A brand reset, a Shopify rebuild,
and a conversion-ready foundation —
shipped in eight weeks.

Built templates/product.nipple-relief-covers.json as the brand's new front door. Eleven purpose-built sections in deliberate order: buy zone first; then why-choose, how-it-works, comparison table, expert quote, lifestyle banner, badge ticker, FAQ, reviews, you-may-also-like, product videos.

Section 1 carries the conversion. Section 2 onward does the convincing for the undecided. A new highlight block (text + optional icon, rendered between rating and price) gives marketing a controllable above-the-fold message — reusable on any PDP, not locked to one product.

Built dedicated landing pages: Emotional angle, Problem angle, Trust angle. So paid traffic from different ad angles lands on a page that matches the angle, rather than the homepage.

All three live.

Each is independently testable. The agency can now isolate which angle wins per audience and per market without redeploying anything.

Replaced the single-threshold shipping bar with a two-tier bar. Customizer-driven thresholds and colors, fully translatable, feature-flagged so the single-tier fallback still works if disabled.

This directly targets AOV. Two-tier mechanics give the merchant an explicit lever to test thresholds without re-deploying — change a number in admin, watch the AOV move.

Custom upsell widget inside the drawer, plus a rebuilt cart footer with savings line, trust badges, and payment icons consolidated in one place. The drawer re-renders on cart:refresh using Shopify's section rendering API so all of this stays accurate without JS state drift.

The old standalone cart-page redirect was removed entirely. One drawer, end to end. Less code, fewer bugs, faster checkout entry.

Rebuilt the global product card. Single-variant items get a true AJAX add-to-cart via Focal's product-form custom element; multi-variant items link to the PDP.

The old card had a hidden second form that bypassed variant selection — a silent cart-error generator that had been quietly costing conversions. Quick-buy drawer and popover paths kept intact. Fix is global: applies to every collection page, every grid, every recommendation widget across the site.

Extracted the per-template delivery_info and trust_badge blocks into shop metafields and a Trust-badge metaobject. The merchant edits trust content once in admin and it propagates everywhere — instead of forking 20 templates.

This is the kind of work that doesn't show up in a homepage screenshot but compounds every month. Future content updates that used to take a full developer pass now take five minutes in the Shopify admin.

Codified the brand into brand-profile.json — documentary, lived-in, Filter Free Parenting voice. Explicitly not white-linen-mom. Explicitly not clinical.

Then built an idempotent photoshoot pipeline (scripts/run_photoshoot.py) that produces five contextual styles plus clean transparent plus shadow-cutout for each product. 29 products fully reshot at roughly $0.55 and five minutes per product. The catalog image set is consistent across every product page for the first time in the brand's history.

After publishing on May 5, RUM showed CLS P75 climbing from 0.02 toward 0.3 — a real-world regression we hadn't caught in pre-launch testing.

Diagnosed and shipped fixes within a week: eager-loaded the homepage hero, gave the badge ticker an explicit width (loading=eager, decoding=async — marquees can't lazy-load), set fetchpriority on the covers hero, and added an eager_load section setting to the lp-image-text-split block. Fixes live May 11–12.

Field-data recheck scheduled for May 18 — one full week after fixes shipped — to confirm trajectory back to the pre-launch baseline.