The Problem
The company distributed products across three warehouse zones and used eight carrier lanes with different rate structures, delivery windows, and priority rules. Every morning, a four-person operations team worked through incoming orders manually — cross-referencing spreadsheets, checking inventory by email, and assigning carriers by judgment call.
The process took three to four hours each morning under normal volume. During peak periods — promotions, seasonal spikes, or supplier delays — it collapsed entirely. Orders waited. Customers called. Staff worked overtime reconciling what shipped, what didn't, and why.
The biggest cost wasn't the time. It was the error rate. Roughly 13% of orders had at least one routing or carrier assignment error that required manual correction downstream — each one a customer service interaction and a fulfillment delay.
Architecture
We built a three-layer system: a React web application for the operations team, a Node.js API handling routing logic and business rules, and a PostgreSQL database for order state and audit history.
Order events from the customer-facing platform triggered routing via AWS Lambda functions, which applied a configurable ruleset — product category, destination region, carrier preference, priority flag, weight tier — and updated the warehouse management system (WMS) via REST API. An Amazon SQS queue absorbed traffic spikes without dropping events or creating back-pressure on the WMS.
The React dashboard gave operations staff a live view of order status, a queue of routing exceptions requiring human judgment, and a configuration panel for adjusting business rules without a code deployment. The team moved from reactive order handling to exception monitoring: they only touched orders that the system flagged, typically fewer than 5% of daily volume.
Technologies
Business Impact
- 70% reduction in order processing time — the daily routing cycle dropped from 3–4 hours to under an hour for the same volume.
- 14 hours/week saved across the operations team, reallocated to customer relationships and vendor management.
- 99.2% fulfillment accuracy, up from roughly 87% pre-launch.
- Zero order backlogs during peak periods — the SQS buffer absorbed a 4× volume spike in the first month without incident.
- Routing rule changes (new carriers, region reassignments, seasonal priority adjustments) now take minutes instead of spreadsheet redesigns.