Case study
Anina
Personalised multi-seller ecommerce marketplace.
A multi-seller marketplace with personalised recommendations, Stripe Connect payouts and automated DHL tracking.
The problem
Most ecommerce shows a generic feed, burying shoppers in options that don't match their style. Sellers get basic tools with no view of orders or payouts, and manual payment handling plus opaque delivery erodes trust at checkout.
My approach
Built a server-rendered recommendation engine over a multi-tenant Supabase architecture with strict data isolation, then automated the money and logistics paths end to end with Stripe Connect and the DHL API.
How it's built
Next.js App Router, server-rendering the recommendation surfaces. Supabase provides multi-tenant storage with row-level security per seller. Stripe Connect handles split payments and payouts; DHL webhooks drive tracking.
Key features
- Personality test driving product recommendations
- Multi-seller marketplace in one shopping experience
- Seller dashboard for products, orders and payouts
- Stripe payments with automated seller payouts
- DHL integration for automated delivery tracking
Challenges & Solutions
Multi-tenancy risked one seller's queries reaching another seller's data.
Enforced isolation with row-level security policies at the database rather than in application code, so a missed check in the app cannot leak data.
Manual payout and shipping enquiries dominated the support queue.
Automated both paths, using Stripe Connect for scheduled payouts and DHL webhooks for tracking, then surfaced status directly in the order view.
Results
- 40%increase in engagement from personalised recommendations.
- 70%fewer support tickets after automating checkout and shipping flows.
Implementation stack
Tech Stack
Frontend
- Next.js
- TypeScript
Styling
- TailwindCSS
Backend
- Supabase
Payments
- Stripe Connect
Logistics
- DHL API
Analytics
- PostHog