AhsanAutos.pk: Headless E-commerce Rebuild

The History

This is the second time I’m building this site headless. Understanding why requires some context.

Before 2022: WooCommerce

When the client first approached me, the site was already running on WordPress/WooCommerce. Standard setup, worked fine, but slow and limited in what we could do with the UI.

2022: Headless BigCommerce

I built a headless version using BigCommerce’s Storefront API with a custom Next.js frontend. The performance improvement was significant—faster load times, better UX, and impressions grew substantially.

Then BigCommerce’s pricing became a problem.

BigCommerce charges based on sales volume, with automatic tier upgrades as your GMV increases. The issue: they calculate everything in USD. In Pakistan, a motorcycle costs around 150,000 PKR—roughly $500. But their system was reading it as $150,000 in sales.

A few orders like that and suddenly we’re on the $40-50/month Pro plan. For a Pakistani retailer selling vehicles where margins are tight, that wasn’t sustainable. Add to that: BigCommerce didn’t actually support PKR as a transactional currency. No native payment gateways for the Pakistani market.

2023: Back to WooCommerce

The client moved back to WordPress/WooCommerce. Not because it was better, but because it was affordable and worked with local payment gateways.

2024: Headless Shopify

The client came back. This time they wanted Shopify as the backend—better PKR support, more payment gateway options, pricing that doesn’t scale weirdly with high PKR values.

So I’m building it headless again. Second time.

Current Architecture

Next.js 16 (React 19 Server Components)
    ↓
Jotai State Layer (cart, payment mode, UI)
    ↓
Shopify GraphQL Client
    ↓
Shopify Storefront API

Same pattern as before, different commerce backend.

The Dual Payment Mode Thing

This client sells vehicles. Some customers pay cash, some want installment plans. Standard e-commerce platforms don’t handle this cleanly.

My solution: products carry installment or cash tags. The frontend filters based on the active mode (stored in a Jotai atom, persisted to localStorage). For installments, there’s an interactive calculator that derives monthly payments from configurable markup rates.

Trade-off: Tag-based filtering is less structured than metafields, but it’s simpler for staff to manage.

What’s Different This Time

Aspect	BigCommerce (2022)	Shopify (2024)
PKR support	None (USD only)	Native
Pricing model	Sales volume tiers	Fixed subscription
Local gateways	None	Multiple options
GraphQL API	Decent	Better documented

Shopify’s Storefront API is more mature now. The ecosystem is better. And critically, they don’t inflate your bill because your currency has more zeros.

Technical Choices

Layer	Choice	Why
Framework	Next.js 16	RSCs stream content, minimal client JS
State	Jotai	Atomic updates, RSC-compatible
Styling	Tailwind CSS 4 + OKLCH	Perceptually uniform colors for dark mode
Commerce	Shopify Storefront API	PKR support, stable pricing

OKLCH colors were a good call. Traditional HSL shifts perceived hue when you change lightness—reds drift pink in dark mode. OKLCH doesn’t. Dark mode worked on the first try.

Optimistic Cart

Same pattern as my other e-commerce builds:

User clicks “Add to Cart”
UI updates immediately
Server syncs in background
Rollback on failure

Shopify’s API responds in ~200ms, which is fine for page loads but feels sluggish for cart actions. Optimistic updates fix that.

Status

Currently in development. The site is still running on WooCommerce while I build out the Shopify version. Once it’s ready, we’ll migrate.

Hosting will be Vercel paid tier once live. After three versions of this project, I’ve learned that cutting corners on infrastructure costs more in debugging time than it saves in hosting bills.

What I Learned (Three Versions Later)

Platform pricing models matter. BigCommerce’s sales-volume pricing is designed for USD markets. It breaks badly when your currency has more digits.

Headless isn’t always the answer. WooCommerce worked fine when “fine” was all the client needed. But when they wanted performance and custom UX, headless made sense.

Some clients come back. Building a good relationship matters more than any single project. This client came back because the first headless version worked well—they just needed a platform with better pricing.

Tech Stack