Web Director & UX Designer
The Mysterious Package Company.
I rebuilt the store on Shopify and ran weekly A/B tests on pages doing $30–50K a day. The tests settled copy calls by revenue, not opinion. “The coolest thing you can get in your mailbox,” says Neil Patrick Harris.
Watch the short version · 55 seconds
Rebuild on Shopify instead of patching five years of custom code.
- Outcome
- Made weekly A/B testing the habit; the best homepage variant showed +18.8% across 70K+ sessions
- Landing pages generating $30–50K daily in direct sales
- Design decision time cut from 1–2 weeks to 1–2 days
- Dramatically reduced fulfillment times through automation
- Problem
- A five-year-old custom eCommerce build was clunky for customers and brutal to maintain, with manual fulfillment eating the team’s week.
- My role
- Directed all aspects of UI and web development
- User testing via A/B split testing
- Rebuilt the flagship eCommerce store on Shopify
- Custom fulfillment automation inside the new site
- Key calls
- Put fulfillment automation inside the store itself
- Weekly A/B tests over big-bang redesigns
The rebuild.
MPC had spent five years on a custom build that was clunky for customers and brutal to maintain. I rebuilt the flagship store on Shopify: mapped the old front-end and back-end flows, cut what did not earn its place, and launched a storefront where the story comes first and the path to purchase is obvious. (The rebuilt storefront plays up top.)
Wireframe to design system.
Wireframes to a full design system to prototypes. One system in the brand’s antique-occult voice carried the homepage, catalogue, membership flow, and checkout, across desktop and mobile.
Fulfillment, automated.
The team was losing a week to manual fulfillment. I mapped the entire member journey, from public homepage through membership approval, experience requisition, checkout, and the fulfillment workflow, then built automation into the store itself so orders moved without hand-holding.
Conversion as a weekly habit.
I treated conversion as a habit, not a project: a fresh A/B test most weeks, read the result, ship the winner, repeat. These are the actual readouts. The homepage copy test showed a +18.8% relative lift across more than 70,000 sessions, and a button test showed +31.6%. I’ll be straight about the method: I called both early once a variant was clearly ahead, so read them as strong directional wins from a steady testing practice, not lab-grade certainty. The discipline is the point.
Hand-coded, before AI.
I built the experience landing pages by hand in 2017, years before AI could write a line of them. No frameworks, no page builders, no generators, just HTML, CSS, and hand-animated SVG. Nearly a decade later they still hold up, and that is the point: get the fundamentals right and the work does not date. These are the real pages, rehosted and running right here, open one and look around. Also live: the members’ sub-page and a self-drawing SVG icon.
Want the story behind this one? Ask the AI me anything about it.
