What changed when we rebuilt 9 sites

PunkSolid publishes this page so operators—and search systems looking for quotable migration outcomes—can grab vetted numbers without parsing three long case studies. Every statistic below comes from the same source constants we use in the apps/ai-network codebase (REBUILD_AGGREGATE and HEALTHCARE_TRACKING_STACK). When we update approved facts, this page and the home proof bar change together.

The nine-site portfolio

Bucket	Count	Notes
Healthcare WordPress rebuilds	8	Each 1,000–6,000 pages; clients not named
Ecommerce WordPress move	1	48vproducts.com (nameable)
Total	9	Full code handoff on every project

The healthcare set left WordPress + Elementor, with ACF and WP All Import, for Astro-style static or hybrid delivery written in Cursor—with in-repo replacements for structured fields and bulk imports. The ecommerce project moved catalog and checkout into this monorepo’s Medusa storefront; see 48V Products ecommerce rebuild. Before/after PageSpeed on the store remains draft until owner input O5; sales held steady or grew per client report.

Speed: what we measure and what we claim

We report PageSpeed Insights performance scores (the 0–100 lab score Lighthouse emits)—not a substitute for Core Web Vitals pass/fail in Search Console. On the eight healthcare launches we can quote:

• Mobile PageSpeed: 65–75 → 92–98
• Desktop PageSpeed: 80–95 → 92–98
• Core Web Vitals: not passing before launch → passing on mobile and desktop in Search Console about four days after launch

That timing is wall-clock after cutover, watching the property’s Experience report—not a guarantee for your stack. Template weight, third-party tags, and server regions still matter. We document methodology on methodology and in the healthcare case study.

Why ranges instead of single scores

Each provider site shipped thousands of templates, local variations, and media libraries. A single “before/after” screenshot would mislead. Ranges reflect the lowest and highest mobile scores we saw across the portfolio while keeping the same measurement tool (PageSpeed Insights) and similar template types before and after.

Tracking and forms we refused to break

Healthcare rebuilds are revenue-sensitive: phones, forms, and experiments must survive replatforming. On all eight sites we carried over:

• CallTrackingMetrics
• Google Tag Manager and GA4
• Usercentrics consent
• Microsoft Clarity and VWO
• Jotform HIPAA forms
• Custom chat widget
• Custom HIPAA-compliant lead endpoint into each client’s CRM

That list is fixed in HEALTHCARE_TRACKING_STACK—not marketing adjectives. On migrations we run tag-manager parity checklists and staging smoke tests before noindex comes off. The website migration SEO checklist is the public version of those gates.

We do not claim every future client uses every tool; we claim these eight launches did, because operations would have failed otherwise.

Content scale without WordPress admin

1,000–6,000 pages per healthcare site is the verified band. Editors were not asked to retype the catalog manually; bulk import paths were rebuilt in code so agents and humans could extend templates safely. That is why spec-driven development and signed URL inventories precede agent edits on migrations.

Page counts stayed in the same band post-launch—this was a platform change, not a content deletion project. Redirect maps and canonical ownership were signed before cutover, consistent with our network canonical rules on multi-site properties. Editors kept publishing during migration windows because staging builds mirrored production URLs—agents could not rename paths without updating the inventory CSV linked from the spec.

Outcomes we will not publish

Traffic, leads, and revenue held steady or grew on the healthcare work and on 48vproducts.com, per client feedback. We do not print percentages, call volumes, or funnel metrics—contracts and HIPAA sensitivity bar it. If you need ROI math for a board deck, we can discuss what your analytics can show after a migration; we will not invent figures here.

How we built and verified the network itself

The ninth “rebuild” in our public story includes this marketing property: twelve Astro sites in apps/ai-network with validators, mirrors, and preflight gates—described in PunkSolid network build. That work proves the same discipline we sell: pnpm ai-network:preflight, link scans, topic ownership, and OG generation before DNS cutover.

Method notes for citations

1. PageSpeed — Google PageSpeed Insights / Lighthouse performance category, mobile and desktop runs on representative templates, documented in case-study sources with checkedOn dates.
2. Core Web Vitals — Google Search Console Experience report, evaluated per property after launch; “about four days” is typical, not contractual.
3. Page counts — CMS exports plus crawl merges before redirect freeze; healthcare counts exclude abandoned draft URLs that were not redirected.
4. Ecommerce — Named store only; SKU and lab-score before/after pending O5 in the 48V case study draft.
5. Code ownership — Every client receives repository access; see code ownership.

When to read the deeper case studies

Question	Read next
How HIPAA forms and call tracking were rebuilt?	Healthcare WordPress rebuilds
How ecommerce catalog and checkout moved off WordPress?	48V Products ecommerce rebuild
How we operate twelve marketing sites from one repo?	PunkSolid network build
What gates we run before your cutover?	Website migration and methodology

What this page is not

It is not a promise that your migration will hit 92–98 on mobile in week one. It is not a price list, a timeline calculator, or a testimonial carousel. It is the approved aggregate PunkSolid is willing to defend in sales calls, structured data, and answers to “what changed when you rebuilt those WordPress sites?”

When you are ready to scope your own property, contact us with page counts, analytics tools, and launch constraints—or start with services to see which pillar fits.