UrbanCart Mobile App

The problem

UrbanCart was a regional e-commerce marketplace connecting consumers with local artisan and specialty retailers. Their web experience was working, but mobile conversion was 40% lower than web — a sign that customers were browsing on mobile and buying on desktop. They needed a proper mobile app, not a responsive website, to capture the mobile-first shopping behavior of their audience.

They also wanted AI-powered product recommendations — using purchase history and browsing behavior to surface relevant items and increase average order value. This needed to work across both iOS and Android without the cost of maintaining two separate native codebases.

What we built

A React Native app covering the full e-commerce experience: product discovery with AI-powered recommendations, category browsing, search with semantic understanding, product detail pages, shopping cart, Stripe checkout, and order tracking. Seller profiles and product ratings built in. Push notifications for order updates and promotions.

The AI recommendation engine uses collaborative filtering on purchase history combined with content-based filtering on product attributes. For new users, it falls back to popularity and category signals until enough purchase history accumulates. Average order value in the app 23% higher than web, driven primarily by the recommendations feature.

Single codebase, both platforms. Deployed simultaneously to App Store and Google Play at launch. Backend Node.js API, PostgreSQL database, Redis for cart and session caching, AWS infrastructure.

Technology

• React Native + TypeScript
• React Navigation (tab + stack navigation)
• Node.js + Express API backend
• PostgreSQL + Redis
• Stripe mobile SDK
• Recommendation engine in Python (collaborative + content filtering)
• AWS EC2 + RDS, Cloudfront CDN for images

Results

Launched to App Store and Google Play simultaneously. Within 90 days: 18,000 downloads, 4.6-star rating on both stores, mobile conversion rate within 8% of web (up from 40% below). Average order value 23% higher in app than web. The client extended the engagement for Phase 2: live shopping events and social sharing features.