DigitalBellcrest Group·2024·Manchester, UK
An e-commerce rebuild that didn't fall over on launch day
The Black Friday storefront that finally survived Black Friday.
The Black Friday storefront that finally survived Black Friday.
“First Black Friday in five years I didn't refresh the status page every ten minutes.”
— Head of Engineering, Bellcrest Group
Bellcrest Group is a Manchester-headquartered retail group whose legacy storefront had a very specific problem: it broke every year on the one day it needed to hold. At 4x baseline traffic the platform buckled, and their Black Friday campaign was already scheduled to drive 10x. Every developer they had spoken to had promised a nine-month re-platform. Bellcrest had four. Echo & Ember runs a Manchester-adjacent engineering team through Slipy Technologies, and our headless commerce and SSR e-commerce practice was built for exactly this brief — brands that need serious performance engineering shipped inside a real deadline, without the classic split between a design studio that ships a Figma file and an engineering vendor that ships six months late. From day one, product design and engineering sat in the same standup. We migrated the store to a headless commerce stack, rebuilt the storefront in TanStack Start for server-side rendering performance, and made a decision that turned out to matter more than any front-end optimisation: the checkout got moved to a fully isolated Cloudflare Worker so that a product-page traffic spike physically could not take the till down with it. The design team shipped a redrawn PDP flow with a live inventory read, a rewritten guest checkout, and an accessibility pass that the retail team had been asking for since 2021. Full stack, design to launch, in eleven weeks. Then Black Friday arrived. 99.98% uptime across launch week, through a 10x baseline spike. Page loads dropped from a p75 of 4.1 seconds to 1.2 seconds. The engineering team stopped being on call for the storefront for the first time in three years, and the marketing team stopped scheduling promotions around infrastructure limits. This is what a headless commerce rebuild from a UK-based creative studio with an in-house performance engineering team can actually ship inside a hard deadline.
The previous platform buckled at 4x baseline traffic. Their Black Friday campaign was already scheduled to drive 10x. Every developer they'd spoken to had promised a re-platform in nine months; they had four.
Because we build in-house through Slipy Technologies, design and engineering ran in the same standup from day one. We migrated to a headless commerce stack, rebuilt the storefront in TanStack Start for SSR performance, and put the checkout on an isolated worker so a product-page spike couldn't take down the till. Full stack shipped in eleven weeks.
99.98% uptime through a 10x launch spike. Page loads dropped from a p75 of 4.1s to 1.2s. The engineering team stopped being on call for the storefront; the marketing team stopped scheduling around infrastructure limits.
