Case Studies / Platform rebuild
We rebuilt our own site like a client project.
146 inner pages migrated into a new design shell — header, footer, conversion paths, redirect map, and deploy packaging. Staged and documented; the homepage migration is deliberately held for founder review.
The challenge
Starting pointA site that had outgrown its shell.
The agency site needed a full design-system migration, but the groundwork underneath needed just as much attention as the pixels on top.
The site had outgrown its fixed-width layout. The mobile navigation drawer rendered broken across inner pages. The live redirect map was incomplete — the www canonical, http-to-https upgrade, and legacy 301 rules were missing from the live server configuration. And deploy packages were shipping file permission modes that caused assets to 403 after extraction on the new hosting.
A redesign that ignored any of that would have shipped a prettier version of the same problems.
The approach
MethodRebuild in place. Stage everything.
Keep every URL, every page's content, and every page's metadata intact — change the shell around them.
Four rules governed the work. Rebuild in place: no URL changes, no content rewrites, no metadata churn — the migration touches presentation and infrastructure only. Stage everything, deploy nothing unreviewed: every build is packaged, checksummed, and previewed before anyone touches production; deploys stay a manual founder decision. Deterministic packaging: one build script merges the redirect rules and security headers, verifies the archive, and rotates the previous build into backup. Reconcile, don't re-add: the legacy redirect map gets checked against the sitemap before anything is reinstated.
Implementation
What changedNew shell, reconciled infrastructure.
146 inner pages now share one shell: a floating pill header with the new lockup, “Check My AI Visibility” as the primary conversion call-to-action, a rebuilt footer with contact card, newsletter signup, and social links — all on the agentic theme with brass accent as the default.
Mobile navigation was fixed across all 146 pages. The sitemap was rebuilt to cover the migrated pages. The 92-rule redirect map was reconciled against the sitemap: 46 rules active, 48 deliberately left disabled — each of those 48 would have 301-redirected a live, indexed page, which would have been an SEO regression rather than a migration fix. Seven dead tool pages return 410 (gone) instead of soft-redirecting to a generic page. File permissions are normalized to 644/755 before packaging so extracted assets serve correctly on the new host.
Results
Verified outputWhat this produced — and what it didn't.
Only verifiable facts. This work is staged, not yet deployed, so no traffic, ranking, or conversion claims are made.
146 inner pages migrated into the new shell, content and metadata untouched. 1 sitemap, rebuilt. 1 92-rule redirect map, reconciled and documented — including a standing decision record for why 48 rules stay disabled. Every deploy package integrity-checked with correct file permissions. The homepage migration is intentionally held back: index.html keeps the old design in staging until the founder reviews and approves the port.
Staged work. Results above describe build artifacts, not site performance — measurement starts after deploy, against a pre-deploy baseline.
What's next
RoadmapHomepage port, deploy, measure.
The remaining work is sequenced, not stacked.
First, the homepage port — a founder decision, already staged as a separate gate. Then the deploy itself, followed by measurement against the pre-deploy baseline: crawl health, redirect behavior, and conversion-path performance. The redirect decision record travels with the build so the next migration doesn't re-litigate it.
Make the next move yours.