Taar
Fashion e-commerce discovery — fewer taps to narrow the catalogue
The problem
Fashion discovery usually makes the shopper do the work — filters buried screens deep, results that don’t move until you commit, sold-out sizes that only reveal themselves after you’ve picked one. Taar flips that: narrowing happens in one place, the count updates live, and availability is honest before you commit to anything.
Taar is a mobile-first product discovery UI for a fictional fashion label — one Autumn Edit of ten pieces. It isn’t a deck or a set of static mockups; it’s a working React build, so every decision in it has to survive contact with real state: filters that actually filter, sizes that actually sell out, a grid that can actually end up empty.
Design rationale
Narrowing happens in one place, and the count updates live behind the sheet.
Because results that don’t move until you commit make the shopper do the work.
Availability shown up front answers the question once, before commitment.
Because sold-out sizes that only reveal themselves after you’ve picked one reopen it on every product.
Quick add from the grid shortens the path to the bag.
Because a product-page detour for every add is the longest possible loop.
Flow
- Browse the Autumn Edit — ten pieces under category tabs.
- Open Filter — one bottom sheet: size, colour, price.
- Results update live behind the sheet; the CTA carries the count — “Show 4 pieces.”
- Active filters stay visible as removable chips under the bar.
- Quick add from the grid — colour, size, done, no page change.
- The bag counter confirms the add without ever leaving the flow.
Filtering is one bottom sheet — size, colour, price — not a stack of pages. Closing the sheet never costs a surprise.
Wireframes
Try it
Open Filter, add a size.
Watch active filters stack up as removable chips under the bar.
Design decisions
- One sheet, live count on the CTA
- Results update live behind the sheet, and the CTA carries the count: “Show 4 pieces.” Closing the sheet never costs a surprise — no “apply and pray.”
- Filters stay visible as chips
- Active filters don’t disappear back into the sheet — they stay visible as removable chips under the bar, so the grid’s state is always readable at a glance.
- Sold-out sizes are shown and struck through, not hidden
- Hiding them reopens the question on every product; showing them answers it once — and explains why a piece can’t be added in your size before you try.
- Quick add from the grid
- Tap plus, pick a colour and a size in the sheet, done — no product-page detour. Sold-out sizes are disabled right there, and the bag counter confirms the add without ever leaving the flow.
The prototype respects the boring rules — the ones that decide whether a design actually ships:
- 44px minimum touch targets everywhere, including the quick-add button on each card.
- The sheet is a real dialog: Esc closes it, the scrim closes it, background is inert and page scroll is locked.
- Pressed states are announced via aria-pressed; sold-out sizes read as “M, sold out”.
- prefers-reduced-motion turns off the entrance and sheet animation.
Visual system
Editorial, not app-template: Fraunces for display type, Hanken Grotesk for UI, a warm-paper palette with a single navy accent. Product images sit in tonal slots, and the grid reveals with a quiet staggered entrance.
- Paper #F6F3EESurface
- Ink #1B1D22Text
- Mute #6B685FSecondary text
- Line #DDD7CCDividers
- Taar navy #1F2A44Accent
The tonal slots are deliberate — the layout is designed around art-directed photography that hasn’t been shot yet, so the prototype doesn’t fake it with stock images.
Build & next
The React build is the study’s evidence — filters that actually filter, sizes that actually sell out, a grid that can actually end up empty. It runs live on this site: the embed on this page is the real app, not a video of it.
- Figma file backing these screens: research, flows, wireframes and hi-fi.
- Product detail and bag screens to complete the purchase loop.
- Real photography to replace the tonal slots.
- No metrics yet — the original app hasn’t been measured, so no tap-count or conversion claims until it is.