Snitch.

Most of the year is easy. The engineering exists for the twenty minutes that are not.

Client: Snitch
Sector: E-commerce
Scope: App + web + admin
Stack: Next.js, Node.js, PostgreSQL

Overview

Fashion commerce that survives a drop. Storefront, ordering and operations tuned for peak traffic.

01 — The challenge

What made it hard.

A fashion drop concentrates a month of traffic into a short window. Overselling becomes a support problem that scales faster than revenue, and every extra checkout field costs conversion.

Inventory truth

Overselling is a support problem that scales faster than revenue does.

Checkout friction

Every additional field costs conversion, and payment coverage has to be broad and fast.

Post-purchase

Returns, refunds and exceptions are most of the operational cost and none of the demo.

02 — The product

What we shipped.

Storefront, catalogue, cart and checkout with the order management and operations tooling that handles returns, refunds and exceptions behind it.

03 — The build

How it holds up.

Load tested well above expected peak, with caching strategy treated as part of the product rather than an optimisation applied afterwards.

The system
Inventory is a single source of truth and every surface reads from it, so the storefront cannot promise stock the warehouse does not hold. Read paths are cached aggressively with explicit staleness rather than hopeful expiry.