CommerceFlow

A multi-category marketplace connecting 120+ vendors with 45,000 monthly active buyers had outgrown its monolithic eCommerce backend. The platform started as a single-vendor store and was stretched into a marketplace through plugins and custom patches. Every vendor onboarding took 2 weeks of manual setup, and the system couldn't handle the transaction volume it was generating.

Vendor onboarding required manual database entries, FTP credentials for product uploads, and a 14-step checklist managed in a shared Google Doc. Average time from vendor signup to first product listed: 12 business days

Product catalog had no standardized schema. Each vendor uploaded data in their own format. 30% of product listings had missing images, incorrect categories, or broken descriptions

Inventory sync was a nightly batch job. Products sold out during the day still showed as available until midnight. Customer complaints about "ordered but out of stock" averaged 85 per week

Order routing to the correct vendor was semi-manual. A fulfillment coordinator reviewed every order and forwarded it to the right vendor via email. Processing time: 4-8 hours per order

Payment disbursement to vendors happened monthly via manual bank transfers. Reconciliation took 3 days per cycle. Vendors complained about delayed payments and opaque fee calculations

The platform crashed twice during holiday sales when concurrent users exceeded 2,000. Each outage lasted 3+ hours and cost an estimated $40K in lost revenue

The marketplace was growing despite its infrastructure, not because of it.

I rebuilt the marketplace as a purpose-built multi-vendor platform with automated vendor onboarding, real-time inventory, intelligent order routing, and automated payment disbursement. Designed to handle 10x current volume without architectural changes.

From 12 business days to 45 minutes.

✅ Self-service vendor portal with guided product upload, automatic image validation, and category mapping suggestions based on product descriptions

✅ Standardized product schema with validation rules catching missing fields, incorrect formats, and duplicate listings before publication

✅ Automated storefront generation: vendor gets a branded sub-page with product grid, filters, and reviews within minutes of completing onboarding

📊 Outcome: Vendor onboarding dropped from 12 business days to under 1 hour. Product listing quality issues reduced 82% through automated validation

No more selling products that don't exist.

✅ Webhook-based inventory updates from vendor systems with sub-5-second propagation to the storefront

✅ Automatic listing deactivation when stock hits zero, reactivation when restocked, with customer waitlist notifications

✅ Multi-warehouse support with location-aware stock allocation for vendors operating from multiple fulfillment centers

📊 Outcome: "Ordered but out of stock" complaints dropped from 85/week to under 5. Inventory accuracy hit 99.4% across 120+ vendors

Every order reaches the right vendor in seconds, not hours.

✅ Automatic order splitting for multi-vendor carts with parallel routing to each vendor's fulfillment queue

✅ Vendor-specific SLA tracking with escalation alerts when fulfillment exceeds committed timeframes

✅ Real-time order status updates flowing from vendor systems back to the customer-facing order tracker

📊 Outcome: Order processing time dropped from 4-8 hours to under 30 seconds. Multi-vendor order fulfillment accuracy hit 99.1%

Vendors paid accurately, on time, every time.

✅ Stripe Connect integration with automatic commission calculation, fee deduction, and vendor payout scheduling

✅ Real-time earnings dashboard for vendors showing sales, fees, pending payouts, and transaction history

✅ Automated reconciliation replacing the 3-day manual process with instant settlement reports

📊 Outcome: Vendor payment cycle shortened from monthly to weekly. Reconciliation time dropped from 3 days to zero. Payment-related vendor complaints eliminated

Find the right product across 120 vendors in 2 keystrokes.

✅ Unified search across all vendor catalogs with faceted filtering by category, price, vendor rating, shipping speed, and availability

✅ Personalized product recommendations based on browsing history, purchase patterns, and category affinity

✅ Vendor comparison on shared products showing price, rating, shipping time, and return policy side by side

📊 Outcome: Search-to-purchase conversion improved 34%. Average session depth increased from 3.2 to 5.8 pages. Cross-vendor discovery drove 22% of all purchases

Why I chose this stack and what tradeoffs I made.

PostgreSQL with tenant-scoped schemas over separate databases — 120+ vendors need data isolation without 120 database instances. Schema-per-vendor with shared infrastructure tables (orders, payments, search index) keeps queries fast and operations manageable

Stripe Connect over custom payment ledger — Multi-vendor payment splitting, tax reporting, and vendor payouts are solved problems. Building a custom ledger would have added 6+ weeks and ongoing compliance burden. Stripe handles PCI, 1099s, and international payouts

SQS for order routing over synchronous processing — Multi-vendor cart splits generate parallel fulfillment events. SQS ensures every vendor receives their order even if one vendor's webhook endpoint is temporarily down. Dead letter queues catch and retry failures

Next.js with edge caching for product pages — 45,000 monthly buyers browsing 50K+ products. Edge-cached product pages with ISR serve sub-1-second loads. Dynamic elements (inventory count, pricing) hydrate client-side