The challenge
Northwind is a US-based supplement brand with seven SKUs across cognitive performance, sleep, and joint support. Annualized revenue was $14M when we started, with 62% from subscription. The brand was on a Shopify 2.0 theme they had outgrown, load times pushed past 4.8 seconds on mobile PDP, the bundle UX hadn't shipped in 14 months, and the subscription upsell ran on three different apps that had all been updated independently of each other.
The internal CFO had built the case for a headless rebuild but couldn't get past two questions from the founder: was the cost actually going to pay back, and would the migration kill their organic traffic during the launch window?
What we found
Two-week audit. Two surprising findings:
- Their three best-converting product pages had never been A/B tested. They had been the "good enough" templates from 2022 and the team had been afraid to touch them.
- They were leaving roughly 11% of monthly revenue on a bundle they had built but never properly merchandised. A "starter stack" of three products at a $189 price point existed in the catalog but appeared on a single secondary collection page.
The performance ceiling on the existing theme was the binding constraint. Even if we ran 50 successful CRO tests, the underlying theme couldn't ship the page templates we needed for them.
Approach
We scoped a 10-week rebuild on Shopify Hydrogen with three explicit goals: hold organic traffic during cutover, lift PDP conversion 25%+ on the rebuild alone, and ship the bundle UX as a first-class surface, not a follow-on engagement.
Audit + IA + redirect mapping
Pulled every URL with 5+ monthly organic visits over the prior 12 months. Wrote the redirect map. Tested it against staging twice before launch.
Design, PDP, PLP, cart, bundle builder
Six page templates designed in Figma. We shipped one direction, not three. The founder reviewed; we did one round of revisions.
Build
Hydrogen front-end on Shopify Storefront API. Custom bundle builder built as a Shopify Function. Klaviyo flows reconfigured during the build, not after.
QA + load testing
Cross-browser, cross-device, cross-checkout. Synthetic load testing at 3× peak traffic. Three Klaviyo cohort dry-runs to verify subscriber data integrity.
Cutover + 14-day stabilization
DNS flip during the brand's lowest-traffic Tuesday window. We watched Search Console, GA4, and Stripe dashboards in shifts for 72 hours.
Execution
The two trickiest engineering pieces were the bundle builder and the subscription migration. The bundle builder needed to feel native (not modal) on PDP and respect the brand's subscription savings logic across mixed cart compositions. We built it as a Shopify Function, clean integration with the native cart, no third-party app, and shipped a six-step build flow that A/B-tested against the existing "buy more, save more" PDP layout from week 5.
For subscriptions, the previous store ran subscriptions on Recharge with custom liquid that had drifted over two years. We migrated 27,000+ active subscribers with their plans, discounts, and next-charge dates intact. Two subscribers got duplicate-charged in the 24-hour cutover window, both were refunded within four hours. That was the only post-migration support burden the brand carried.
4.8s
Old Shopify 2.0 theme, image-heavy variant swatches
1.6s
Hydrogen build with preloaded variant images, deferred third-party scripts
The bundle UX rebuild deserves its own callout. On the old theme, the starter-stack bundle lived on a forgotten collection page and accounted for 2.3% of monthly revenue. On the new build, it became a PDP-native experience accessed from every product page, the cart, and a dedicated /build-your-stack URL. Bundle revenue climbed from 2.3% to 14% of monthly revenue inside the first 8 weeks post-launch.
Results
(See the full headline-results grid at the top of this page.)
The organic traffic number is the one the founder cared about most. Migration losses are the standard risk on a replatform, most brands take a 6-12% organic dip for 3-6 months. Northwind ended the 90-day post-launch window 4% ahead of pre-launch trend, which we attribute to the redirect map and the improved Core Web Vitals (Google rolled out a Page Experience signal weight increase 6 weeks after launch).
What we learned
Two operating insights came out of this build that we now run on every supplement engagement:
- Bundles are an entire merchandising surface, not a feature. Every supplement brand we audit has a bundle product hiding in the catalog. The brands that treat it as a first-class PDP experience compound 2-4× faster than brands that treat it as a collection-page footnote.
- Headless is justified by specific UX requirements, not by performance. The performance lift on Hydrogen mattered for SEO, but the conversion lift on Northwind came almost entirely from UX patterns the old theme couldn't ship. If the requirement is "faster site," Shopify Plus would have been the cheaper answer.
The boring engineering decisions, the redirect map, the subscription migration, the cutover Tuesday, paid for the entire project. The flashy stuff was secondary.