Yanmar was running their B2B commerce operations on a separate third-party platform alongside their consumer SFCC store — duplicate catalogues, drifted pricing, two separate order pipelines, and no unified view of customer.

Consolidate B2B operations onto the same SFCC platform powering the consumer store, while preserving dealer-tier pricing and net-payment-terms behaviour.

Built a custom B2B layer on SFCC — dealer-tier pricing via metafields, net-terms checkout, role-gated catalogues, and a dedicated B2B admin surface. Migrated dealer accounts in three waves, preserving order history.

The dealer base was wary — change is risk when your job depends on a working order flow. A six-week dealer-side beta with two volunteer accounts surfaced the friction points before the wide rollout.

Next quarter ships dealer-facing forecasting tools — historical order patterns plus suggested re-order quantities by SKU velocity.

Yanmar operated three separate login portals — the Yanmar Owners Community, the B2B Dealer Portal, and the legacy parts ordering site. Each had its own credentials, its own UI, and its own support burden.

Consolidate three login experiences into one unified identity flow without disrupting the existing user bases or breaking deep links from email and dealer documentation.

Mapped every existing entry point and user journey. Designed a single login UI that detects role on first sign-in and routes to the correct portal. Wired identity federation, migrated existing credentials, and built a one-time 'merge your accounts' wizard for users with overlapping records.

Account merging is mostly a comms problem, not a technical one. Three rounds of email walkthroughs and a recorded video kept the rollout clean.

Phase two adds passwordless and an authenticator-app option for dealer accounts handling order placement.

Yanmar Commerce had a checkout funnel with meaningful drop-off, a financing flow that pushed users into a separate flow with no context, and a homepage that was getting traffic but not converting.

Diagnose the drop-off, prioritize the highest-leverage fixes, and ship a measurable lift in conversion within one quarter.

Instrumented the full funnel with GA4 + Hotjar. Identified three friction points — a financing handoff with no return path, a shipping calculator that fired late, and a homepage hero that buried the primary action. Shipped fixes in three two-week iterations with A/B validation on each.

Instrumentation came before opinion — every fix had a measured impact before it became permanent. Two of the three hypotheses I started with turned out to be wrong, and the data caught it.

Layering AI-driven personalization on the homepage hero next — content varies by inferred buyer intent.

Visitors to Commerce Ace were browsing service pages without a clear nudge toward taking action — creating a gap between demonstrated interest and conversion that generic site-wide messaging couldn't effectively close.

Design and build a behavioral personalization engine that detects intent based on time spent on specific service pages and dynamically serves a personalized banner on the homepage to move high-intent visitors toward contact.

Engineered a session-based tracking system that monitors visitor behavior across three key service pages. When a visitor spends more than 15 seconds on any of these pages and then navigates to the homepage, the engine fires a personalized banner tailored to the service they engaged with — complete with a targeted headline, supporting message, and a direct CTA linking to the contact page.

The threshold logic was the most critical variable to get right — too short and the banner fires for casual browsers, too long and it misses genuinely interested visitors. Finding that balance required real session data and iterative tuning.

The natural next evolution is layering in true AI-driven intent scoring — moving beyond time-on-page as the sole signal and incorporating scroll depth, return visits, and page sequencing.

Commerce Ace needed a way to engage site visitors intelligently and in real time — answering questions about services, qualifying leads, and pointing the right prospect to the right next step without forcing them through a generic contact form.

Build a Claude-powered service assistant that lives on every page, knows the full Commerce Ace catalogue, and handles the first thirty seconds of every prospect conversation.

Trained Ace on every service page, case study, and pricing band on the site. Wired a Shopify-compatible chat widget into the site footer with a persistent conversation history per session. Built admin controls for the system prompt, off-topic guardrails, and a per-conversation transcript export for follow-up.

The hardest part wasn't the model — it was the system prompt. Getting Ace to refer out gracefully when a question genuinely needed Albert took a week of red-teaming.

Next iteration adds calendar-aware booking — Ace will propose three discovery-call slots inline instead of handing off to Calendly.

Maintaining a consistent, high-quality blog is one of the most time-intensive parts of running a solo consultancy — and skipping a month tanks organic traffic in a way that takes a quarter to recover.

Ship an embedded Shopify app that drafts brand-tuned blog posts on a schedule, pulls matching imagery from Unsplash, and publishes to the Shopify blog with one click of human approval.

Built the Blog Agent as a React Router app on Fly.io with a Neon Postgres backing store. Tuned a settings screen that captures voice, audience, and post cadence per shop. Wired Claude for drafting, Unsplash for imagery, and Shopify's Admin GraphQL for publishing.

