B2B SaaS · cloud cost and performance engineering

Domain Migration and Redirect Integrity

Challenge: A 300-row redirect map was ready to ship days before a combined rebrand, domain move and CMS replatform. Only 22 of its 300 targets actually resolved.Results: Shipped at 493 rows with 478 verified single-hop 301s. No chains, no www hops, no broken targets.Timeline: Pre-launch QA against a fixed cutover date

The challenge

A rebrand, a domain change and a CMS replatform landing on the same fixed launch date, with the information architecture consolidated at the same time. The build team had produced a 300-row redirect map. We were brought in to QA it, read-only, days before cutover.

  • 278 of 300 map targets returned 404. Only 22 resolved. The map had been written against a content structure that did not exist on the new site.
  • 165 live URLs had no redirect mapped at all, including an entire /en-gb/ language tree and three core service pages.
  • 283 targets pointed at the www host, which 301s to non-www, adding a second hop to the majority of the map.
  • The new domain was still serving robots.txt with Disallow: /.

Approach

Crawl for an authoritative inventory. Rather than trusting the map's source column, we crawled the live site to establish what actually existed: 397 URLs, 395 returning 200, zero noindex. 137 map sources were not in the sitemap at all.

Probe every target against the live new site. Each destination was requested and its real status recorded. This is what surfaced the 404s, and it is the step that is routinely skipped because the map looks right in a spreadsheet.

Find the pattern behind the failures. The 404s were not random. 150 rows pointed at a flat /insights/<slug> path while the migrated articles actually lived at /insights/blogs/<slug>. The content had migrated; the map had not. A further 157 rows carried a trailing slash that redirected onward, creating avoidable chains.

Close the gap, then re-verify everything. We built the 165 missing rows and verified each new target before committing it. When the client edited the sheet directly, we reconciled their 57 value changes and re-probed the full set rather than trusting the earlier pass.

Escalate what is a business decision, not an SEO one. Thirteen URLs had no correct destination. Those went back to the client as explicit decisions rather than being quietly pointed at the homepage.

Results

  • 300 to 493 rows in the final map — the original 300 reconciled and corrected, plus 165 previously unmapped live URLs.
  • 478 verified single-hop 301s — every committed redirect re-probed live: all 200, no chains, no www targets, no 404s.
  • Zero redirect chains, www hops or broken targets in the shipped map, down from 278 broken targets and 283 www hops.
  • 13 decisions escalated to the client rather than guessed, plus 2 deliberate 410s for retired content.

Key insight

A redirect map is a claim about a website that does not exist yet. Until every target has been requested against the live build, it is untested. Probing the destinations, not reviewing the spreadsheet, is what turned a map that would have 404'd 93% of its targets into one that shipped clean.