KAAL · Website Redesign

KAAL — custom front-end rebuild design (Shopify backend)

2026-07-08. The concrete "how we'd actually build it" plan. Extends 2026-07-08-kaal-north-star-commerce-build.md with stack, phases, feature designs, feasibility.

Could we do it? — Yes, realistically.

Not a weekend, not a moonshot. With Omar (dev) + Claude (AI-assisted build) + Sam directing, a modern Shopify-headless storefront is very achievable incrementally. The risk isn't capability — it's scope discipline (don't big-bang) and owning more maintenance. We keep Shopify checkout (Shop Pay), so we never touch the hardest, highest-risk part.

Approach — strangler-fig, PDP-first, parallel to Dawn

Never a big-bang cutover. Stand the new front end up beside Dawn and migrate page types one at a time, proving each converts before the next (earn-the-right-to-scale):

Phase 0 — Data spine (start on current Shopify): fit/size customer profile + product intelligence. The foundation.
Phase 1 — PDP (highest traffic × leverage): rebuild the product page headless, A/B vs Dawn PDP. Ship the fit engine, real bundles, reviews-at-price, source-aware hero, sub-1s load.
Phase 2 — Collections + search + home.
Phase 3 — Account / retention (fit passport, reorder, membership, back-in-stock).
Phase 4 — Content / SEO / gifting acquisition surfaces.
Checkout stays Shopify throughout. Every phase reversible (route traffic back to Dawn).

Stack (recommendation)

Front end: Next.js + Vercel + Shopify Storefront API. Why over Hydrogen: content/SEO acquisition is a core KAAL goal (Next excels), marketing-page velocity, strongest AI-assisted-dev path, and it fits Sam's existing stack (Supabase, Vercel/Cloudflare). Hydrogen/Oxygen is the Shopify-purist alternative if we want tightest native integration.
Auth/accounts: Shopify Customer Account API.
Data layer (fit profile, personalization, events): Supabase (already in the stack) + Shopify customer metafields.
Reviews: Judge.me via API (headless), rendered natively.
Search/merch: Shopify Search & Discovery API (or a headless search later).
Checkout: Shopify + Shop Pay. Untouched.
AI layer: Barton OS / Claude for fit recommendation, concierge, merchandising, content.

What it would look like — the flagship ideas (KAAL-specific)

Fit Passport (the killer feature). 60-second one-time fit quiz → KAAL remembers her size forever. Every PDP then shows "Your size: M — 3 left" pre-selected. Solves the #1 lingerie objection (fit fear + your no-refund policy) AND drives retention (she returns because KAAL knows her) AND builds a proprietary fit-data moat.
Gifting Mode. A toggle that transforms the experience — gift message, gift-ready/discreet reassurance, occasion reminders, "shop his & hers." Because gifting IS your demand engine (Gift Box in every co-purchase pair, 57% multi-product orders).
Source-aware landing. Cold Facebook-feed traffic lands on a trust + offer-clarity-heavy hero; Instagram/warm gets a fast path; SEO gets content-rich. Tailored per UTM/referrer.
Real bundles / Complete-the-Set. One-tap set at an honest price, correct component inventory, no TAKE30 trap. First-class, not a variant hack.
Honest live scarcity + back-in-stock loop. Real per-variant stock ("only 2 left in your size"); turn your 27% OOS problem into a demand-capture waitlist/drops engine.
Reorder & replenishment. Lingerie wears out — "reorder your Black Pearl" + gentle replenishment nudges.
Inner Circle membership — service, not points (no discounts, on-brand): early access to drops/restocks, saved fit, faster checkout, concierge fit help.
AI fit concierge on the PDP (leverage the already-built Roxy chat widget + Barton OS).
Post-purchase fit feedback loop: after delivery, "how did it fit?" → feeds true-to-size → improves everyone's recommendations. Compounding data moat + fewer exchanges.

How it addresses the CRO problems we actually found

Diagnosed problem	Rebuild solution
Fit fear + exchange-only/no-refund (cold-buyer killer)	Fit Passport + fit engine + true-to-size at the selector
Split/opaque sets that oversell	Real bundles, component inventory
Weak trust for cold high-ticket	Reviews/UGC + guarantees at the decision point
Gifting is the demand but there's no gifting UX	Gifting Mode
27% SKUs out of stock = lost demand	Back-in-stock / waitlist / drops loop
Retention (email 26%, no account experience)	Fit Passport account, reorder, membership
90% Meta dependency (single point of failure)	SEO/content/gifting acquisition front end
Theme app-bloat (~3.2MB JS)	Custom front end, only the JS you need

Honest limits (what a rebuild does NOT fix)

Traffic quality. Facebook feed's 0.6% ATC is a Meta targeting/intent problem. A better page converts that traffic only marginally — the real fix is the IG-weighted ad-set test. The rebuild's CRO ceiling on cheap traffic is capped by intent.
Checkout speed. Shopify's checkout-web (~926KB) is unavoidable; the rebuild wins everything up to checkout, not checkout itself.
So the rebuild is justified by personalization + retention + data moat + acquisition + up-to-checkout speed — NOT by "it fixes the blended conversion rate" (which is mostly a traffic-mix artifact).

Cost / risk / what I'd need

Team: Omar + Claude build it; Sam directs; Roxy owns the brand/content surfaces.
Rough magnitude (AI-assisted): PDP prototype in ~1-2 weeks; incremental migration over ~2-4 months; fully parallel to Dawn, reversible at every step.
Main risks: maintenance ownership (you own more than a theme), scope creep (mitigated by phase gates), and not letting the shiny rebuild distract from the bigger non-front-end lever (traffic allocation).

First move

Stand up a headless PDP prototype for Black Pearl on Vercel (Storefront API) with the Fit Passport v1, real bundle, and reviews-at-price — then A/B it against the Dawn PDP on live traffic. Proves the thesis on the highest-leverage page, one reversible step, before committing to the migration.