← All Projects

B2B E-Commerce Platform

Headless commerce platform built for a B2B client — handling complex pricing tiers, bulk ordering, and integration with legacy ERP systems.

Next.js.NET CoreAzureTypeScriptPostgreSQL
GitHub ↗

B2B e-commerce is a different beast from B2C. The pricing alone will make you question your career choices — customer-specific pricing, volume discounts, contract rates, regional variations, and all of it needs to sync with a 15-year-old ERP system.

The Problem

The client had been taking orders by phone and email. Their sales team was manually entering everything into SAP. As they grew, this stopped scaling. They needed a self-service ordering portal that their business customers could use directly.

What We Built

A headless commerce platform with:

  • Next.js storefront — server-rendered for SEO, with customer-specific pricing displayed after authentication
  • .NET Core API layer — handles business logic, pricing calculations, inventory checks
  • Azure Service Bus for async communication with the SAP ERP system
  • PostgreSQL for the commerce database, with Redis caching for price lookups
  • Azure Blob Storage for product catalogs and technical spec sheets (PDF)

The pricing engine was the hardest part. Each customer can have:

  • Base catalog prices
  • Contract-negotiated overrides
  • Volume discount tiers
  • Time-limited promotional pricing
  • Regional adjustments

We built a rules engine that evaluates all applicable pricing rules in priority order. It’s not glamorous code, but it’s correct, and that matters more than elegance when money is involved.

AI Addition (2025)

After the core platform stabilized, we added AI-powered features:

  • Smart search using Azure OpenAI embeddings — customers can search for products in natural language (“I need a heat-resistant gasket for 200°C applications”) instead of knowing exact part numbers
  • Order prediction — suggests reorder quantities based on historical purchasing patterns
  • Automated product descriptions generated from technical specifications

Results

  • 40% of orders now come through the self-service portal (up from 0%)
  • Average order processing time dropped from 2 days to 15 minutes
  • Sales team focuses on complex deals instead of routine order entry
  • Smart search reduced “product not found” support tickets by 60%

Timeline: Core platform built Q1-Q3 2024. AI features added Q1 2025.