Beauty Studios is a £4M ARR premium beauty brand. They'd hit the ceiling of what OS 2.0 could deliver: editorial-led PDPs that needed sub-1s LCP, an interactive shade finder, and AR product try-on.
Two agencies had quoted Hydrogen builds at £80k–£120k. Both quoted 6 months. Both had shipped one Hydrogen build before — neither in production for more than 90 days.
Brief: ship a Hydrogen build with the editorial UX they'd designed in Figma. LCP <1.0s on mobile p75. Live in 14 weeks. Editorial team must be able to update content without a developer.
Decided on Hydrogen + Sanity CMS for editorial content. Storefront API for catalog. Edge caching strategy on Oxygen. Performance budget: LCP <1.0s, INP <100ms, CLS <0.05.
Built a Tailwind-based design system from the Figma file. 47 components shipped to Storybook. Editorial team trained on Sanity Studio.
PDP, PLP, cart, checkout (using Shopify's native Hydrogen checkout). Editorial sections shipped in parallel — Sanity content live by week 8.
Built a small embedded React experience for shade finding. AR via 8thWall on iOS Safari + Android Chrome. Tested across 12 device classes.
Fine-tuned Edge caching. Image preloading. Bundle splitting. Shipped LCP at 0.9s mobile, 0.6s desktop. Soft launch behind password.
Phased migration of OS 2.0 → Hydrogen with redirect map. 30-day daily monitoring of CWV, conversion, error rates.
"Two agencies told us 6 months and £100k+. They shipped in 14 weeks at £68k, and the LCP went from 3.2s to 0.9s. That's not an iteration — it's a different store. Mobile conversion is up 39%. The shade finder paid for itself in the first quarter."
Headless isn't for everyone. When LCP <1s and bespoke UX matter more than editor convenience, Hydrogen earns its complexity. Talk to us.