(ZIP FILE ATTACHED - SCROLL TO THE RIGHT) One of our members, Oliver, closed a €10,000 build plus a €2,500/month SEO retainer using exactly what's in this post. Not because the site looked good. Plenty of people can make a site look good now. Because it ranks, and because the client can run it without calling him. Everything I used is attached. The prompt that builds the site, the prompt that makes it rank, and the prompt that gives the client a CMS. THE THREE WAYS TO BUILD THESE SITES Every animated site I've built lands in one of three buckets. Pick one per site, don't try to do all three. 1. Scroll hub. One video, exported to frames, and the scroll position plays it forward or backward. That's the whole trick. Best for a single product launch, a flagship service, or a brand that needs one bold story from top to bottom. 2. Component build. Not one video. Separate elements that move at different speeds and react to each other. The plumbing site in the video is this: a 3D house that pulls apart as you scroll so you see the bathroom, then the kitchen, then the hot water system. Best when you have several services to show. 3. Custom interactive. One bespoke mechanic that only exists on that site. A tear, a pull, a grab. Silly, and it works: people stay on the page longer, and time on page is a metric Google pays attention to. You don't invent these, you steal the idea and hand it to the AI. Where to find them: - Scroll hubs: godly.design and motionsites.ai - Components: 21st.dev/community and app.spline.design/community - Custom mechanics: tympanus.net/codrops/demos and awwwards.com/websites/interaction-design Most of these let you copy the prompt straight off the page. Paste it in, change the product, and you have the build in about five minutes. THE PART THAT MAKES IT WORTH $10,000 A beautiful site nobody finds is a hobby. Run Lighthouse on anything you build and check four things: - Performance: 90+ - Accessibility: 90+ - Best practices: 90+ - Agentic browsing: 4 out of 4, or 3 out of 4 at worst