The starting point
TerraCommerce came to us running a 5-year-old WooCommerce site that had outgrown its foundation. Their flagship product line was strong, their margins were healthy, and their paid traffic was efficient — but somewhere between the ad click and the checkout button, nearly 99% of visitors were vanishing.
The diagnostics weren't subtle. Page load time averaged 4.8 seconds. Mobile checkout broke entirely on newer iOS devices. Product images loaded sequentially instead of in parallel. The cart icon updated on a full page reload. A baseline Lighthouse audit scored the site at 31/100 on mobile performance.
Conversion rate sat at 1.2% — roughly half the industry benchmark for their category. Every dollar of paid spend was working twice as hard as it needed to.
What we did
We made a hard call early: this wasn't a tuning problem. The old stack couldn't be incrementally improved to where the business needed it to be. We proposed a full rebuild on a modern foundation — and committed to shipping it in 90 days without breaking SEO, ad infrastructure, or live orders.
1. Next.js + edge-optimised delivery
We rebuilt the storefront on Next.js 14 with App Router, deployed on Vercel's edge network. Product pages now ship static HTML, hydrate progressively, and stream personalisation (cart, recently-viewed, recommendations) after first paint. Time-to-interactive dropped from 5.1s to 1.4s.
2. Purchase-intent-led product page redesign
The old PDP was a wall of features. We redesigned it around the three questions a buyer actually asks: Is this for me? Is it worth the price? Can I trust this brand? Hero image, social proof, and pricing live above the fold. Specs, ingredients, and FAQ live below — but loaded eagerly so they're instantly available on scroll.
3. Headless CMS for the marketing team
We migrated content to Sanity, with a custom schema that lets the marketing team launch product collections, landing pages, and seasonal campaigns without touching developer time. The team now ships ~12 new landing pages per month vs. roughly 1 before the rebuild.
4. Conversion infrastructure: analytics, A/B, CRO
We layered in PostHog for product analytics, GrowthBook for feature flags and A/B testing, and Klaviyo for browse/cart abandonment flows. Within 30 days of launch, the CRO loop had identified and shipped seven winning experiments — each contributing to the final conversion number.
"The new site paid for itself in the first month. We went from embarrassed to send people to our website, to using it as our #1 sales tool."
James M., CEO — TerraCommerceWhy it worked
Most e-commerce rebuilds fail in one of two ways: they ship a beautiful site that doesn't move the metrics, or they obsess over the metrics and ship something visually downmarket from the brand. We solved for both by making conversion-rate improvement the explicit success criterion in week one — but treating the brand bar as a non-negotiable.
The 90-day timeline was deliberate. Long rebuild projects fail because teams lose the urgency that creates the right trade-offs. A hard deadline forces honest scoping. Half the value was what we shipped; the other half was what we deliberately didn't.
What's next
TerraCommerce is now expanding into two new product lines and three new markets — using the same storefront as the foundation. We're in retainer with their team handling ongoing CRO, performance maintenance, and new landing page production. Monthly revenue continues to compound.