01 / 05
The challenge
Rkiza had been on Shopify for four years and had grown to $6M in annual revenue. The theme was solid, the operations team knew the platform inside out, and the founder was — reasonably — worried about touching anything. The problem wasn't Shopify. The problem was that the theme had been extended a dozen times by three different agencies, and every optimization now cost more than it returned.
LCP on the product page was 3.8 seconds on mobile. The cart drawer took 1.2 seconds to open. The recommendations engine was firing a 400 KB script before the buy button was even interactive. On top of that, the marketing team was fighting the theme every day: A/B tests required a developer, product pages couldn't vary layout by category, and Klaviyo flows weren't attached to any structured event.
A quiet audit showed the real damage: an estimated 8–12% of visitors were bouncing on the product page before the buy button was interactive. On a $6M topline, that's $500K–$700K in annual revenue silently walking out the door.
02 / 05
Our approach
We built a plan the founder could believe in: incremental headless. No big-bang replatform. We spun up a Hydrogen (Remix) storefront alongside the existing theme, migrated one product category at a time, ran everything through a shared cart, and cut over each category only after conversion held or improved for two weeks.
By week three, the men's category was live on Hydrogen. LCP dropped from 3.8s to 1.1s. Conversion held. By week eight, all four categories were headless. By week eleven, the theme was decommissioned and the shop was 100% on Hydrogen with a lean, opinionated component library.
03 / 05
Key decisions
We stuck with Shopify as the commerce backend. This was a deliberate choice — the operations team knew it, the accounting integrations were already in place, and nothing about Shopify's admin was holding them back. What was holding them back was the theme layer. Headless let us keep everything valuable and replace only what was broken.
We built a Klaviyo events adapter directly into the Remix routes. Every add-to-cart, view-item, view-item-list, and begin-checkout now fires structured Klaviyo + GA4 events with rich product context. Klaviyo abandoned-cart flows now recover at 14%, up from 6%.
We ran server-side GTM from day one. iOS 17 and Consent Mode v2 were already breaking client-side attribution for lookalike audiences. The server-side stack means Meta CAPI, TikTok Events API, and Klaviyo all get the events they need — without the cookie decay that would otherwise starve them.
04 / 05
What changed
Conversion rate lifted from 1.9% to 3.2% site-wide over 90 days — a 68% relative uplift. The blended CAC across paid channels dropped 23% because the ad platforms could finally see what was happening after the click. Klaviyo revenue as a share of total climbed from 17% to 28%. LCP is now 1.1s on 4G; INP is under 200ms across the site.
The finance team could tie an additional $1.2M in annual revenue to the replatform — measured against a 12-week pre/post window with paid spend held flat.
05 / 05
Where we're headed
We're now designing a first-party recommendation engine trained on Rkiza's own purchase and browse history — replacing the generic Shopify app they were paying $499/month for. Early A/B tests suggest another 4–6% lift on category pages.
