Case Study
Multi-site content engine
6 sites started on one engine in 45 days
Custom production app · Publishing
The goal
Launch new content sites in days instead of months, each one built to rank in search, get cited by AI answers and turn readers into email subscribers from day one.
The problem
Each new site used to start as a copy of the last one: CMS schema, editorial blocks, renderers and about 50 operator scripts. The copies drifted apart. In one session, three schema modules had to be ported into a live site by hand, and a new block type reached production before the site could render it, breaking a live page.
What we did
Every site runs the same stack: Next.js, an embedded Sanity CMS, and an answer-first page structure (direct answer, key takeaways, FAQ, sources and structured data) that search engines and AI assistants can quote. Each site keeps its own repository, domain, deployment and CMS project, so any one of them can be frozen, handed off or sold.
The shared layer is a set of versioned packages: a content contract, CMS schema, renderers that flag a block they don't recognize instead of silently dropping it, and an email pipeline. A fleet registry lists every site, so audits and analytics run the same way on each one.
The results
- ✦6 site builds started on the engine between June 10 and July 24, 2026, including the site you're reading
- ✦4 shared, versioned packages replace hand-copied code
- ✦Unknown content blocks are flagged at render time instead of breaking a live page
What it does for revenue
- Speed to market
- Six site builds started in 45 days, because each new site starts from the engine instead of a blank repository.
- Visibility
- Every page ships answer-first with FAQ and structured data. One guide on the engine started receiving ChatGPT referrals while its Google ranking hadn't moved.
- Leads
- Email capture is part of the engine: every site stores subscribers and can forward them to an email platform, so traffic starts building a list on launch day.
- Margins
- Shared packages replace the hand-porting that used to eat whole working sessions, so the cost of running each additional site keeps falling.