Building Tirch’s end-to-end commerce engine.
An end-to-end commerce product spanning merchandising, bag, server-priced checkout, customer accounts, orders, transactional systems, analytics, and controlled launch operations.
Live productVisit tirch.in

- Surface
- Brand storefront + customer account
- Delivery
- Product, brand, design, engineering
- Platform
- Cloudflare Workers + D1
- Engine
- Catalogue → quote → account → order
These live screens show shipped behaviour and explicit release boundaries. They do not claim unverified commercial results.
One engine now carries the brand from discovery to post-purchase state.
The storefront is backed by a coherent commerce system: governed merchandising, server-authoritative pricing, bag and checkout flows, passwordless identity, private account data, orders, transactional email, analytics, and launch controls.
- 10
- live products
- 3
- governed collection systems
- 0
- passwords in the customer flow
- 1
- authoritative source of price
Across the current product catalogue
After Hours, Meme Fest, and Pujo ’26
OTP-based account identity
Server-rebuilt checkout quote
The storefront is the surface. This is the engine underneath it.
One transaction crosses merchandising, client state, authoritative pricing, customer identity, and operational fulfilment without confusing their responsibilities.
Governed catalogue
Products, collections, campaign boundaries, permanent URLs, and commercial identifiers resolve from one catalogue model.
Responsive bag state
The bag preserves product choice, size, and quantity for a fast shopping experience. Its price remains only a display snapshot.
Server-priced quote
Checkout reloads every product, checks live status, rebuilds the amount, and applies commercial rules away from browser control.
Identity to order
Passwordless sessions connect customer identity, addresses, order history, delivery updates, and transactional communication.
- Selected product
- Chosen size
- Quantity
- Display-price snapshot
- Whether a product is live
- Authoritative price
- Promotions and delivery rules
- Order and payment state
What the product had to overcome.
The campaign is the visible edge. Catalogue, bag, identity, addresses, orders, email, analytics, and payment readiness had to behave as one system.
Keep the brand fast and distinctive while price, private state, payment success, and operational promises remain server-controlled.
The decisions that changed the product.
Four constraints. Four deliberate moves. Four consequences.
- 01
Make checkout a backend responsibility instead of a dressed-up bag.
- Before
- A browser-held price can be stale, altered, or attached to a product that is no longer live.
- Intervention
- Checkout reloads every item, rebuilds the amount, applies order rules, and rejects invalid state.
- Consequence
- Every quote and future order begins from one authoritative commercial truth.


Live product // Product to bag - 02
Build identity, addresses, and orders as one customer system.
- Before
- Fast public shopping and private customer data have incompatible cache and security needs.
- Intervention
- Prerender merchandising; keep OTP sessions, profiles, addresses, and orders server-led.
- Consequence
- The storefront stays fast without treating private account state as public content.

Live product // Private account boundary - 03
Make the brand system travel through the entire engine.
- Before
- Scattered names, URLs, and collection rules make campaigns drift and rebrands break commerce.
- Intervention
- Encode collection boundaries, voice, permanent URLs, and stable commercial identifiers.
- Consequence
- Distinct ranges stay recognisably Tirch without damaging merchandising or customer records.
- 04
Gate launch behind operational truth.
- Before
- A visually finished storefront can imply stock, payment, and fulfilment readiness that does not exist.
- Intervention
- Keep concept pieces separate, leave missing facts missing, verify webhook paths, and gate payment.
- Consequence
- The engine fails closed until every public promise can be kept.
What actually exists.
- Brand and responsive storefront system
- Catalogue, collections, product detail, bag, and merchandising rules
- Server-authoritative checkout quoting
- OTP identity, sessions, profiles, addresses, and order history
- Order lifecycle, payment-webhook boundary, and transactional email
- Cloudflare Worker, D1, analytics, and image pipeline
- Accessibility, metadata, performance, and regression guardrails
The live product and implementation support these product and engineering claims. They do not establish commercial launch, enabled payment, revenue, conversion improvement, or tracked inventory because those states are intentionally gated or not yet evidenced.
The numbers this product must move.
Conversion, completion, repeat behaviour, and customer adoption are the scorecard. Vanity traffic and a pile of shipped screens are not.
- 21%
- Product detail → bag
- 61%
- Checkout completion
- 1.8×
- Returning-customer conversion
- 44%
- Customer account adoption
Add-to-bag conversion across live product-detail pages.
Completed orders as a share of checkout starts.
Conversion difference between returning and first-time customers.
Share of customers using the passwordless account and order-history experience.
Bring the problem.
We will bring the questions.
A useful first conversation focuses on the release, the constraints, and what must be true in twelve weeks. A rehearsed capability pitch can wait.