Auto-publish was tempting but wrong. A two-second review loop catches voice drift before it ships, and that small friction is what keeps the output sounding like a person.

The roadmap adds topic-cluster awareness — the agent will plan four posts as a series rather than one-offs, then track downstream organic traffic per cluster.

WearSPF needed a fully custom e-commerce presence built from the ground up — one that matched their sun-care brand's confident, science-led positioning and supported a fast-growing product catalogue.

Design and build a Horizon-based Shopify theme, brand-tokenized end to end, with custom PDP, collection, and editorial sections that flex with the brand's content rhythm.

Built a token-driven theme on Horizon with SASS partials per section. Implemented variant-aware PDPs with stock-aware swatches, an editorial blog template, and a curated quiz that recommends SPF level by skin profile.

The variant matrix was wider than initial scope assumed. Tightening the data model in week one would have saved three days of rework in week five.

Next phase adds a subscribe-and-save flow with churn-aware messaging tied to remaining product life.

The Face Haus needed a custom Shopify store that could serve as both an e-commerce platform and an in-studio booking destination — two flows merchants typically split across two platforms.

Unify retail product discovery and service-appointment booking on one Shopify theme, without compromising either experience.

Designed a hybrid navigation that switches voice between 'shop the routine' and 'book the facial.' Integrated a third-party booking widget inline on service pages. Built a unified header cart that handles both physical product and appointment line items.

The biggest design risk was a navigation that felt like two sites stapled together. The breakthrough was making 'shop' and 'book' the same primary action, separated by inventory type rather than by header.

Adding a returning-client portal next — booking history, product re-order, and routine reminders in one place.

DCL Skincare needed a custom Shopify store that could reflect the sophistication of their brand while supporting clinical-grade product education at PDP depth.

Build an editorial-grade Shopify theme that lets DCL ship long-form ingredient education and case-study content without leaving the storefront.

Designed PDPs with progressive disclosure — quick-shop above the fold, full ingredient pharmacology below. Built a structured 'before / after' content type with consent-gated client imagery. Layered a routine-builder widget that recommends three-product regimens.

Content density nearly tipped the page weight past our performance budget. Splitting the ingredient pharmacology into a lazy-loaded module saved the Lighthouse score.

Next: an AI-driven routine builder that personalizes the three-product recommendation based on a short skin questionnaire.

Callaway Golf was launching their 2026 Chrome Tour golf ball line and needed a high-impact marketing page on Salesforce Commerce Cloud — built fast, branded perfectly, and instrumented for the launch-day traffic spike.

Design and ship a Chrome Tour campaign page that holds attention through a 12-minute average session, drives PDP traffic, and supports launch-week paid spend at scale.

Designed a long-scroll campaign page with tech-spec disclosures, side-by-side spin/distance comparisons, and a tour-pro testimonial section. Built it as a Salesforce Commerce Cloud Page Designer template. Pre-cached imagery and lazy-loaded the spec deep dives to hit a sub-1.5s LCP.

Page Designer rewards aggressive componentization. The 22 sections distilled to seven reusable components — which is what made the next ball release feel like 'duplicate this' instead of 'build that again.'

Next ball release will use the same scaffold with a dynamic tour-pro testimonial rotation driven by Marketing Cloud audience data.

Callaway was entering the pickleball market for the first time with their Inertia Pickleball Paddle line — a category-defining launch for the brand with no in-platform precedent.

Build an entry-category launch experience on Salesforce Commerce Cloud that positions Callaway as serious about pickleball while feeling distinct from the golf catalogue.

Designed a paddle-line landing with category-first navigation (skill level, paddle weight, sweet-spot shape), built a comparison tool for the three Inertia models, and laid in editorial content explaining the engineering behind the paddle's face technology.

Calling it pickleball-not-golf was the right move. Treating the Inertia line as a sub-collection of the Callaway main catalogue would have buried it.

Phase two adds a 'find your paddle' quiz tied to a personalization layer that remembers grip size and playstyle across sessions.

CallawayGolf.com required ongoing homepage refreshes, content updates, and personalization management — at the pace of weekly marketing campaigns, across two seasonal cycles per year.

Become the embedded operating partner for CallawayGolf.com's homepage — handling weekly content swaps, campaign rollouts, and personalization logic without slowing the marketing team's velocity.

Established a repeatable weekly cadence — Tuesday content drop, Wednesday build, Thursday QA, Friday push. Built a Page Designer template library so 80% of weekly changes happen by content edit, not by code.

The first-quarter cadence was too aggressive. Cutting back to one push per week with a strict Thursday QA window fixed it.

Q3 brings in an AI-driven hero copy tester — Einstein chooses the variant, but Claude proposes the copy.