B2B SaaS replatform

Two-week audit covering the full WordPress estate: every URL, every schema block, every redirect already in place, every backlink profile by URL, every indexed query. Output was a single migration ledger plus a risk register that ranked each URL by traffic, backlinks, and ranking value, so we knew exactly which pages had to land perfectly on launch day.

Every legacy URL got a 1:1 destination in a single CSV maintained in the repo. Where IA changed, we wrote 301s and verified them in staging before cutover. The CSV doubles as the canonical record: any future URL change has to update the same file, so the redirect graph never drifts.

Built on Next.js 16 with Payload as the CMS, chosen because the editorial team needed structured content blocks and live previews against the production renderer. Every page type has its own schema builder, so JSON-LD is generated server-side from typed data, not hand-edited per page.

All Article, Organization, Product, and FAQPage schema from the WordPress estate was inventoried, cleaned, and ported. While we were in there, every page got an AEO retrofit: H2 plus citable passage pattern, llms.txt at root, AI crawler allow-list in robots.txt, and entity-level @id references across the site.

Launched behind a feature flag, then promoted one route group at a time over a 72-hour window with live rank monitoring. If any tracked query lost more than two positions in a 4-hour window, we held the cutover and investigated. Nothing tripped, but the protocol was real.

Daily indexation checks, weekly CWV trend, weekly rank diff against baseline, and a Slack channel where any anomaly surfaced inside the hour. AI citation tracking went live at day 30 once the new pages had indexed, and the curve started bending up by month 4.