An editorial store for clothing and the home. A storefront, a shopping app and a seller back office, all working prototypes, covering the flows that make the money: product page, checkout and returns.
Maison sells clothing and homeware made in small batches in Portugal. Its old store looked like every other template: sale banners, crowded grids and a checkout that felt rushed. Customers loved the products but rarely bought more than one.
The brief: an online store that feels like the brand’s printed lookbook, a shopping app for returning customers, and a back office the small team actually enjoys using.
Baseline figures are assumptions written into the brief for a fictional client, to validate in discovery. They are not measured data.
Buys a few good things a year, mostly on her phone. Wants to see how pieces live together, checks the returns policy before a big purchase, and hates being rushed.
Runs the studio with two people. Packs orders herself on busy days, handles every return by hand and knows every workshop by name.
“I don’t want to buy a vase. I want to buy the evening it belongs to.”
Proto-personas built from desk research and the benchmark, not from interviews.
An editorial back office: a black masthead sidebar, serif page titles and magazine-like grids for orders and products. It is built for a team of three who pack orders between other work, so the next action on every order is one button.
Most fashion apps shout: banners, countdowns, pop-ups. Maison’s app is quiet on purpose, and it is built for returning customers: saved pieces, restock alerts, orders and returns. The website does discovery for new visitors; the app keeps the ones who come back.
The storefront reads like a magazine: a full-bleed cover, a shoppable room, new arrivals, a workshop story, plain delivery and returns terms, and a journal. The bag slides in from the side so you never lose your place, and checkout happens in the same panel.
The product page is where a €390 coat is won or lost. Four details answer the questions people have before they add to bag.
Rejected: A size chart in a modal as the only fit help. Most people never open it. One line of advice next to the sizes does more, with the full chart one tap away in v2.
The old checkout felt rushed. The new one has three short steps inside the bag panel, so you never leave the store, and the total, delivery date and returns promise stay visible the whole way.
Rejected: A single-page checkout. It looks shorter, but on a phone it becomes one long form where errors hide off screen. Three short steps keep each error next to its field.
| Hypothesis: carts are lost when | What changed | What we’d measure |
|---|---|---|
| Delivery cost appears first at checkout | Delivery cost and the free-delivery line live in the bag | Bag to checkout start rate |
| Account wall before paying | Guest by default, express wallets on top | Checkout start to details complete |
| Card errors shown after a page reload | Inline error next to the field, in words | Payment error rate, time on the Pay step |
| Wrong size, then a refund | Fit note on the page, exchange before refund | Return rate by reason, share of exchanges |
For clothing, returns are the margin story. The policy is written once and shown in the same words on the product page, in the bag, at checkout and in the footer. Returns themselves are self-serve in the app and land in the back office with a reason code.
“Exchange comes before refund, because a size swap keeps the sale.”
Rejected: Printable return labels. Cheaper for the studio, but people without a printer stall and the return becomes a chore. Courier pick-up is part of the premium promise.
Serif headlines, generous white space and no countdown timers. The brand sells slow, considered pieces, so the interface should feel the same. The newsletter offers early access to restocks instead of a discount.
Rejected: A 10% pop-up on first visit and a sale countdown banner. Both lift sign-ups in the short term, and both teach customers to wait for a code.
Customers buy into a mood, not a SKU. Hotspots on a styled room let them add one piece or the whole corner, and the room skips sold-out pieces and says so. We expect room orders to be larger; the pilot target is 2× to 3× the items of a single-item order, measured by average order value on orders that include a room add.
Rejected: A “Complete the look” carousel under each product. Cheap to build and common, but it shows pieces out of context and is easy to scroll past.
A thin progress line in the bag shows how far you are from free delivery. The €200 threshold sits just above the assumed €192 average order, so most bags are one small piece away. It nudges a second piece without a pop-up.
Rejected: An upsell pop-up at checkout. It interrupts people at the moment they are most ready to pay.
The team lives in the admin all day. Serif page titles and calm tables make order work feel less like a spreadsheet, and packing, shipping and returns happen from one drawer.
Rejected: Working only inside the default platform admin. It does the job, but the brief’s problem was four tools and many clicks to pack one order.
Ivory paper, ink black and one clay accent borrowed from the ceramics. Cormorant Garamond for headlines and prices, Jost for everything you read quickly. Every colour has one job.
Maison is designed to be buildable on Shopify Online Store 2.0 sections, so the studio can change the store without a developer. Each storefront block maps to a theme section:
| Storefront block | Theme section |
|---|---|
| Cover | Image banner |
| Shop the room | Custom section, one block per hotspot product |
| New arrivals | Featured collection with quick add |
| Our story | Image with text |
| Delivery and returns | Multicolumn, text from one policy page |
| Journal | Blog posts |
| The Maison letter | Newsletter |
| Product page | Product template: variant picker, inventory note, delivery date, collapsible tabs |
Checkout stays inside what the platform allows: express wallets on top, delivery dates from shipping settings and branding through checkout settings. The back office is the part that is not stock: it would be a custom admin app reading orders, stock and returns through the platform’s Admin API. The shopping app would use the Storefront API.
| Metric | Baseline (assumed) | Target | How measured | Guardrail |
|---|---|---|---|---|
| Items per order | 1.1 | 2.1 | Order data | Return rate must not rise |
| Abandoned carts | 68% | 31% fewer | Checkout funnel events | Payment error rate stays flat |
| Orders with a room add | New | 2× to 3× items of single-item orders | Orders tagged by room add | Returns on room orders |
| Returns kept as exchanges | To measure | 1 in 3 | Return reason codes | Refunds still within 5 days |
| Tools to run orders and stock | 4 | 1 | Studio team, week 1 vs week 6 | Time to pack an order |
Target outcomes, not results. No pilot was run. These are the numbers I would agree with the studio team for a 6-week pilot.
If I did this again, I would design the product page and checkout before the room. The room is the memorable part, but a premium brand loses trust fastest at the moment of paying, and those screens decide whether the room ever gets used.
The hardest part was staying quiet while still nudging a second piece. I kept exactly one nudge, the free-delivery line, and cut the rest. The photos are stand-in sample imagery, so a real launch still needs an art-directed shoot with a clear crop and colour brief.
What I would test first with real people: whether “Add the whole room” feels helpful or pushy. Five unmoderated sessions on phones, counting how many room items get removed from the bag.