Rebuilding an Acquired Domain Without Breaking Historical URLs

What we observed at the first technical release

The acquired domain had previously hosted a Chinese agency website. The supplied backlink investigation identified historical destinations; it did not authorize copying the former operator’s claims, clients or relationships. We used it as a route inventory for a publication under new ownership.

After the first technical phase on October 16, the recorded release contained 46 canonical pages, eight articles in two languages and two browser tools. That is a dated snapshot, not the live inventory of every later release. The public verification recorded 474 assertions, including HTML comparisons, redirects, language pairs and error responses.

Download the recorded HTTP-check snapshot. It contains public URLs, statuses and destinations without account credentials or private planning documents. These checks establish the tested behavior at publication time. They do not establish recovered authority, search positions or visits.

The important difference was inside the query

An old address such as /?news/78.html is a request for the root path with a literal query segment. It is not the same route as /news/78.html. A replacement site that serves its homepage first can accidentally answer every old query with the homepage and a 200 response.

Our request handler reads the legacy segment before serving static HTML. It recognizes the documented mappings, removes the route token and retains unrelated tracking parameters. If a request contains conflicting route tokens, it returns an error rather than guessing which article the visitor intended.

A useful equivalent, a retired page, or an unknown address

The four article topics have relevant successors, so their historical queries use permanent redirects to the Chinese editions. Documented old service, case and customer routes with no equivalent return 410. An unknown historical query or an undocumented path returns 404. The error pages offer navigation without pretending that the missing material still exists.

We did not redirect every former page to a single homepage. Google’s migration guidance recommends mapping old URLs to relevant new destinations and avoiding irrelevant bulk redirects. That guidance informs the mapping; the HTTP snapshot supplies the evidence that our implementation behaved as intended. Read the migration guidance.

Root and www, and HTTP and HTTPS, are consolidated with the route mapping so the documented cases reach the final destination in one hop. Additional parameters survive. This makes the link useful to a reader even when an old URL appears in a source we do not control.

Why the historical redirects still lead to Chinese

English is the principal language of the new journal, but the old article queries belonged to Chinese content. Keeping a Chinese successor preserves that immediate reading context. The corresponding English page remains available through the page’s language switch.

Each complete edition has its own canonical URL. Language annotations reference English, simplified Chinese and the English fallback, and both pages list the same alternatives. We use zh-Hans in those annotations. This is a publication decision based on the available content; it does not prove demand or search visibility in any country.

The Hreflang Builder can prepare an equivalent map for another project. Review the actual deployed pages as well: the tool never fetches a URL or verifies that its translation is complete. Google explains reciprocal language annotations.

Repeat a public check without privileged access

For a header-level check, request an old query and inspect its Location header, then request that destination. These examples use public addresses and do not require a Cloudflare account:

curl -I 'https://www.jinchenggeo.com/?news/78.html'
curl -I 'https://www.jinchenggeo.com/zh/geo/what-is-generative-engine-optimization/'

The first documented response is 301 to the Chinese article; the second is 200. Also test a retired query such as /?service/, an unknown query and the plain path /news/78.html. They are separate cases, with 410, 404 and 404 respectively in this implementation.

A header check does not establish the quality of the replacement text. Our release process also compared the served HTML with the release artifacts, checked language links and inspected pages in a browser. Preserving a redirect while replacing the destination with unrelated copy would not fulfill the reader’s task.

What another publisher can reuse

Build a private inventory of old URLs and evidence before drafting redirects. For each address, record the old topic and language, the relevant new destination or retirement decision, the expected response and the result after deployment. Keep the evidence separate from private access tokens and unapproved content.

Use the Content Brief Builder to define the new page’s scope and unanswered questions. Use the Robots.txt Tester to review rules, then verify actual hosting access independently. Test routes after publication, not only in a local build.

The verified outcome here is a working bilingual route contract and a recorded first-release check. Whether search systems choose, index or cite the new pages requires separate observation. For that next task, continue with our indexing diagnostic.