CISNET: IT certification training.
Marketing site + SEO rebuild.

Legacy PHP that couldn't grow.

CISNET's existing site was a hand-coded PHP build from 2014. Maintainable because one person knew where everything was. Breakable because nobody else could ship into it without an outage. Marketing wanted new certification pages. Paid campaigns into specific cohorts. Leads captured into a CRM. Engineering reality: every new page was a Friday deploy with the lights flickering.

The deeper issue was SEO momentum. The legacy site ranked for ~70 cert-related queries. Any redesign that didn't preserve those URLs risked tanking organic traffic for a quarter or more. The business depended on that organic flow for ~40% of intake.

• Brittle stack: PHP 5.6, no test suite, manual FTP deploys
• No lead infrastructure: contact-form-to-email only, no CRM, no admin view
• SEO at risk: 70+ ranked legacy URLs that couldn't break
• Marketing blocked: couldn't ship landing pages without engineering

Next.js 16 + programmatic SEO + lead infrastructure.

Three parallel workstreams over 12 weeks. Stream 1 (weeks 1-4): stand up the Next.js 16 + Supabase + Vercel foundation. Port the marketing pages 1:1. Run the redirect map in dev. Stream 2 (weeks 4-8): programmatic SEO. Generate 200+ cert × city pages from a structured content model. Ship sitemap.xml + 7 distinct schema types. Stream 3 (weeks 8-12): lead pipeline. Facebook Lead Ads webhook with HMAC verification. Supabase storage. Admin portal. Resend transactional email. Stripe Checkout for paid courses.

Hardest call early on: how aggressive to be on the redirect strategy. I mapped every legacy URL individually rather than rely on Next.js catch-all rewrites. Slower to build. Google indexed the new URLs with full link equity inside 6 weeks. Not 6 months.

• Foundation: Next.js 16 (app router), TypeScript strict, Supabase, Vercel
• SEO surface: 200+ pSEO pages generated from a cert × city matrix
• Redirects: 70+ legacy URLs mapped individually (not catch-all)
• Schema: 7 types: Course, LocalBusiness, FAQPage, Organization, BreadcrumbList, ContactPoint, WebSite
• Lead pipeline: FB Lead Ads → HMAC-verified webhook → Supabase → admin portal → Resend
• Payments: Stripe Checkout + course-enrollment webhook

Boring, fast, debuggable.

Default opinionated stack: Next.js + Supabase + Vercel. Could've used a CMS for the marketing pages. Editorial volume didn't warrant the operational tax. Page content lives in MDX in the repo. Engineers ship copy through PRs.

12 weeks. 3 streams. No scope creep.

Locked scope in week 0. Anything that came up after went into a "v2 list." None of it shipped during the engagement.

What it enabled.

Headline numbers from the engagement. Technical surface only. Business KPIs are CISNET's to disclose.

Beyond the numbers: marketing ships new certification pages through MDX without engineering. Lead routing is observable end-to-end. Ad campaigns target specific cohort × city combinations that previously had no landing page.