-and-replace across the database, done carefully with a dry run first, fixes thousands of links in minutes.

Images deserve their own pass

Blog images from AI builders are frequently served from the builder's CDN. Those URLs stop working when you cancel the subscription, so every image must be downloaded and re-uploaded into the WordPress media library, and every reference in post content must point at the new local copy. Do this before you cancel anything.

While re-uploading, keep the original file names where possible and fill in the alt texts. It is dull work, but image search traffic is real traffic, and alt text is also an accessibility requirement. For large blogs I script the download and upload; for small ones, an afternoon of manual work is fine.

Verify against the inventory

After the import, walk the inventory row by row. Every URL returns a 200? Title and date correct? Meta description present? Redirect working where the URL changed? This sounds tedious and it is, but it is exactly the kind of tedious that protects rankings. When the blog is large or the URLs are messy, a conversion service like Convert2WP that maps pages systematically saves days of manual checking and removes the risk of human error.