Enterprise engagement · Commerce · Omnichannel
ShopSense AI — Shop smarter, personalised by AI
An intelligent omnichannel shopping platform designed to remove friction from discovery, decision and checkout
- Conversion rate
- +31%
- Average order value
- +22%
- Cart abandonment
- −34%
- Repeat purchase
- +18%
Overview
The shopping experience had grown faster than its design could keep up with: thousands of near-identical products, unreliable reviews, inconsistent sizing and a checkout that punished mobile users — a 71% cart abandonment rate and declining repeat purchases from premium members.
The ask: design an AI-powered experience that reduces friction at every step and rebuilds customer trust.
Research
Listening before designing
18 in-depth interviews, 420 survey respondents, 2,000 session replays and six competitor teardowns, distilled into four personas: Ananya the first-time shopper, Rohan the returning customer, Priya the premium member and Marcus in customer support.
- Overwhelming catalogue — too many similar products
- Reviews are noisy and hard to trust
- Sizing is inconsistent across brands
- Checkout is long, especially on mobile
- Order status is unclear after purchase
- Returns feel like a punishment
Opportunity map
Turning insight into design bets
The journey — discover, explore, compare, decide, purchase, post-purchase — was mapped against six opportunities: AI discovery, conversational shopping, visual search, one-tap checkout, live tracking and rewards.
IA & search
Making everything findable
Navigation was restructured around intent — Discover, Shop, Assistant, Orders, Rewards — collapsing 42 top-level categories into eight thumb-friendly entry points.
Semantic search interprets natural queries such as “linen dress under 3000 for a summer wedding” and returns filtered results, AI suggestions and complete-the-look bundles.
AI features
Where intelligence meets craft
A shopping assistant grounded in the live catalogue answers questions, compares items and builds looks. Visual search turns the camera into a query with confidence-scored matches. Recommendation carousels are explainable — every card says why it is there. Comparison summarises AI-generated pros, cons and best-for-you callouts.
Checkout & post-purchase
Removing friction end to end
Checkout was cut from six steps to two with saved addresses, wallet support and inline validation. Live tracking adds a real-time map, ETA and proactive delay alerts with an inline assistant. Self-serve returns use QR pickup, an instant refund preview and reason tags that feed back into the catalogue.
Design system
Sense DS — consistency, accessibility, speed
Tokens, 60+ components, motion primitives and shopping-specific patterns for price, rating, stock and tier — fully themable and WCAG 2.2 AA compliant: 4.5:1 minimum contrast, 44px touch targets, focus-visible states, full keyboard support and reduced-motion variants for every animation.
Outcomes
Craft, evidence and impact
Across five rounds of moderated testing task success moved from 62% to 94% and SUS from 68 to 86. Sixty days post-launch: +31% conversion, +22% AOV, −34% cart abandonment and +18% repeat purchase.
Next: AR try-on, deeper personalisation for premium members, group shopping sessions and multilingual assistant coverage.
Next case study
LabSync AI — One diagnostic platform, three roles
Healthcare · DiagnosticsContact
Let’s talk about the problem, not just the solution.
I’m open to senior product design, UX leadership, design systems, and AI-native product opportunities.
Get in touch