I designed a fashion-tech startup’s brand dashboard from a blank page, then redesigned the shopper-facing 3D fitting room. Brands now upload garments without developers; shoppers see a recommended size for every item.
0 → 1
brand dashboard, designed from a written spec
140+
screens and states, desktop and mobile
3 / 3
milestones approved; scope grew from one product to two

fig. 1
The dressing room after the redesign: the mannequin stays the hero, and every item carries its own size.



01 · context
A technical 3D product, run by people who aren’t technical
Shoppers could already try clothes on a 3D mannequin. Brands had no way to get their garments in: no upload, no setup, no billing.

fig. 2
Before: size in one window only, tabs that behaved like a hidden wizard, download locked behind a face upload.
The brief
“Very clean and professional, yet still looking somewhat fashionable… very simple to use even for non-technical people.” — the founder
Our question: how might we make 3D try-on simple for a merchandiser, and make the size decision obvious without turning a luxury fitting room into an admin tool?
Goal
A shopper knows which size to buy for every item, and brands run the tool without developers.
Why it was complex
→ brands upload .glb 3D files, but the user is a merchandiser, not a 3D artist
→ every brand setting changes what shoppers see: two products, one data model
→ plans, quotas, add-ons and 4 payment methods had to feel fair, not like paywalls
→ no PM, no research budget, and gaps in the spec to turn into decisions
02 · my role
Sole designer, from a written spec to developer handoff
I was the only designer, reporting directly to the founder, with no PM or design lead. Product decisions, UX and UI were mine to make, explain and defend, across four design sprints and three milestones.
03 · the dashboard
A self-serve admin a merchandiser can run alone
One test for every screen: can a merchandiser get a garment live without asking a developer?
Key decisions
Hub · four KPIs and a quota bar that warns before the limit, not after
Add product · a wizard to create, one page to edit
Defaults · set pose and backdrop once for the brand, override per item
Billing · upgrade, downgrade and payment failure, each with a recovery state

fig. 3
Add product: a three-step wizard with progress, from a .glb upload to a styled default look.
01 Basic info
name, category and .glb file
02 Category rule
9 types, mapped to the body
03 Default look
pairing or neutral underwear
04 Images
generated from the 3D model
05 Product page
embed, analytics, defaults
04 · the fitting room
Five skippable steps, and a size next to every item
Same features, new order. The mannequin stays the hero: hair colour comes first so the 10 styles update to match, and on mobile the panel becomes a bottom sheet.
The shift
One size decision per item, visible at every step.
Before → after
Navigation · tabs that acted like a hidden wizard → a 5-step stepper, every step skippable
Size · one window, one global size → a size badge on every item, plus “leans towards L”
Look · unclear what’s on the mannequin → a look bar with Buy and Remove on every item
Export · download locked behind a face upload → with my face, or as is

fig. 4
Mobile: the panel becomes a bottom sheet, and the look bar keeps Buy and Remove one tap away.
05 · collaboration
Founder feedback as research, without losing the product
With no shopper access before launch, the founder’s feedback was the closest thing to research. Each note became a rule, not a patch.
The rule
Kept the features. Cut the noise.
Founder note → what I designed
“…so we know which part of the mannequin it goes on” → a 9-category taxonomy with pairing rules
“default the pose/backdrop for each item separately” → a two-level defaults model
“where it takes the user after clicking Manage plan” → upgrade, downgrade and payment-failure flows
“a bit too crowded and distracting” → the same features, behind a stepper


fig. 5
Fitting room v1 → v2: the same features, reorganised behind a stepper so each step shows only what it needs.
06 · results
2
products on one design system
9
garment categories with pairing rules
Day 1
first milestone approved
Every milestone approved, the first on the day it was sent, and the engagement grew from the dashboard to the shopper side. “Design-wise perfect!” — the founder
Restraint is a design decision · a wizard to create, one page to edit · design the rules, not only the screens.
next: measure after launch · completion to export · share rate · size-tooltip use
© 2026 Maria Sidorova, Amsterdam
