Retail
Maison Ardent
A headless replatform that made speed the brand feature
- Industry
- Luxury retail
- Engagement
- Design system + build
- Team
- 5 specialists
- Duration
- 16 weeks

The problem
Maison Ardent's themed storefront took over six seconds to render product pages on mobile. Every campaign needed developer time, and the brand team had run out of patience.
- 6.2s largest contentful paint on mid-range mobile devices
- Campaign landing pages took two weeks of developer effort each
- Twelve conflicting button styles across the storefront
- Checkout abandonment 19 points above the category benchmark
Our approach
Next.js · Shopify · Vercel
Design system first
We built a tokenised component library with the brand team before touching templates, retiring the twelve button variants down to three.
Headless with Shopify retained
Commerce, inventory and payments stayed on Shopify while the storefront moved to Next.js, keeping operational tooling the merchandising team already knew.
Composable campaign pages
Marketing assembles pages from approved blocks in the CMS, so a new drop goes live in an afternoon without a deploy.
A performance budget with teeth
Image policy, route-level budgets and a CI check that fails the build when LCP regresses beyond the agreed threshold.
Timeline
Audit & design system
Weeks 1-5Performance audit, component inventory, tokens and a documented library in Storybook.
Storefront build
Weeks 6-11Catalogue, product detail, search and cart on Next.js with Shopify as the commerce backend.
Campaign tooling
Weeks 12-14CMS block library, preview workflow and editor training for the brand team.
Launch & tuning
Weeks 15-16Phased traffic cutover, A/B validation on checkout and post-launch performance tuning.
Results
- 1.4s
- Median largest contentful paint on mobile
- +34%
- Increase in conversion rate
- 1 day
- Campaign page turnaround, down from 2 weeks
- -22%
- Checkout abandonment
- Marketing ships campaigns without engineering involvement
- One documented design system across storefront and email
- Performance regressions caught in CI instead of by customers
“Our team launches a drop in an afternoon now. The site finally feels as considered as the product.”


